Crimefighting with a Data Model, w/ John Hancock

CEO, Hubstream

Listen Now:

Data guy by day, crimefighter also by day!  John Hancock has revolutionized the Law Enforcement space with his company Hubstream.  His team uses advanced data techniques and tools to assist investigators to battle all manner of illegal activity.  His story, his data journey, and even his accent are fascinating!

Episode Timeline:

  • 1:35 – Project Gemini is how Rob and John met, the evolution of BI tools, and that pesky status quo
  • 13:50 – The story of John Hancock the crimefighter AKA Captain Correlate…and how Hubstream was born
  • 31:25 – How do you build a data warehouse that doesn’t fail? In part, it’s understanding Humans.  And an explanation on the marvel that is Hubstream
  • 1:30:45 – The team is…everything!
  • 1:40:10 – How are Power BI and the Power Platform involved in Hubstream

Rob Collie (00:00:00): Today's guest is John Hancock. And as he told me, when we very first met back at Microsoft, like in 2007, no, he did not sign the declaration of independence. John and I worked together for several years, working on power pivot, the forerunner of Power BI. And I had just enormous respect for this guy, but I had no idea the double life that he was actually leading. John didn't choose to fight crime with data. It chose him. Left Microsoft, started a company, inspired others to join him, built it the hard way, and it's paid off.

Rob Collie (00:00:35): It's really blossoming both for them and for law enforcement. Oh, and did I mention that it's built a hundred percent on the Microsoft platform, including things like Power BI. That's what we call badass. He even rocks a little bit of a superhero accent. As usual, we talk about a lot of things. Talk about data. We talk about building startups. We talk about people, the value of teamwork. And at one point he even uses the words, " the human element" completely unprompted. Such a good time, we're definitely going to have to have him on again at some point in the future. Don't take my word for it though. Let's get into it.

Announcer (00:01:10): Ladies and gentlemen, may I have your attention please?

Announcer (00:01:15): This is the Raw Data by P3 podcast with your host Rob Collie and your co-host Thomas LaRock. Find out what the experts at P3 can do for your business. Go to power pivotpro.com. Raw Data by P3 is data with the human element.

Rob Collie (00:01:33): Welcome to the show, John Hancock. How you doing, man?

John Hancock (00:01:37): Good thanks. How about you?

Rob Collie (00:01:38): Doing all right. You and I, our way of getting to know each other was we intersected so briefly in the grand scheme of things back on project Gemini Power Pivot at Microsoft.

John Hancock (00:01:50): Who would've thought that would take us all this distance, right?

Rob Collie (00:01:53): Yeah.

John Hancock (00:01:54): It seemed like a pretty revolutionary thing at the time, but it's super cool to see how it's all unfolded over these many years now.

Rob Collie (00:02:00): It felt special when we were working on it. I don't think any of us really imagined deep down that it was going to be anything as special as it actually turned out to be. I had to leave Microsoft to discover how amazing it was. Just such a revelation to me that like, wow, this thing we were working on actually is good. In fact, it's better than we hoped.

John Hancock (00:02:22): It turned out way better than and I think any has had any right to hope. It seemed like so controversial and weird at the time, this basic idea. Hey, you don't have to have these super high paid BI consultants of which I was one to actually get any value out of your data. Like actual normal humans. One day in this far off future land we'll be able to connect to information and build things themselves. And we go around and tell people that, and it'd be like, "No, that's never going to work, no chance."

Rob Collie (00:02:47): And of course, people telling us that didn't have any vested interest, and us being wrong, cause who are you going to talk to at those times, you're going to talk to BI consultants, right?

John Hancock (00:02:57): Yeah, exactly.

Rob Collie (00:02:59): That's the only people you're going to ask. It even turned out to be this radical way of doing things that was so controversial back then. Turned out to be... That's actually a better tool set. Even for those existing high priced BI consultants. A lot of them have really, really embraced this tool set. It just turned out that it was just better for humans.

John Hancock (00:03:20): Yeah definitely. It's sort of sure. I used to do these huge projects. It would take forever. They'd wheel me in there as this SQL server analysis consultant and I'd be part of a team of six and we'd go build these ETL pipelines and all this stuff, all the while having somebody translating the business. It'd be out there as talking to the business and come back to us and they'd tell us things and we'd build this whole big thing, get it out there in the world. And of course, no one would use it, because by then the business had kind of cruised along. And so, all the rigor that we had, all the data integrity rules and all the stuff like, we're so much better than you with your crazy spreadsheets and your formulas, which are bound to be wrong.

John Hancock (00:03:58): Well, actually the Excel sheets and that iterative process was just so valuable that real world businesses, they need that. They just needed that to be better. So, it's interesting when I've had quite an evolution over the years, in terms of thinking about that from somebody who, as you say had that vested interest. I was the number two data guy at Microsoft, Canada, heart, helping customers with that technology set. And you go out, you do enough of those engagements. You realize, holy crap, this isn't working. The projects we did were successful on their face, but every time you went back a year or two later, it was like, no, there's these dashboards that sit there and everybody else in the meantime, that's reverted back because there's now a new business unit and new business prices, all these things just keep on going at the scale or speed of the business.

John Hancock (00:04:41): It's super cool to see now, people just take this for granted. The terminology's even changed and people just think of this as BI now, this is how it works. It's not these big projects anymore where it's actually more agile. And of course there are all kinds of problems as a result of that. But those problems pre-existed on any of this technology and they'll live with us for a long time. How do you know this is right? Where's your data coming from? Those sorts of questions.

Rob Collie (00:05:04): We've always had those. I've got all these potential fight starters, that I'll say at conferences or whatever, for example, there has never been a traditional BI project that was a success.

John Hancock (00:05:18): That's a good one.

Rob Collie (00:05:19): And what you're describing, your previous experience is exactly what I mean by the traditional... It's a team of six and you're translating the business and it grinds forever with tons and tons of infrastructure and plumbing. And then by the time you're done, it probably doesn't even actually deliver on the original requirements, be perfectly honest because you need that iteration to deliver. But even if you did hit that initial target, that target has moved in the two years that it took to execute. I think you're right that the new way of doing things is now considered to be like the right thing. It's sort of like intellectually, however, the real world hasn't caught up to that reality at all yet. It's like the intelligency have decided, "Okay, this is the new way," But you go out and you witness the reality and reality is very stubborn. It's got a lot of inertia. We still probably got a decade maybe even, I don't even know of catching up. The real world has to actually adopt and change to meet this new agreed upon best way.

John Hancock (00:06:26): I agree. I also see a lot of people out there where you feel like they have to age out of the system for change to happen. Anyone who comes from that world, who hasn't made the leap by now into look, your job is to actually service the business. They're moving faster than you can. Anyone who uses the word ontology, if you hear that word or something like it, variations on that thing, that's it. You are the anchor. You are not the person who's facilitating what needs facilitating here right.

Rob Collie (00:06:52): Yes. On our previous podcast, we actually talked about the Max Planck Theory of scientific revolution, which is, that it's not this meritocracy of ideas that everyone wants it to believe. The old guard has to leave.

John Hancock (00:07:05): Yeah, definitely. And I think that the other interesting part of this, from my sort of part of the world, which is really about helping people to investigate crime, what I see as well, is this big revolution going on where the sort of analytics used to be something that you bolted on and it was something you did in the rear view mirror. And even in the new agile thing is still, most of the stuff I see is rear view mirror stuff, right. We are doing this new thing and we need to be able to count the metrics or do the KPIs of this new project, which is great. But you're always looking in the rear view mirror. You figure out what data you're collecting, you clean it up, you build analysis around it. And there you go, look, now you can see behind you.

John Hancock (00:07:45): So what we are finding interesting in our space, is trying to figure out how you sort of marsh that data analysis in a forward direction, because for investigating crimes, the value isn't in the rearview mirror, it's going into the process that they're doing on a day to day basis and saying, "okay, every single step you take along your journey to do these investigations, how can we marsh all this information of which there's tons, it's messy, it's large. How can we bring that into your day to day work and help you see forwards?" Because what we're trying to do with crime is you're trying to bring all that together and say this is where you should focus your attention, all the sea of noise, here's where you're headed.

John Hancock (00:08:24): So it's kind of like for us, we want to be more sort of like ways, where you're driving along and it's not looking in the rear view mirror. You get these little popups along your day today, which is, "Hey dude, there's a big traffic jam ahead. You want to go to the left" or "There's this good opportunity over here" And so trying to martial all of that into their day to day is what I think is interesting. Especially since with SAS, I think the next 10 years as it evolves is going to be more and more about that forward looking approach to data analysis. How do you like start altering people's behavior on their day to day as they're executing, whatever it is you're trying to do, rather than just having these projects where you can measure the thing that you're doing anyway?

Rob Collie (00:09:06): So I want to make the transition into your alter ego crime fighting, doctor data. We're going to come up with some sort of superhero, Marvel-ish name for you, but...

John Hancock (00:09:18): That's fantastic. You guys should totally do that. I suck at marketing personally.

Rob Collie (00:09:21): [inaudible 00:09:21]

John Hancock (00:09:21): My name is John Hancock. That's not weird enough. I mean it's...

Rob Collie (00:09:26): Yeah. I remember. And when you introduced yourself the first time I met you, you introduced yourself to this whole room and you said, "Okay, and first of all, Americans, yes, my name was John Hancock." And that's my first memory of you. It is literally that.

John Hancock (00:09:41): It's pretty memorable. That's right. I still, when I was out in the world, I used to get one to two jokes a day on average about that one. And so, you kind of get a little bit bored of it, but whenever it was somebody at US Customs making that joke, I would laugh like it was the first time I heard it. They look at your passport. They make the John Hancock joke. You're like, 'Oh, you're so funny, Mr. Customs guy. Yes. Please let me in."

Rob Collie (00:10:04): Don't don't send me back. But before we go there, I want to almost pick a fight, but I actually think in the end we really agree. So I hear this a lot about BI, the rear view mirror thing. I want to seize this apart a little bit for our dear listeners. So first of all, I think one of the reasons why BI has had, or has this reputation as a rear view mirror is first of all, it was so damn slow, right? And even when you had reports that you liked, which of course we've already agreed, never happened. They would be run like monthly and it would be too late to make any difference whatsoever. And as you increase the frequency, as you increase the velocity and the speed at which you're able to see, I start to think of it as no longer the rear view mirror.

Rob Collie (00:10:51): I start to think of this as the windshield, because you know, like truly seeing the future, that's predictive analysis, right? And so we want to make a distinction there, but I think the point is, is that if you're using data to look backward, this is a subtle philosophical shift. If you're using data to look backward, versus you're using data to try to decide what to do today factually mathematically, it might be in some sense, you could argue it's the same thing, because you're using things that have happened.

John Hancock (00:11:25): Yes.

Rob Collie (00:11:25): The other way to describe things that are happening in the rear of your mirror is facts.

John Hancock (00:11:31): That's true.

Rob Collie (00:11:31): They're the only things that have happened, everything else is a guess, right? But you're using it to synthesize your move. I agree, that's one of the things we've been talking about a lot is that it's all about the action that you take. Being informed is worth nothing on its own.

John Hancock (00:11:45): Right. It's about the behavior that results from that. So I think that is kind of maybe a way to slice this problem a little differently, which is that data analysis or data that's changing, where you take the car, changing the direction you would've steer. And if you didn't have that thing available to you, you would take different actions as a result. So that's, I think the next sort of 10 years of how to measure the value on this. It's really about what did you do differently that you wouldn't have done. And then you go back and you say, well, because of course, as I was doing the thing, the system around me was aware of all the information and was telling me all the good things we should go do. And so we did it, and that just becomes a normal way of doing it.

John Hancock (00:12:24): And you see this showing up, I think in micro areas where you already had people doing their work on a day to day basis, that needed guide like CRM, where like here, or call center's a classic one, "Hey, I'm in a call center." And the system's telling you stuff like, "Hey, ask them about this and tell them to add, telephone service, because it'll save them, and we'll make more money." So what are these systems like that? But they've kind of pointed at the kinds of areas of the world that are very easy to, relatively speaking, to guide people. There's a finite set of actions. They bring them in. When you start getting into the more complex areas, which I think fit the bill for most of what the value is that's being created in the world, then it gets much more complicated.

John Hancock (00:13:07): And I think we're starting to enter into a new era where giving you useful guidance as you go, it's becoming more and more a possibility out of the sea of noise, the unstructured and the structured and all the things that are happening to start to be able to learn from that and go look, you are sitting here as an expert in your domain and the value I'm giving you is things that you couldn't possibly have gone out and found yourself. And that thing is actually helping you to change the direction. You're going to do a different thing now, because I've told you out of all the thousands of things, here's the most relevant thing you need to know. And the most actionable thing you need to know.

Rob Collie (00:13:41): It's almost like to continue the metaphor. It's like, we've finally, as an industry developed the art of windshield and window glass. Right. And so now we can talk about steering. John, you and I overlap for probably two years.

John Hancock (00:13:55): Something like that, yeah.

Rob Collie (00:13:55): Too good, well, they were tumultuous years for me. But two years back in the formation of essentially what became Daks and the tabular model and the beating heart of Power BI, but all this time, our friend John was leading this double life. I don't remember you ever once talking about it. I'm the kind of guy, if I'm living a double life, all you're going to hear about is my other life. Like that's all I'm going to be... You're going to be inundated with it. Like, "Okay, come on now, Rob stop." But you, I don't think you were keeping it a secret. It's just wasn't the thing... It just wasn't that important to the work at hand that we were doing every day. And so it just never came up. But can you tell us about this double life as a crime fighter?

John Hancock (00:14:39): Sure, absolutely. It has been an interesting journey. So back at the beginning of all this, I was working, as I said, as a business intelligence consultant. I ended up working at Microsoft consulting services in Canada. And so I would go out there and help people do those projects I talked about, right. I'd be the Microsoft guy on that six person project team. And we'd go and help banks make more money. In that whole process, one of the days I went back to the office and walk in the hallways and one of the engagement managers comes out to me and he says, "John, I'm so glad to see you. You're one of our top data people. I have this great opportunity for you to volunteer to go do something." I thought, wow, that's a pretty weird thing for one of these guys.

John Hancock (00:15:15): I bail out at X amount of dollars per hour for Microsoft. You're asking me to volunteer. This is weird. She says, "Yeah, yeah. We got this opportunity for you. You can go in and help Toronto police investigate crimes against children, like child pornography offenses." And I was like, oh, no way am I doing that. No chance. Thanks dude, but no, I'm out. So I went home and thought about it and talked to my wife. And I was like, "Can you believe this guy wants me to go help Toronto police for this?' He was like, "Oh, so you're going to... You're just going to carry on helping those banks make more money then so, okay." So kind of molded over and went back in. I said, "Look, idea of like helping investigate crimes against kids is horrifying to me. I don't know what value I could add. And I probably can't take it." Many people in the world, if you say those words to them will be like, "Hey, I can't handle that." And some brave individuals were out there dealing with this, not me.

John Hancock (00:16:07): Long story short, I ended up getting pulled into this project, had Toronto Police and it turned out that back then sort of 2003, they really just started realizing that the technology revolution had carried on and kids were chatting on MSN Messenger and all these other products. And they were out there in the world. And there were all these predators who were starting to reach kids in domains over the internet, as opposed to just the real world, which is terrifying enough. And so I ended up going in and trying to help Toronto police who had just formed a unit to go after a crime type, which they're called child exploitation, which covered a vast array of bad things that happened to kids online and in the real world, and then published online.

John Hancock (00:16:51): And so I spent about three months embedded with Toronto police and trying to figure out what to do. And it's big challenge. As a data guy, I showed up there and expecting that there would be drowning in data, just like the banks and all the rest, which is why I got sent in. And I showed up and biggest source of data they had was an Excel spreadsheet, they're managing. Now they would go out and then make a risk and they'd come back with machines and they have vast troves of information. But the day to day things they were using, they didn't have a data problem to solve. So as a BI guy, you kind of taken away my typical act. There was no way I could go in and go, 'Well, you need to assemble your data and build an ETL pipeline and blah, blah, blah." So I ended up having to actually listen to them and spend time figuring out what they needed to do. And it was a pretty low point actually. I was over there in the summer and I was like, "Oh my God, I've really got myself into something here. People are expecting me to use information, to help these guys, and I really don't know what to do." So two months into it, I thought to myself, "Well, this is going nowhere." I'm not coming up with things. They're not coming up with things. So I started to just take people out individually for coffees, right. So I went for one of the senior people in the team, took him out for a coffee. It'd be in Canada. It was Tim Hortons. I have to do a Tim Horton shoutout. Maybe he can get a sponsorship for your podcast.

John Hancock (00:18:06): And so we sat there and I said to her, "Okay, look, just tell me about any success you've had. Let's see... Can we talk about your successes you've had so far?" She said, "Yeah, funny you should mention that. It was a really great case we finally did last week." And so what happened was as Toronto police, we got a report from a parent in Toronto saying that their kid was on MSN Messenger and they're chatting to a friend on MSN Messenger who they thought was an 11 year old boy. And the parent had found these messages that included pictures and all kinds of stuff. And so the parent was super worried. So they reported her to Toronto police. So they get the report. The only information in there is, this is a made up email address just for everybody, but they get an email address like [email protected], that's it. All they have to go on.

John Hancock (00:18:51): So the cops said, "I spent the day trying to find information online, found nothing. We've read nothing, ready to go on." So it's not actually illegal behavior. It's just shady. So there was nothing I could do. So he said, "I went out for drinks that night with my buddy from Vancouver police. And we were sitting there having a few beers and talking about our days, and he's also in this new child exploitation domain. So I was bitching about how I've just had this really bad day. I've got this parent, I've got this kid out there. I've no idea who this [email protected] is." She's just complaining about how much this day sucked. And the Vancouver cop goes, "Wait, did you say [email protected]?" And it turned out that they'd had this huge, big, undercover operation for months in Vancouver.

John Hancock (00:19:35): And those Bob54 alias had kept coming up, but they didn't have enough information on that side either to do anything about it. So the two cops then put their information together. Now they've got enough to actually go and get the information. They put in a request to Microsoft with all the legal backing to get the guy, so result. So at that point, my brain just completely exploded because I finally got it. This is the actual problem. Finally, I found the real issue here. And the issue is, while the criminals are starting to adopt all these new technologies and they're out there completely unbound and completely borderless, the way we've organized law enforcement response is completely geography based. It's been like that since forever. And that's how we've organized all of our criminal investigations, which is Toronto police is reporting to the Toronto mayor. They're paid by Toronto taxpayers. And if you drive down a particular street, you cross over a magical geographical boundary. You're now in the territory of a different police service. Those two different agencies are not connected. They're not sharing information. They are completely siloed.

John Hancock (00:20:42): And so right there in 2003, I just had this huge revelation of the fact that all the crime that starts to get to be mediated on the internet or anything that revolves around the internet, all of a sudden is this totally borderless space. But all of our response as a society is organized around geography. So I'm like, got it. Finally, I'm not going to sit here being a useless data person with no data anymore. I'm going back to the office and I'm going to tell him the news, right. So I go in there and go talking to the president of Microsoft Canada at the time who was awesome guy, who also had two young kids.

John Hancock (00:21:14): And I said to him, "Look, I get that you probably sent me in there hoping we donate a few copies of office, but I've got some great news. Instead of doing that, we're going to build a giant system to unite the whole of Canada. We're going to build this massive project to integrate all the people who are doing this kinds of investigations together. So that every piece of information that comes in can be correlated and matched against all the other stuff." And so we basically pitched that to Microsoft corporate and we ended up building this huge system for Canada as a totally pro bono effort out of Microsoft citizenship project. We basically over a year and a half built a system that is still being used to really manage the flow of information in, into Canada and help them to figure out that there are these different connections. And so that was really the start of my double life, right. Because I switched from doing just BI consulting for banks into running this project as the solution architect for a number of years. And even when I ended up leaving that project and moving to Redmond to start work on Project Gemini and all the other BI technologies, I was still kind of acting as an informal advisor to that project, which ended up at Microsoft's digital crimes unit for many years. So I always had this double life, as you said, where I would do my day job of PM on the Project Gemini team and in lead and eventually GPM. But all the time, I was also like driving over every now and then into digital crimes units and talking to the people and still staying in touch with the law enforcement, who, as you can imagine, 2003 was actually pre-Facebook.

John Hancock (00:22:46): And so the world had continued to evolve at an incredibly rapid rate. And so the things that we thought were going to be tough right at the beginning, just got tougher and tougher every single year, the amount of information coming in doubled, so this is something that we've seen over and over in all types of investigations, which is just as the whole of society moves online. Everybody around you is using technology. You're Skyping your mom in a foreign country. Everybody is just technology based. Well, of course, crime moved with it. So there's no such thing as offline crime anymore, right? Everything that is crime in the world has some online footprint. And because of that, the information that you have available as investigators is huge, right, because there's all these different things that are collected, but the problem switches over to how do you actually figure out what's important?

John Hancock (00:23:36): How do you actually, as you said before, how do you steer in the direction of the worst harm? So that as an investigator, you can do something about it. You've got a finite amount of time, finite amount of resources. You have to be able to synthesize all that information together. And then on a day to day basis, be focusing on the things where you can make the most difference. And that's really where the company that I found at Hub Stream with some great people out of Microsoft, we've put together a company that really builds the software that takes that project we originally did at digital crimes unit and has kept on making it available for that crime type, but also many others.

Rob Collie (00:24:12): At least one of the people that went with you to Hub Stream was one of our absolute ACEs in the hole on Project Gemini. Let's come back to him in just a second.

John Hancock (00:24:22): Sure.

Rob Collie (00:24:23): I want to appreciate that guy. Yeah. I really do. So a lot of things about this problem space that are unique and sort of domain specific, I'm not going to try to pigeonhole it into saying it's the same as everything else, because it isn't. At the same time there's still some themes here that are familiar. It's this siloization is a huge problem in the business world. And some of times it does sort of take a geographic silo, right? Like if you were a company that has grown through mergers and acquisitions.

John Hancock (00:24:52): You have all these subsidiaries and... Absolutely.

Rob Collie (00:24:56): You got the operations in different countries and everything. So you have that silo problem, but you also even have a silo problem across line of business systems. That's been one of the big, big, big themes, really that Power BI is amazing at. See the silo problem and you recognizing that it was a silo problem, right? It was kind of the moment where the light bulb went on and your life basically changed forever. It's a dramatic way to say it, but it's true.

John Hancock (00:25:19): Absolutely.

Rob Collie (00:25:20): This is the sort of problem that once you get a hold of it or once it gets a hold of you, it's never letting go. It's easy to see and understand how it maintained a place in your life. You were still going back over to the digital crimes unit.

John Hancock (00:25:35): It's a combination like a really challenging problem in data generally, but just a generally challenging problem, but also the opportunity to make a big impact, which was really just so driving me. And that's how I managed to persuade people to quit their great jobs on Microsoft and come over to this crazy adventure. It's that combination, right? It's a super hard problem that's going to keep growing every day we show up it gets more complicated and we're constantly having to innovate and come up with new things. And at the same time, the impact that you can make in the space, particularly crimes that are so obviously horrific and that where everybody watches TV and movies and they see these like fancy CSI type systems. But you actually walk the hallways of any police department or service or agency that's responsible for investigating crimes.

John Hancock (00:26:23): And what you see is ancient technologies completely out of date, three or four different things, many national federal agencies. If you go there, they have three terminals on their desks. There's these three systems that don't interconnect and they swivel their chair in order to integrate. And so it's just such a huge opportunity as those law enforcement agencies have started to modernize. And particularly now that they've started finally to move towards the cloud, there's these massive opportunities to make a difference in their lives and in the effectiveness of their day to day investigative activities, as a result of that. Even large corporations, which increasingly are customers of ours in areas like pharmaceutical companies, that sort of thing. They do vast scale investigations where they're getting so much information in, product counterfeiting is another great one. You know, if you've got a big brand and you start to dip your toe into, "Hey, there are people selling my brand on the internet.

John Hancock (00:27:21): And it turns out to be totally fake or it's been diverted," It's actually gray market or whatever. The, the volume you need to actually to assemble and put together and actually have a reasonable understanding of is just growing and growing and growing. So that's how come we've ended up starting this hub stream adventure, which was, I saw back then what was going to start happening, and we've really spent the last several years at Hub Stream building that technology platform. And we just wake up every day with that problem space. We go back to it and we keep pushing forward and some days we make progress and we're ahead and it's an arms race, right. As soon as you make something better, the criminals adapt and then we have something else to go solve. So it's this never ending journey, I think, which has been a really exciting thing.

Rob Collie (00:28:05): It sounds to me like you helped Canada build a watch list.

John Hancock (00:28:10): Yes, kind of. What we really did was we re-engineered how they think about investigations.

Rob Collie (00:28:16): Okay, but you essentially built them a data warehouse.

John Hancock (00:28:21): Yeah. We started out doing this simple thing, right. Which was, he said, "Look, if you're in this place and you're in this place and you both touch upon the same entity, we should let you know." And so that's a great start, but then you scale it up. And so where that goes after that it's complicated, right? Because now, well, where's that information coming from? Well, you start out with humans typing. And that was where we started. 2003, I was a human typing that stuff in there, Rob54@hotmail, so great. Then fast forward a few more years and you start to get feeds of information in, and this is the same thing you see happening in companies as well right? They start with a sales team of one, they're out there prospecting, they're finding stuff. Then they hire two people. Then they start bringing in data feeds, same thing with investigations. And so the problem starts to show.

John Hancock (00:29:03): Bringing in data feeds, same thing with investigations. And so the problem starts to shift from connect the dots of what humans are doing in any given day over to, okay, I'm signing on as an investigator this morning, I have a hundred things that got reported to me. Okay. I'm now going to have to take action on one of them and I might get to three. By the end of the day, I'll have figured out three of them to go do. So which three? That's the challenge that comes in. How do you start to alter the day to day flow? And then when they start to get those data volumes, the other big challenges that start coming in are how do you help people understand that much information? We hit this tipping point I would think about five years ago now, where issue used to be how do I acquire the information that I need in order to do these investigations?

John Hancock (00:29:46): And today the problem is I've got so much damn information, whether it's law enforcement or corporate investigation, whatever it is, you can get feeds, there's crawlers, there's all sorts of sources of information. Now, the problem that people are dealing with is in a world where you've got so much information, you've got so many opportunities to go do things, how do you put that in front of somebody in a way that they can understand quickly, make sense of it and also have some ideas about what the options are and do that in a scalable way. And so that's why, like I say, every day we come in with different challenges and it's really about how do you start taking those emerging things. Like Bitcoin comes up. Well, now Bitcoin has a whole thing about it, it's a whole different element.

John Hancock (00:30:32): That's starting to play into your investigations. Six months ago that was a completely irrelevant side show. And today 30% of your investigations are featuring Bitcoin, because it's a nice easy way to move money around. So, okay. How do you take an agency that has probably persisted for a hundred years in the case of these large national law enforcement investigations and help them change that quickly? Six months in Bitcoin's now a thing. Six months ago was a totally non entity. How do you start to shift? And so building the pipeline of data of visualizations of actions into the day to day systems and day to day work of the investigator, that's really what we've ended up focusing on. So our system for it to work, your investigators are using that as the system. They show up, everything they do is through that system. Everything's tracked, everything's audited, everything's logged and then we have the opportunity to put all that information together and help them do the right thing.

Thomas LaRock (00:31:28): So we had started earlier, we were talking a little bit about the old school data warehousing and basically all data warehouses, in my opinion, are destined to fail. I've never worked on a successful one. I ask a room for of the people I say, "Who here has worked on successful data warehousing project from start to finish?" And everybody's like, "Yeah, actually no." So I was going to ask you, how are you able to maintain a level of success? I mean, is it just your background? Do you know that if you just let things go, it would just fail? So what are you doing? What are the iterations you're doing to allow it to be successful?

John Hancock (00:32:06): Yeah, that's a great question. So I think part of the reason why if you'd diagnose the data warehouse malady it's that thing where super smart people get together and they're way away from the actual day to day problems. And so there's this ivory tower issue. I see that over and over particularly in big law enforcement projects or investigative projects. There are some amazingly smart people out there. And so I think the way that we constantly approach this problem is with humility. We don't do investigations, we aren't investigators. We don't use our software to do investigations. We are our software people. And so the way that we've tackled this issue with the help of our awesome chief design officer, Joe Mylan, who also was one of the people who worked with us back in the project Gemini days, it's really about talking to people, talking to the investigators.

John Hancock (00:32:56): So the reason I had that funding revelation was because I was sitting there having coffee with somebody who was trying their hardest to do a very difficult job, listening to that person, and then going back and thinking, how can I help that person be more successful? That's what we do all the time. And so I've assembled a team of people who lead with humility. If you meet them all, they're all the good ones who are able to check their technical skills at the door and actually come and listen, and then go back into the technology space and marshal the things they can think of, but do it in a way that you keep constantly in touch with the investigators who have the problem to try and make their lives better. And we don't always succeed. A lot of the time we put so mean together, we get it in front of them and we realize we just haven't hit it.

John Hancock (00:33:43): We've built 80% of the useful feature. Now we've got to go back and spend double the time building the extra 20% and actually make it show up. We know this pattern and yet we still keep repeating it. We just built some software, we got it all the way through, looked at it and realized once we started showing it to the humans that previously talked to, we built it in a way that was not going to be in their day to day flow. It just wasn't going to make a difference. It was some cul-de-sac. So got to go back and redo that and put it right in front of their actual flow. So that's the thing. I completely agree with data warehouse failure as a patent. I do think it's that super smart people who are not the domain experts as a patent that keeps cropping up.

Rob Collie (00:34:24): That thing about the ivory tower and smart people getting together and doing something completely detached from reality while self reinforcing each other.

John Hancock (00:34:34): Yeah. That's the other part, right? Yeah. The people who are on that project, I bet you if you went and talked to those data warehouse projects, quite a lot of the people who led that project got promoted, right?

Rob Collie (00:34:44): Yeah.

John Hancock (00:34:45): And so there is a self fulfilling thing about that.

Rob Collie (00:34:47): It's weird, isn't it? This is like just an ongoing theme for me is the technologists of the world, the people that you need in order to be good at doing something like that, to execute a data project, to execute a software project, you need a certain kind of background. And that background almost overwhelmingly brings with it this other mindset, this ivory tower academic mindset. And it's when you can have that tech background, but break that ivory tower mindset. The technology isn't some way to insulate you from the world. There's a lot of refugee mindset in tech. If I can just mathematically encode the world and all these human beings that I've struggled to deal with, I can finally make sense of it and it's a fools' errand. So yeah, the success has happened when you have that humility, when you step out of that, but retaining your technical skills obviously.

John Hancock (00:35:43): Yeah, absolutely. I think that there's another psychology element of that, which comes into play as well, which is that the people who are interested in data in particular and data technologies and all the rest of it, many of us including me are quite introverted. And so the actual act of communicating with other humans, getting up, walking to somebody, picking up the phone, it's counterculture to a lot of people who are interested in data. I still see that the extroverts of the world are not flocking to data science jobs, right?

Rob Collie (00:36:13): Yeah.

John Hancock (00:36:13): It is like that old joke how can you spot an extroverted DBA? An extroverted DBA is one who looks at your shoes when he speaks to you. So that joke has stuck with me because it's just so damn true, right? So I think that's that intersection that the introversion as well plays into that where a lot of people like me and data, they have to push themselves out of that to go and actually talk to humans. It's a big challenge. It's a big psychology thing that I think is inherent in all the things we do.

Rob Collie (00:36:45): On the video feed here, I'm watching Tom rethinking his entire life.

John Hancock (00:36:51): Yeah. I think I have compassion for that because as I say I'm the CEO of a company now. I actually, I do a lot of sales myself, but I'm constantly pushing myself out of that psychology issue. So I think that's part of our foundational DNA as a company and Hub Stream is that a lot of us are data nerds going back and we just constantly try and push ourselves out. The other nice thing about our culture is that because we have these super interesting customers, that's the other thing, the things they're working on are not boring by any stretch of the imagination. So that helps you to push yourself out because you can be genuinely curious, what is it like, what are you trying to do? What are you trying to achieve? How can I help you?

John Hancock (00:37:29): And then you do something and it actually makes a difference. And there's this huge reinforcement loop. Because if you can actually make a difference in the day-to-day lives of any investigator... We work with a pharma company that's doing investigations that medications that are counterfeit are killing people all over the place. If you can help that team be a little bit more efficient, a little bit more effective, we wake up every day trying to figure out how to do that. And so that gets us out of that silo and that psychology trap of moving into just let's think about the technology space and back into the real world and back into the day to day lives of real people wherever we can.

Rob Collie (00:38:05): When did you leave Microsoft to do Hub Stream full time?

John Hancock (00:38:09): Seven years ago now unbelievably.

Rob Collie (00:38:13): Wow. That's crazy. Okay. So what triggered it? What triggered the departure? But to me just as even more interesting perhaps is how? Did you get funding? I mean, did you just walk away from that and how did you get people to follow you? It was like this Jerry McGuire moment where you're walking out with the gold fish saying who's coming with me.

John Hancock (00:38:39): Yeah. It's been definitely a... It's so interesting. Such an interesting journey. So the thing that precipitated it was while I... I was having a fantastic time at Microsoft. I loved working there. Working on PM at Microsoft in the various jobs that I had was awesome. Loved it. It was just so fantastic to be working on that. And I finally was working with a fantastic team who was starting to do new stuff. We got to hang out with Bill Gates occasionally, which was so great. And we were just having a wait over time. And so I was settling in for the duration, right? This is it, here I am now, I'm going to stick around and do this cool thing. And like I said, I had stayed in touch with the digital crimes unit. And what had happened was it had come to the end of its life.

John Hancock (00:39:22): The project that we'd been working on or they'd been working on, it was out there in the world. It was being used by big law enforcement agencies. And for various reasons, its really hard to maintain a product at Microsoft which this was without a billion dollar market. That's just the reality of it. And so digital crimes unit had done their best, great people, but they couldn't figure out how to continue on the innovation that was going to be required in order to keep up with the demands. And so the guy in charge, the fantastic guy, he just basically eventually said, "Look, we can't continue on doing this. We're not doing them any favors anymore. We're actually in the way of innovation rather than facilitating it." So they had decided that Microsoft was no longer continuing on this journey. So at that point, once I had that conversation with them, I had this massive overwhelming dilemma here, which is okay, I know I could make this work, right?

John Hancock (00:40:20): I don't need a billion dollar business for this to be successful. I know some awesome, fantastic people that I could take with me and we could go build this company where we could make a difference in the world, we could have an impact and we could also make a successful business out of it that would grow and grow into new areas. And funny enough nowadays when I talk about that philosophy of I'm going to make a company that's going to make a difference and also have a sustainable business model actually quite widely accepted now. If you look around, there's a whole slew of companies now that are impact driven. They are sustainable, they are organized as profit making enterprises and that's how they keep the lights on and pay people salaries and all the rest of it. They're not throwing Galla dinners, they're not trying to get philanthropists to give them money. They're actually trying to build a sustainable business, but they're also have this direct focus on making an impact in the world.

John Hancock (00:41:12): But back day when I was talking about that, people were like, "Wait, what are you saying here? Are you a for-profit or are you an NGO or a charity or what are you? What are you guys even talking about?" I was like, "I don't have to choose. I can do both those things." That's literally how this is going to work. I'm going to make a successful product. People are going to pay me because it makes things better for them. And also I can make a difference in the world. I didn't have to choose.

Rob Collie (00:41:35): I love that. Love it, love it, love it. There's something to a small fraction that I find similar in my experience which is that when you set out to be a good deal for all involved, there's something about that that even today sort of runs antithetical to like what's quote unquote considered "good business." There almost has to be a loser somewhere, right? Or someone that's being harvested or milked in order for the real businessy types to like say, oh, you're on the right track. And it turns out if you ignore all of that, you can go and build something that's successful and it is for profit and it has a good business model, but still is good to all the human beings involved in the ecosystem. It's such a thing for us that we even have an internal... It's mostly me that does this, the nobody believed in us, nobody gave us a chance. I didn't watch the super bowl post game but someone said, everybody counted us out, nobody thought we could. It's just guaranteed. And to hear you saying these sorts of things really resonates with me that if you just ignore all of that and go figure out how.

John Hancock (00:42:48): Yes, that's a challenge, right? So if you switch your brain and, and try and take the, okay, I'm going to do this. There's not some magic line between making the world a better place and being able to do that on a day to day basis for a long time, which is a sustainable business model. If you think that there's a one thing you're going to go do, then you can start planning for that eventuality, right? Like, okay, how do I do that? And that's going to involve thinking very carefully about the business you're going to go build, where's the money coming from, who's paying it, what are they paying it for? And also what opportunities do you have to make the world a better place? And how are you going to go and execute on those while not going out of business because you're doing things that aren't also then leading into inbound revenue.

John Hancock (00:43:28): So it's a funny thing, even now after all this time, I'm in touch with quite a lot of companies that are trying to pull that off. Whether they're doing human trafficking or whatever the domain is, or video interviews for child victims. There's so many things out there where people are trying to build a sustainable business that has an impact in the world. What we still see even today is there's still biases left where people who start a not for profits can tend to be a bit looked down on for profit companies. Like, oh, well you're not really focused on impact. You're actually in this for the dollar. I see your people are well paid and you have a nice office and yet you claim to be for impact and making the world a better place.

John Hancock (00:44:10): Somehow the not for profits are better in some sense. But that mental model I think is shifting because the people who fund the impact in that space are also starting to ask questions like, well, what is your long term business model? How are you going to sustain this thing other than us constantly writing checks to you guys and then trying to figure out how to measure your impact. And so we had a lucky thing in our case, which was, I never bothered to try that because in order to go down the road of building a not for profit, you actually have to have background and experience in doing that. I know some people who try to do it without that background and experience and it was a catastrophe because you got to figure out where you going to get the money from, how are you going to measure it? Are you throwing Galla dinners here?

John Hancock (00:44:51): I realized that if I threw a Galla dinner, no one would want to attend it. It was just the basic fact of the matter. No one's coming to John's Galla dinner. So, okay. Well, if that's not going to work, as you say, you've got to go focus on what's the art of the possible. And I had done a startup before in the late nineties, which coincided with a dot com boom and crash although it wasn't a dot com company. It was this wacky idea we had which was what we called data analysis. No, we didn't call it that. What was it called back then? Business intelligence probably. Business intelligence over the internet. No, it was decision support over the internet. That's how old it was. 1999, 2000. And we built this BI thing, which was a Java front end and we were FTPing from our customers AS400 systems and then giving them back a driver front end that they could log into.

John Hancock (00:45:38): And so I did that startup in 1999, 2000 with a bunch of other great people and we took money for that startup. And that taught me a lot about what happens when you take money for a startup, right? That ended very badly for all concerned, including those who put the money in but also those of those who worked our bets off was a very bad outcome. We were all stuck with... I try to work for a while, which was just terrible. And the thing that that taught me was that who you take the money from and what their objectives are is going to maybe align with you in certain respects is maybe not going to align with you in other respects. And when things start to go bad or get complicated, that's going to be the most important factor about whether this great venture you have is going to succeed or fail.

John Hancock (00:46:21): So getting back to there I am at Microsoft, digital crimes unit tells me that they're not continuing on this journey, I've got this business plan. Frankly, I had this little notebook of ideas that I'd been writing in for years. I was like, "Well, of course, I'm not going to do this. But if I did do this thing, what would it look like? How would it work?" And so brought out that thing and looked at it and I was like, "Well, I'm not going down the not for profit road because I don't know how to do that. I'm not going to go and raise money because explaining what I'm doing here is just too complicated. It doesn't fit into the nice buckets that investors are going to go and feel comfortable with. And so I'm really just going to go and do what I see a lot of companies doing in this year, which is bootstrapping. Find a customer where you can add value to them, do the thing with them, keep going and going and build and add more customers."

John Hancock (00:47:10): And it takes a long time, I'll be the first to testify on that. We went way faster in the company where they injected a bunch of cash, but we also crashed into a wall way faster too. And the fall art was bad for all concerned. They lost all the money. We lost all our work. It sucked just wasn't good for anybody. So the bootstrapping model is a great model, but just you commit to growing at the speed of you can bring customers in and if you want to do different things, you have to try and find a way to carve out that time in the context of giving these customers value. And so that is what when I'm out there in the Seattle world at the moment talking to other people who are doing this impact thing, pretty much everybody is following a similar model. They're trying to build a bootstrap business. Sometimes they manage to get creative models where they can involve the philanthropic world some way in that journey. But most of us are just trying to figure out how to put together profitable businesses, usually cloud based, usually SaaS while also making the world a safer place in whatever way we can. So now it's a thing, it's out there.

Rob Collie (00:48:12): Again, a lot about that resonates with me. We ended up going the bootstrapping approach to build P3 and you're right. You have to be a lot more patient. It takes a long time to build to the same place that you might have been able to build to if someone had just cut you a check. But I know that reliving the history of our company, there are moments in the history of our company where an investor would've forced us into a short term view of things, like a local minimum. The incentives would've driven us in a direction that was not the most resilient and robust version of us. And we probably wouldn't have made it. I can absolutely imagine that. And so you've got to wait for that what they call the, let me tell you about my 10 year overnight success.

John Hancock (00:49:02): Yeah. That's exactly it. You see this it's happening quite a lot now that I think the world of venture particularly is changing so radically. Because now you can get going so much more quickly than you could before and that means you can hit customers pretty quickly. By the time I checked my Microsoft badge and exited, I already had line of sight to... Of course we had an advantage because we had the product already from the digital crimes unit area, but we also had customers lined up and we could service them from day one. So as soon as we exit, we had one customer and then another customer, some time went by, we got another one. So we started to build up quickly and that gave us really the freedom to do what we wanted to do. So in addition to growing the business, we've also done insane things. We run this huge global system as a pro bono project. We've never figured out how to get funding for that thing, but it's used by thousands of agencies. And so we just carried on doing that.

Rob Collie (00:49:55): So John, if I understand it correctly, the crimes against children stuff, that's something that you just continue to provide basically pro bono.

John Hancock (00:50:03): Well, it's an interesting one, right? Because a lot of the things that happen in the world are local like I said. And trying to get a local police department or the local task force or whatever it is, software, particularly complex hosted software is super hard. So for those kind of cases, we've just basically take the approach of, okay, we're going to have to do our best and do it as a pro bono thing. Sometimes we've figured out government grants or whatever to try and bridge the gaps. But for the other agencies that are doing those kinds of investigations, the big federal ones, they needed more than just technology, they also needed help, right? And so some of them also needed hosting when we finally got to the cloud. And so for them, we've got a bunch of paid services that we can offer that are annual base where we basically work with them.

John Hancock (00:50:49): We've ended up building an interesting capability as well to go along with that, right? Which is like I say, we make technology. The vast majority of people at Hub Stream are software engineers making the product. It's one big product that used by everybody and then it's tailored to the different domains. It's like a model driven platform, much like imagine if you took the power BI approach of building a model and then you applied it to an active process. So you'd need to have descriptions of the information and things like that and visualizations, but you'd also need workflows and rules and business logic and stuff like that. Take that approach. That's really what Hub Stream, the direction we've gone. So it's a cloud service, it's model driven. So we can go and provide it to people, we can tailor the model to their day to day activities and then they can use that to do the investigations, which may be radically different.

John Hancock (00:51:41): So to pull that off, we started out with a few different models. We had consultants for a while, that sucked because every day, the bit one thing would come up and then we'd have to say, "Sure, we can help you with that. Let's put together a project proposal." So that model sucked and really got in the way of actually helping customers. We also had a support thing where you could phone up and say, "Hey, I've got this problem," and support would help you. Eventually we learned from what other people are doing in SaaS generally, but particularly in SaaS where it's trying help people achieve a business outcome, which is we've got a customer success team now. And so we stole all of the reproaches and terminologies from the big companies that went ahead of us and customer success is basically a team that is set up to do what it says, help your customers be successful. They are funded essentially by the fact that your customers will renew. So if you help them be successful at the end of the year, end of the three years, whatever the contract is that customer's going to sign up again for another subscription because they've been successful using the software. So you can fund your work to help them as part of that recurring revenue bucket. So our customer success team is really the key to what we're doing.

Rob Collie (00:52:49): Can I be a little difficult for a moment and say you had your consulting and support and those didn't work. So you renamed those two things customer success, and that works. What is it that's truly different?

John Hancock (00:53:05): Yeah, that's a great question. So the first and most important thing is how's that group funded? Where's the funding for this? Like for consulting, it's easy. You have to make 20% margin. So if you do fixed price things, you figure out how to quote people that plus 20% and that's how you... So you can go, how do you know you can hire another one? Well, when you've got more work than you have opportunities, you go hire another one, right? And if that goes the other direction, sorry, Mr. Consultant, or Miss consultant, you're out of here. We don't have enough work and you've been on the bench for three months. So that model, I understand having been one of those consultants. So for customer success, the big difference is we price this subscription knowing that we have to fund X amount of people to help be successful.

John Hancock (00:53:49): So right from the get go, our pricing model, we don't have a line item, we don't have hourly rates generally. We've gotten completely away from that. Now our subscription, when I pitch it to people, they go, wow, that's really simple every time. We work with you we figure out what your domain is, we figure out how many users you've got, what kind of data you've got, what the volumes are we talking about? And then we use patent matching to go, okay, you're like this customer that we've had already. And so we know roughly how much time we'll need to spend to make them successful. And then the way that model works in practice is you make a huge loss in year one because they need a ton of help, right? They're getting going. Year two, they start to get involved in the system and there's a lot of stability, okay, but they do still find these new opportunities like, oh, we realize we've now got this hub of all our information, but there's this thing over here that we haven't incorporated yet.

John Hancock (00:54:41): Year two, you start to balance out and break even. And then year three, then you can be profitable because by then the customer has figured out what they want to do and they're actually incentivized to churn more slowly in terms of their requirements, because they've got a bunch of people all using this thing and they start to only move in smaller increments so that the wave of innovation slows down but still keeps going, but in smaller little increments. So once we figured model, now you can fund somebody whose job it is to help your customers be successful. And on a amortized all customers long-term basis, you can balance out the books. And so that was a huge thing for us. Once we figured out that we needed to incorporate that mentality into how we price the product, that was what shifted us over into a much better place.

Rob Collie (00:55:31): John, the more I hear from you about all this, the more I'm glad that you're on the side of good, because your ability to play the long game is something else. And it's one of those, again, we really need to write the comic book. Because somewhere there's your alter ego that's playing on the other side.

John Hancock (00:55:54): That's a good one, man. Yeah.

Rob Collie (00:55:56): So look at it. In the early two thousands, you saw where the world was headed in terms of law enforcement, something you said earlier, you just saw it all ahead of time. I recognized the way that voice you did see it, I believe it. As a humorous aside at the beginning you said, well, I was the number two data guy in Canad, right? And you'd been trying to catch that number one for so long and you couldn't. And so instead you said, "You know what? Screw this. I'm going to go to Microsoft and help build the platform that destroys the domain in which he was number one. I'm going to take that supremacy from him." And once that was done, that's when you went after crime.

John Hancock (00:56:36): Yeah. It sounds great. Actually, to be honest, I was just trying desperately to get to know as much as that guy did it all the time because it was such an inspiration, oh my God. When you get a really good data nerd and they walk in and they look at the query pan with one eye closed and they're like, "Yeah, all you need is an index," and all the lights just come back on, everything's great.

Rob Collie (00:56:57): So when we started talking about databases, data warehouses, things like that, we got pretty deep down that particular rabbit hole but I want to bring us back to what is it that Hub Stream is doing? I'm sure it's doing lots of things, but at its core, it sounds like you have a tremendous amount of disparate types of data and more in new types of data every day that you need to be correlating. If you go back to the bob54@hotmail example, right? Am I on the right track? Is that close to the spinal column of upstream?

John Hancock (00:57:33): Absolutely. And it goes into the big challenge with investigative data, right? Is that when I came into the space, maybe just tell a little story, that'll probably be the easiest way to frame this. So I was working, helping banks, right? So you're in a bank, you go and collect the transaction data, you bring it up there, you add it up, you slice it, you dice it, you're good, right? It's facts. Like you said before, it's what it is, right? So in investigations, there's this weird thing that happens and I learnt this...

John Hancock (00:58:03): In investigations, there's this weird thing that happens and I learned this pretty early on, very publicly and embarrassingly actually, which was, I was starting to get into the space and we had finally got to the point with our conversations with the police agencies we were working with, where we started to put together data sets and starting to look into them.

John Hancock (00:58:18): And as you said the first thing you think of is correlation. Okay, let's go and see which ones match together. And so we loaded it all up, matched all the things together, we looked for email addresses, we looked for phone numbers, we looked for all kinds of information that would relate things together right? Correlations. Links. We're super excited, we finish up the query, it completes, and it pops to the top the most correlated thing in the system.

John Hancock (00:58:42): We're like "Wow, we found the most correlated thing and oh my God, it's like correlating 26 different cases. This is like a criminal mastermind here. We have found the criminal mastermind. This is awesome." And the cops go "Well, who is it?" And I say, "It's [email protected]." Ah Shit, in that moment I had this radical revelation again, which was, "Oh my God, criminals lie."

John Hancock (00:59:06): Oh yeah. The cops that were there, they're looking at me like, "Yeah you, Microsoft guys are smart in like a very specialized kind of way right?" Because, yes my brain criminals lie. That's absolutely true. The data that you're getting in, the signal that's coming into you, is full of incorrect data as we would call it in the old world. But in criminal world, it's people trying to cover their tracks. They are literally out there on purpose trying to deceive you.

John Hancock (00:59:34): They are trying to make these things not facts. Because if you can find enough facts, you can go kick in some doors right? So the interesting part about Hubstream is, trying to say, "Okay, we know we're building you your day to day system, because that's the only way we can really influence your behavior. So you sign on, now every single thing you do from receiving a bunch of reports into you, all the way through to the end game of arrest and beyond is in the system, you're in Hubstream, that's what you're doing." So now in order to help you, the trick about Hubstream is how do you actually show them the relevant information, even in a world where [email protected] is a thing? And so at its core what hub stream's doing basically in addition to the usual, line of business type system things which is transactional, audit security, all those kinds of workflow, that's a huge thing for us, workflow automation, all that stuff.

John Hancock (01:00:27): The interesting data part of it is about taking in all that different data from all these different sources and then trying to match it together in a way that helps you see these patterns while also excluding links that are really not useful to you. So that's where the system tuning and all the rest of it, all the user experience is really built around that. Because the superpower that you actually have in the scenario turns out not to be technology, it turns out to be the human. So the human staring at the screen has these superpowers, which is that they are trained, they are skeptical, they're cynical, they know what people do, they know if you look at the word [email protected], they understand, "Oh yeah that's a non-validated email address." Like people could put in whatever they wanted to. So of course that's what they did.

John Hancock (01:01:11): So the superpower is the human and that's actually what makes the arrests. When people talk about Hubstream being involved in things, I'm always humble about that. Because, we are not the ones making the arrests. All we're trying to do is, we augment the human that's sitting there in front of the screen. You try and bring all the relevant things together, you try and tune out the noise and you try and actually empower that person to do what they do well, which is investigate. I'm not one, but I know the good ones. And all we're trying to do is power them up. The data nerds among those are probably thinking, "Well, how the hell did you do that? how do you actually model that?"

John Hancock (01:01:45): So the cool thing about this actually in a funny way was our experience on Power Pivot. Because in Power Pivot, we spent all this time trying to build models that represent the real world. And now Power BI today, as you said, the beating heart of all that stuff comes back to the model. It's a model driven thing. So when we started building the Hubstream technology, when Devarajan, our CTO, and I were talking about it and we're starting to think this one through, we decided to do something a little different in this world. Which is, we said, "Okay, we are actually starting with a model. This model is going to be driving more than it would do in a BI scenario, which is mostly about understanding data, it's going to be put to even more work than that because it's going to also say what workflows happen and what permissions there are." It does a ton of stuff, but it's all built around an idea of a model at its core. And so the software is set up so that tomorrow, when you come in and Bitcoin is now a thing, instead of having to re architect your system and build these new things, you go in and change the model. So you go and say, "Right now I have Bitcoin as a thing" or payment method or whatever, you end up modeling it as, and you go and deploy that into the system. And voila, now when you sign on next time you hit refresh, you sign in in the morning, "Oh, now Bitcoin a thing."

John Hancock (01:03:02): So that evolution, the system's set up to evolve in that way. So it's all completely model driven from the get go. Literally that was the next thing we did. The endeavor after our adventures in BI, we spent all these years in BI so we are a system for investigators built by a bunch of BI people. So it is in many ways the spiritual successor to the semantic models we've done in BI for many years, but applied to this particular domain.

Rob Collie (01:03:29): That's fascinating. Okay, I'm not necessarily smart enough to have understood everything you just said in that compactible form. But I do recognize when I'm hearing something intelligent.

John Hancock (01:03:42): It's just my accent. It fakes everybody.

Rob Collie (01:03:48): I think we're probably better off sneaking up on people. We give you Southern United States draw and people really wouldn't see you coming then which is basically half my family by the way. Smart people with that voice, they kind of catch you off guard sometimes. But yeah, you fit the role. You live up to your voice John. Don't you fret.

Rob Collie (01:04:06): So there's code in your product, it's a software product. But are you saying that when something new comes along, like a new type of data, you can go and sort of integrate that into the model without having to really mess with the core code of the rest of the software.

John Hancock (01:04:27): That's exactly right. See, here's how that's even cooler than you might think at the beginning. We thought of this originally and thought "This is a cool idea." And me and Devarajan super excited with our whiteboards and tap dancing and, this is great, this is going to be awesome. Like I've never seen anyone build a system like this before we are going to go build it like that. We're going to merge the best of this BI model driven thing with the world of investigations. Voila, a whole new thing.

John Hancock (01:04:52): It turned out way better than we had hoped even. Because, it turns out that every organization we worked with has different nuances to how they treat that information. Not only that, but if you look at the different domains of investigations, investigations turned out to be bigger and more complicated than we thought, basically more diverse, more wide ranging.

John Hancock (01:05:12): We found investigative scenarios that we're now used in that are radically different than what you would've thought of originally. So for example, one of our biggest projects at the moment is to build a national transportation safety investigation system. So, that's an entirely different domain. If a plane crashes, you're going to go do an investigation, which involves hundreds of people and thousands of pieces of evidence and all this stuff. So the reason we're able to do that project is because of what I just said there, we have a model. Now we walk in there and there are a few concepts in that model that are the same from online crime versus transportation safety investigations. Like, you have an investigation, you have notes, you have documents. But there's a ton of stuff that is completely specific to that. Not just that domain but that particular country's implementation of that domain.

John Hancock (01:06:02): And because we're a model driven thing, the process of actually deploying the technology is, Hubstream brings the technology. We usually have consulting partners for a big project like that, who are sitting there talking to the transportation safety investigator saying, "Oh, there's a thing called an aircraft. Okay. What kinds of information are you tracking on and how do you use it?" And they sit there and they set up the model and they hit deploy, time goes by, refresh and then they see, "Okay, great. Now I can relate aircraft to my investigation. Now I can relate all these different elements over to it."

John Hancock (01:06:34): So that's actually has been the value of what it was we built. We wanted to have it able to set up to evolve quickly. And boy, was that a good idea. Because like I say, real world is always more complicated than you think. And that's actually been the way that we've got the technology out in increasing scope and increasing variety of investigations because of the fact that every customer you walk up to that's choosing Hubstream, If you go and look at what their model looks like and what their screens look like and their workflows, it's quite different. Even within theoretically the same domain, two countries, two different agencies, they may approach it differently, they may model it differently.

Rob Collie (01:07:10): Wow. Okay, so if I didn't know you as well as I do, and I know you well enough, if you were anyone else, if you were a stranger after what you just said, I would've gone, "Oh, come on that's bullshit."

John Hancock (01:07:27): Wow. I finally hit the bullshit meter. Yay.

Rob Collie (01:07:29): I know. No, but I mean you've actually hit the holy grail. So what you just described in some sense, it's kind of like every computer scientist's dream. Especially like a graduate computer scientist, before they go out into the world, there's just this thing with computer scientists, and I was very much guilty of this at least for the first half of my career at Microsoft, which is "Ooh, everything can be reduced to these symbolic models and really everything in the world is best understood through a mathematical lens." And it's all, as I said, that's all bullshit.

Rob Collie (01:08:03): You've acquitted yourself of this crime.

John Hancock (01:08:05): In advance.

Rob Collie (01:08:05): Multiple times on this recording already. It is about the humans right? And so that thing where you're saying like, "Hey, we can come up with this abstract modeling approach that will flex to completely different domains. And the software kind of just sort of keeps working." That doesn't happen, that plan has been drawn up. There's right now, 5,000 people in the world right now putting the finishing touches on a plan of similar abstractness and not one of them is going to work.

John Hancock (01:08:40): Yes, I completely agree. They're going to use the word ontology.

Rob Collie (01:08:40): Well that's right. Yeah.

John Hancock (01:08:44): Which is of course fatal. So I think there are some factors as always, because here's the thing, we're not actually trying to solve everything. So what I see out there a lot at the moment is this whole, no code or low code revolution where to some degree they're trying to make claims about that. But the thing is that to truly do this as like a universal version of what I said, that can handle apps for collecting project information versus the analysis of, or the sale of tangerines or whatever it's. To do this on a super generic way, I think is impossible in our lifetime. It's not going to work. But what we did was we said, "No, we are focusing in very specifically on investigations" which when you're Microsoft sounds like the tiniest of niches in the universe.

John Hancock (01:09:33): But in the actual world, there are thousands, tens of thousands of people who have investigator in their DNA or their job title or their team's name. But by focusing in on what to other people is a very small niche, we can take out of that the basic concepts, not the domain concepts, but the basic ideas of what those humans are trying to do, because that's what repeats. And that's what we can actually say, regardless of whether you're looking in detail at an aircraft, that's shown up in multiple crashes, or looking at an IP address that's shown up in 400 reports of malware, what you as the human are doing, as an investigator, bears an awful lot of resemblance between those two different scenarios. You're trying to manage a large flow of information, you're trying to use your powers of observation, your skepticism, your natural curiosity to go after connections, prune things that don't make sense, add things that do.

John Hancock (01:10:28): So that's why I think we've actually been successful with this approach. I wouldn't say that some no code platform could do what I just said and solve every kind of use case that's out there, but having had the track record over the last six years I can tell you that investigators share a lot of common DNA. And so by focusing on the user side of this, you can take away the things that are truly irrelevant actually, which is the structure of the dates, what the domain is, what those concepts are. You just take that away from the core of it. And that becomes the flex part. That's the part that flexes, but you actually sit down to Hubstream, what's common across all of our customers is that they get in a lot of information, they need to resolve it, they need to investigate, and that's actually the common thread the ties it altogether

Rob Collie (01:11:11): For each sort of type of information, are you using similar storage for each kind? What is your storage technology of choice?

John Hancock (01:11:23): Yes, it's totally proprietary. And I cannot tell you that without... I joke a little bit, but actually, boy have we spent half a decade working on that question. And every generation of our technology, we go out there in search of the universal answer. And actually there isn't a universal answer to that question. So what we actually end up doing is taking a combination approach. Every time we come back to the practical which is, you know what, SQL server, SQL Azure or SQL something is awesome for these things. Blob storage, variations on that, awesome for these things. A NoSQL world with a flexible schema less thing, awesome for these kinds of things. So we end up basically assembling and building integration, building layers on top of, building translations, building, building, building to knit these things together.

John Hancock (01:12:17): Now the world of data at the moment, data platforms are constantly... every time I go out there, someone's promising that they've solved this. They have the truly heterogeneous thing. Yeah, that's not true. We haven't found anything that will truly answer that question. I'm sure that all the vendors out there'll be like, "No, ours is the one we have finally solved this truly heterogenous platform thing. It can do NoSQL, it can do Blob storage, it can do relational." Yeah, not for us it can't. So the answer is, the world's a complicated place, our storage layer is complicated. Every generation, it gets smarter and uses more and more cool things. But at the end of the day, there isn't one answer to that problem.

Rob Collie (01:12:55): Well, so now I can check off my check mark is that I finally got close enough to something sensitive that you can't tell me all the details. And I love that, but you still answered the question though.

John Hancock (01:13:04): I didn't say how we solved it, but I'm telling you, it's a knitting together of the best of things we can find at the generation that we're at right now and next generation will be the same thing again.

Rob Collie (01:13:15): Yeah I found that meal to be very satisfying though. That wasn't empty calories. I'm a satisfied customer. Moving on, you mentioned the similarity between investigations, been reading a lot about COVID lately and questions like how effective are the vaccines against the variants. How long does the immunity last, can you get COVID twice, all these sorts of things. And every time a real medical expert gets on the hook with a question like this, the answer's always, "Mm-mm (negative) you don't really know." It's really squishy, it's not facts. And I was wondering do you think that at some point there might be a medical investigation, like this might be within your purview?

John Hancock (01:14:02): Absolutely. We actually have a project with one of the big pharmaceutical companies at the moment. That is a really tricky area I have to say, we worked with law enforcement for a lot of years, which I thought was complicated and had a lot of regulation, had a lot of procurement issues. Boy, does healthcare blow them out the water. They are the most complicated area that we've ever worked in. There's a lot of constraints on healthcare generally. But I also see when I go in there that because of that, it's kind of been a [inaudible 01:14:33] into some degree. Like there's not a lot of, I would say from what I've seen, this may not be universally true, but because of all those constraints on them, there's a real sort of limitation in terms of what it is that they can get done.

John Hancock (01:14:44): So I think there's actually a ton of opportunity to make the world a slightly safer place by going in there and starting to help them out. For example, law enforcement for many years has not been able to touch the Cloud. Cloud, super scary, not going anywhere near it. Well, in the last three or four years, that's really started to shift and people are moving over. For pharma, it's still not the case. They're still stuck within the constraints. I'm just looking at that as probably one of the bigger areas for innovation in the future. Not just us of course, but all the other people out there who have technologies that they can bring to bear, because it's such a critical part of the world. You look at things like counterfeit medications in Africa that kills a lot of people, counterfeit cancer medications, boy is that a scary term.

John Hancock (01:15:28): So I think that there's a huge opportunity to go in there and bring some of the investigative technologies and approaches from, even from things like investigating cyber crime and bring that into the world of medical and healthcare. And so that's definitely something I'm really looking forward to get further into. But boy, is that a tricky problem.

Rob Collie (01:15:46): I bet, but I would've never guessed that it was harder than law enforcement.

John Hancock (01:15:50): Yeah. I think it's because law enforcement's regulated, but healthcare is just one of those... Such a regulated world. Every step that they take is governed by regulation. It's all private organizations and because of that, how we as a society have chosen to maneuver is we say, "Okay, a lot of the innovation and services are going to be provided by private sector companies but healthcare is a tricky thing, human lives are at stake, so you've got to regulate the hell out of it." So It's this weird combination of private sector, but super critical to as a society and therefore highly regulated. That's been the interesting and fun part about tackling that area.

Rob Collie (01:16:29): At first blush, I would think that the regulations would make it easier because everything kind of fits into, one size fits all in a regulated... like wild west would be harder. Are you saying that regulation suppresses data? Is it like the HIPAA stuff? You can't link it back to the individual, so you really can't correlate.

John Hancock (01:16:48): Well, it's more like every step you take has regulation. So whatever it is you're doing, you can't color outside the lines because the lines are everywhere. You're surrounded by lines. And so, it's not like you can't do anything, but everything you do, you have to be aware and conscious of the regulations. And so we spend a lot of time talking to the [inaudible 01:17:11] customer about, and they tell us things that surprise us constantly. Like "Really? If you do that, you have to do this? I'm quite surprised by that." So once you get into the actual day to day operations, they're not free to move at all. We've decided as a society that they are going to be constrained and every step they take, they're going to think it through, they're going to think of ramifications. And so that's what makes it challenging.

John Hancock (01:17:34): We are learning from them. I'm sure it must be a big learning curve getting into healthcare as a starting point, what exactly the spaces that they have to maneuver and what the responsibilities are that they have as a provider of healthcare solutions, whether that's medication or something else. So for example, if they're find out that something has caused an adverse effect, there's a lot of regulation that says what they have to do next, they can't just go "Oh, interesting." And move on to the next report, they have to actually do something.

Rob Collie (01:18:02): And there's also things like even just taking this data from over here and that data from over there and putting them on the same server at the same time. If you do you that you've committed a felony or something and without ever knowing it.

John Hancock (01:18:15): That's exactly it. So everything is constrained and that's what's the challenging part about that. But like I said, I really do feel like that area of our society is really ripe for some help, in terms of bringing in technologies to help people understand what's going on. Because there's a lot of crime out there that revolves around this stuff, counterfeit medications is just a starting point. It's a massive, massive thing. As someone who grew up in Africa, it's a lot more often in the news over there than it is in the U.S. but it's definitely something that causes a lot of pain and suffering in the world.

Rob Collie (01:18:50): I believe it. And I do believe that it could use a lot of help. That's for sure. I asked that question about the medical field, thinking maybe I was going to stretch your brain.

John Hancock (01:18:59): It would've done a year ago. Boy, has it been a year.

Rob Collie (01:19:01): I should have caught you a year ago. I'd have felt better about myself. Instead you're like, "Yeah, yeah all over it." Which is awesome by the way. Okay, so something about the way your technology works, the way that the data model is separate and flexible from the code itself, the actual instructions that you've programmed into the software over time. My first job at Microsoft, I was a software tester, a test engineer on Office 97 setup.

John Hancock (01:19:29): Okay. Thrilling.

Rob Collie (01:19:31): Oh my God you have no idea how exciting it was. For the first six months I was just blissful. I was like, "Oh my God, I'm working at Microsoft." I'm like, "Oh, Office 97." And the setup scripts, the setup engine, Office had it's own setup engine that it wrote for installing software. It was called ACME. And by the way, this became the source of a band name that I had never formed. I never formed the band, but we did name our cross country racing team at Microsoft after this. One time I saw a build report for ACME, something about Swedish ACME had failed, the Swedish language version of the ACME installer had failed its build. So that we named our team Swedish ACME from that point forward, just sounded like some weird Wile E. Coyote thing, but there was another installation engine in development.

Rob Collie (01:20:22): The problem with ACME and really with all installers at that point was that it was procedural. The scripts that you would write to install stuff would be like, go write this registry key, go put this file there, go do whatever. Like it was a set of instructions, which is what you would expect to go install the software. Guess what? Then you need to run the other modes like reinstall, to repair if something has sort of gone wrong or uninstall, heaven forbid you need to run an uninstall. Because the uninstall script was then a completely separate pathway through the... It was separate code. You had to write code that says, go remove this file, go remove that registry key or whatever. And so what you found was is that the setup part, the setup script was actually well tested and well debugged relative to every other pathway. Uninstall almost never worked.

Rob Collie (01:21:24): And it was just so dumb in a way. So a couple people on that team had a really smart idea, which was to take installation, take software installation and make it a non procedural language, make it a declarative language. Instead of writing code that says, go write this file, put it here, write this registry key. They came up with this idea that there'd be a database that describes the application. The application consists of these groups of files and registry keys, and shortcuts and things like that. And then we built a separate engine. Its job was to read that database and decide how to install things, read that database and decide how to uninstall things. And then we could test that code, the uninstall reinstall code in the main engine, we could test it and really thoroughly debug it, make it solid. And then all that people had to do was basically describe the contents of their application to what's in these databases.

Rob Collie (01:22:24): It was my first brush with a concept in the nerdy ontological sense, it's called declarative programming. It's hard to do, it's hard to build systems that work that way. This one ended up being called MSI windows installer. And it actually worked, I mean it was a hard project. It was the first project that I was program manager on. That was my first PM job. And I was allowed to be PM of that project because no one else wanted that crappy job.

John Hancock (01:22:56): That's foundational though, that's big technology you end up building there. That sticked around for a long time.

Rob Collie (01:23:01): Yeah. It's one of those things that is now amazing because you never notice it and that's what you want in a setup technology and it is a game changer. It actually has made windows far easier to manage. It really has been a great success. And it's also how I got involved in all kinds of places that I shouldn't have been at when I was 25 years old at Microsoft, I was interacting with some people that I had no business interacting with and it showed. There's something about your story, in another show that we've already recorded, but it hasn't gone out yet. We were talking to Chuck Sterling about these power virtual agents and how the code of the front end, the chat bot doesn't change. You're just feeding it new topics behind the scenes. So it's this more of a declarative data driven, leave the code alone as much as you can. And when you can build something like that and it works, it's a home run. It's hard to do.

John Hancock (01:24:03): Yeah, It is. We are fortunate in that we started with this idea in mind. I don't fancy anyone's chances of grafting that on later, if you haven't started out with that in mind, good luck to you. But the huge advantage we have is I get out there now and I see a couple of different approaches to building systems for investigations. The one is, that sort of traditional case management thing. And when you go out and look at that stuff, they're written code that says, "Okay, I need to manage a person." The page has this on it, it has a name, it has a picture, it has a save button and they've written the code and everything is all tied together. So you look at those systems and of course they're all horrifically out of date and slow moving. Then I go into large government agencies and there's a name of a company that I shall not mention, but they're a very big multi-billion dollar provider of investigative solutions basically, platforms. When they're out there and they go into their clients, they deploy the technology, they have these sort of black cloud engineers that come in and do the data science and set it all up.

John Hancock (01:25:04): And then the customer gets going. And of course the world changes and Bitcoin happens or something else happens. And then they go to that vendor and they say, "Yes, we now need to have a column that says Bitcoin." And they go, "Yes, we'll get to right on that for you. That will be $20,000 to add that field." There's this whole consulting driven world where they're basically building these things, using smart data sciencey engineers. All this is the world where everything is super static and written by developers 14 years ago and imported over through generations.

John Hancock (01:25:35): Those two worlds don't work when you go out there and you see it in action. So like I say, we're fortunate to have come from the background we had, the weird sort of unique journey that we got to this place, but then day one was saying, "Okay, we aren't going to go and write hard coded screens in the system and we also sure as hell are not building a consulting company." I was a consultant. I knew I wasn't building a consulting company from day one. So what else is left given that the fact that you're a bunch of guys that just came from Power Pivot. Hey, you start with the model right? Everything has come around that model. And that's actually been the sort of superpower behind it all.

Rob Collie (01:26:10): In all sincerity. Congratulations, not just on the product, even just the nerd in me, the nerd that's sort of been tamed and turned into more of an ambassador to the humans. The system that you've built, it just sounds gorgeous in all the ways that count. And it also sounds like you put together an amazing team.

John Hancock (01:26:31): Those two things went together definitely.

Rob Collie (01:26:34): You're going to be famous. Don't forget about us, John.

John Hancock (01:26:37): I hope not. I'll settle for just really, really rich.

Rob Collie (01:26:40): Okay. Do you ever worry that at some point your system will be so widely used that criminals will start to try to understand how it works so that they can cover their tracks better against it?

John Hancock (01:26:54): Not really, because it's true every single day that there's an arms race, and the arms race is between the criminal and the investigator. So we write tools that help.

John Hancock (01:27:03): [inaudible 01:27:00] is between the criminal and the investigator. So, we write tools that help the investigator, but really what's actually going on is people are on a daily basis trying to figure out how to cover their tracks from those people who are society as sent to enforce the law. We are just a part of that machinery. We help build the platforms and the tools, but there's nothing about how the Hubstream system works, that is some kind of unfair advantage to the investigator. Like I said, the only real secret or the truth behind it all, is that the smarts come from the human in front of the screen. Every day investigators show up to work and the criminals are doing something weird that they weren't the day before. That investigator has to adapt or they have to retire.

John Hancock (01:27:46): We're basically supplying the platforms that those people use. It's never going to be the case that there's someone's going to find out something about how investigations work, just by looking at either our technology or even the models that our customers have put together. What you really got to do if you... as a criminal, if you really want to get ahead of law enforcement, is you've got to reverse engineer the brain of an investigator and figure out how to trick them. That's actually what criminals do all the time. They're constantly figuring out ways to look like something else. That's what they wake up every day and try and do. They're covering the tracks from the other humans, not from the technology.

Rob Collie (01:28:20): Okay. Okay. Fine. But if you ever snapped. If you just had enough, right and you decided to Break Bad... you'd be a better criminal now, wouldn't you? You'd be able to avoid your own system?

John Hancock (01:28:34): I don't think so. Here's the thing, you spend all your time in one brain, right. You're going after one thing. What I am is really understanding, and empathetic, and compassionate, about the challenges of the job of an investigator. Particularly in the darker areas where they show up for work every day and they're looking at pictures that the rest of us humans would see one in an entire lifetime, and be traumatized. They look at hundreds an hour, thousands an hour, tens of thousands an hour. Where the sort of weird learnings that I've had over these years have a lot more to do with a real understanding of the job of what investigators have to do. I think, if you were trying to be a master criminal, you should probably get up every day and try and evade law enforcement. That would be the way to go. Just like everything else you'll have to put in your 10,000 hours of evading law enforcement and then you'll get better and better at it. Maybe you'll go hang out with some master criminals. That's really the way to go.

Rob Collie (01:29:33): I don't know, man. It seems like that knowledge of their humanity, those investigators, their deep humanity and knowing what makes them tick-

John Hancock (01:29:41): Maybe I could turn it against them is what you're saying.

Rob Collie (01:29:44): That's precisely what would make you so dangerous. Don't ever become evil John, because by the way, your accent would work well with that too. I could see a movie coming. No problem.

John Hancock (01:29:57): Okay.

Thomas LaRock (01:29:58): I'm fascinated by the work, the data. Part of my journey has been historically production database administrator, and that's kind of what I was when I came on Rob's radar. Although, I had jumped to a software company. Part of my journey is the last four years now, four plus years, I've been getting much more involved in data science. I've probably touched Excel more in these past five years or so, than ever. What you've built, I just have so much respect for the challenge you've taken on, how you've done it, and the work you're doing is important work. Tip of the hat to you, good sir.

John Hancock (01:30:48): Yeah. And this is a probably good opportunity to talk about the other success factor on this, which is definitely team, right. That was the other part you said, we were talking earlier about the whole customer success journey. We had to learn our way into that, right? All these things. We've screwed up things. We've messed up. We've made mistakes. Then we've kind of had two course-correct. I'm really happy where we've ended up. I think the key thing about all of this is definitely team. If I look at the successful products or projects like this, generally, you got to have a different set of personalities all working together to make it work. My center of gravity is about product. It's about... I can spot the problem that needs solving and I can come up with a rough parameters of what a solution's going to look like. That intersection of real world, and then translating all the problems you could go solve into, we're going to focus on this one.

John Hancock (01:31:42): These are the shape of it. This is the parameter. That's one element. I don't do so many other things. That's where I had such a great success. I went in mind, all of the relationships that I built in those great people at Microsoft. I found the people in that domain that kind of filled out the requirements. In addition to product, you also have to have design, which I maintain is different. Somewhat controversially, sometimes. I've worked with this fantastic guy, Joe Marline, for many years. When he and I are working together, the lights come up much brighter because he understands the human element and he pushes towards that all the time. He's just this great creativity that comes into figuring out how are we going to do this? What is this going to look like? How's it going to feel?

John Hancock (01:32:25): That's just such a key piece. Then the other part of that is, I've... I was a professional programmer back in the day, but I was never at the level to build these big systems. Luckily at Microsoft, there's a lot of people around who can build those big systems. Some of them are not as much fun to work with as others, I will say. I had the really fortunate experience, as did Rob, of working with an absolutely fantastic guy [Devarajan Muthukumarasamy 01:32:55], who is just an awesome dev lead on our team. Then had grown up into this guy who was doing the architecture for these big cloud services and things like that. He was one of the first people to start architecting these large scale things for BI. I was fortunate enough to have a really good relationship with him and he's come over for all these years as our CTO.

John Hancock (01:33:14): That means that I now have the super power. I have this fantastic design person who can help figure out the solution. Then I have someone who can build these fantastic large-scale systems, who can manage teams, who can build a team, keep that team productively occupied. Then, the third part of that is the customer element. The third sort of part of this journey is about, how do you get people who can go out there in the world and understand the domain, which is the crime domain? Also have this technology sort of consulting, consultative experience. That's our director at Na VP of Customer and Partner success, Paul Whitaker. He came from military background. Then he went and worked for Microsoft consulting for many years and was helping the FBI to do their systems.

John Hancock (01:33:58): I met him at Microsoft and was just so impressed by his ability to join those two things together. Technology and the customer and get them good solutions. That's really, I think, where I lay the sort of credit for how this has gone well, to the extent that we've helped our customers and built the company is, to solve the problems, which are super excruciating hard. As you said, they change all the time. They keep evolving. That's been a real fantastic element of it, is having all these people together who can bring those different sides. The technology, the design, the customer angle, and product, and then build out the product around that.

Rob Collie (01:34:32): I want to emphasize something you said about team. I completely agree. This is something that in general, I think the entire society, we should be reminding ourselves of this over and over and over again, multiple times per day. We need it so much. Remember that a single human being alone on the wilderness is a meal for a 60 pound cheetah. We're just so helpless. 10 human beings with pointy sticks though. Hmm. Drive the wooly mammoth to extinction. This belief that you can survive as an island or whatever, or it's us versus them. All of that. Is just so dumb. It's just so dumb. By the way, what's the most dangerous thing to 10 human beings with pointy sticks? Another 10 human beings with pointy sticks. If you're not cooperating, whether at a company or at any level of organization, cooperation is just insane in terms of what it can do.

Rob Collie (01:35:38): Having started at Microsoft right out of college, I sort of just joined this massive organization. I got to take it for granted from the very beginning. Starting essentially from zero and building something from the ground up is so educational. I will never view the world the same again. I will never think of myself the same way. The way that I view my own capabilities is different as a result of it. Of course, just the way things work. You would never mention yourself necessarily in the same glowing light, but you are an incredibly important part of that at same team. You, John, that you just talked about. Those people had to leave. Deva had to leave the Microsoft job and go with you. At some point, right? He had to believe in you. I understand that.

John Hancock (01:36:29): Yeah. It's a huge learning adventure. I mean, I had done a couple of companies around us, but this has been such a journey of education. It's interesting hearing you point out the office thing. Office, at Microsoft, was one of those organizations that did recruit right out of cottage and grew people up through the office system. Whereas SQL server, where I landed, was very much not that. They kind of tended to recruit from people who had experience and then came in. It did create this differing culture. When I left, I had the opportunity to start thinking through how we would do that. Where we've ended up actually, is, as we scaled up the company, we've actually taken the office model, which is we've been recruiting the vast majority of our company, comes in from as university graduates now.

John Hancock (01:37:14): We bring them up in engineering on Deva's team, and then they grow up as a Hubstream engineer through that cycle. We've invested all our energy into basically coming up with a model, that we can scale. Where we bring people in and then we mentor them and we grow them. Then that team grows up. That's not how we started out, because that's not what my background was. It's become a very interesting way to grow the company in terms of how to scale all those different elements I was talking about and still have the way to keep your culture whole and keep growing that kind of office model of growing your own constituents in your own culture. It's hard. It's difficult. It can be slow. At the end of it all, you get more control over how that company's unfolding.

Rob Collie (01:38:00): It's really funny, that I grew up in the office system, and at Microsoft, you grew up in the SQL system. You've chosen the office model and we only hire experienced, in our company.

John Hancock (01:38:10): You flip over. I think the other element though, is that in customer success, it's different obviously because you know, having some experience helps. The most of my company is actually our engineering. They're building a software product. For that, the office model works fantastically. You can really go hire fantastic people at the very beginning of their career. If you can identify the right ones, if you can give them the right support, then you can grow them into these great teams without having to go out and do recruiting for more senior people. You can grow your own. Yeah, I've very much appreciated that model. That's something we've been doing well over the last couple of years, I think.

Rob Collie (01:38:47): We cheat a little bit because we sort of halfway, intentionally, halfway, unintentionally, we sort of select for the culture that we want. Our consulting team is only... is a 100% experienced hire thing. One of the two questions I wrote down, that I want to make sure I get to before we wrap. One of them is, what is the profile of the customer success managers? What are those people's backgrounds? How do you recruit them? Where do you go looking for them?

John Hancock (01:39:11): Yeah, that's one that we'd... I wouldn't say there's a template for. That one's a super tricky hire. We're growing that team pretty slowly. One at a time. Everyone is a custom. That's a one where we're trying to find people. We look at them, we look at our customers and we try and see does this puzzle piece, which is a complicated, unique, special snowflake, does that fit into the team that we've got? And how they're going to grow out? That one is, it's very hard. I would say, we will probably have more failures in that area than anywhere else. Trying to recruit into it is super hard. You can get a bunch of benefit out of different angles. If you have a lot of law enforcement experience and investigative experience, that's great. But do you necessarily have the technical experience to help somebody who knows how to investigate, but doesn't know the technology to get more successful? Right?

John Hancock (01:39:58): On the other hand, if you have a lot of technology expertise, which you don't know the domain very much, well, that's going to take you a bit of time to get, to be able to talk to those customers. Yeah, balance is what we're trying to go for. I would say the hardest role that we have at the company to scale.

Rob Collie (01:40:12): I can imagine that. Another thing that I was wondering about throughout this is that, you're a software company. You build software, overwhelmingly, I think on the Microsoft platform?

John Hancock (01:40:21): Yes. Entirely. One of our values is actually that we are completely Azure. We do have some little things that we're trying to get rid of. The Microsoft slow to adopt. Generally, we are pitched to people as no, if you're signing up Azure, which most government agencies are, we are an Azure, a native Azure service. We do all the Azure stuff. Other players in the market are taking the heterogeneous approach where they'll say we'll run on anything. We've gone driving very hard in the opposite direction, which turned out to be a really great bet. Microsoft Azure and government, Microsoft has done a fabulous job in getting to security-level and a level of commitment from government agencies worldwide. That has been second to none. They have won that battle. I would say I would take credit for having correctly forecasted that. I can't accept the real reason we did. It was because we're a bunch of Microsoft people, right? We just naturally built on Azure and now we just tell people it was our strategy all along.

Rob Collie (01:41:17): Yeah. I mean, there's going to be plenty of bad luck in building a business.

John Hancock (01:41:24): Yeah. That was a good fortune.

Rob Collie (01:41:24): You might as well take the good luck. Yeah. You know, you can't... It's like playing blackjack and going, "Oh I lucked out and got the 21 there. I'll give that money back." No you're never going to win if you don't... You have to take the good luck and the bad luck, don't you? A lot of parallels actually sort of emerged in this conversation about Hubstream, as when we had Austin Senseman on talking about Conserv, which is their for-profit. But cultural benefit... I'm starting to sound like Borat, 'Make benefit cultural,' for museums and their collections helping preserve their collections with IOT monitoring. Some of the things you've said, he also has reflected that industry, the museum industry, if you will, if you call it an industry, is the conservatorship type people are very suspicious of for-profit businesses.

Rob Collie (01:42:14): Probably for good reason. Maybe not in his case. If you think about Power BI, let's just talk about Power BI very specifically for a moment. If you wanted to relax it a little bit, you could talk about the Power platform as a whole. If you think of it as a BI platform, you're already kind of... not you, but we are already kind of missing the point. Really, the Power platform, Power BI is a software development platform for a particular niche. It's software development. You started Hubstream seven years ago, officially?

John Hancock (01:42:47): Mm-hmm (affirmative).

Rob Collie (01:42:47): The Power platform was way, way, way less mature. It doesn't even look the same as it looks today. It's a completely different animal now. Today there's sort of a natural progression when you're a software company aimed at a particular domain, such as crime. Whatever it is, right? Is that sooner or later, you eventually arrive at a point where almost everything you've got is custom software built for that. You're not using any of these sort of, Azure as a platform. When you start out from scratch, sometimes you end up using some of the, these sort of general purpose shortcut technology platforms. If you were starting today... First of all, let me ask you a question. Does Power BI appear... in your offering?

John Hancock (01:43:26): Yes. Absolutely. It's actually quite an important component of it.

Rob Collie (01:43:29): I should have asked that first.

John Hancock (01:43:30): Yeah. Well it's for are an interesting reason. Right? Which is, so we provide this end-to-end platform. Okay. Like I say, we do everything from data ingestion all the way through to triage, investigations, referrals out, cold cases, research, trend analysis. The whole shebang. Our whole point is to be the Hub. It's right there in the name. Maybe from Microsoft are not good at naming things, but that one we named right. It's... Things stream into the Hub. That's the Hub. That's what you use it for. The role that the Power BI thing particularly plays for us is, you get this fantastic, highly tuned workflows. All these things go in your day-to-day business. At the end of the day, there's also things that, that particular organization, whether it's law enforcement or private sector or whatever, they need to have certain access to visualizations and scorecards. All these kind of things that Power BI does so fantastically.

John Hancock (01:44:23): The way that we've kind of threaded that needle is we go, here's your end-to-end platform. What we've done to facilitate that the development of these organizations-specific, dashboards, visualizations, et cetera, is you can actually connect up to the data which will provide you in a very Power BI friendly way. You can build your Power BI reports. We've let you publish them back into the application. We've done a very interesting integration with Power BI embedded, where the Power BI dashboard that you build can show up as dashboards. You can actually make custom visuals. When I say custom visuals, I don't mean as in the Power BI custom visuals, I mean, you made a Power BI report that looks like X and you can plug it into our application UI so that if you're doing a case and for example, your investigation has tons of IP history, well there you can have a tile that launches your custom Power BI report that does that kind of analysis that you need to do.

John Hancock (01:45:17): That's how we've kind of layered in what Power BI does brilliantly, which is the ability for normal humans, as you say, with some Power BI experience, to hook up to data and do specifically the thing need to do. They can still publish that back into the end-to-end platform. That's been a... It was a great bet on our part. Power BI embedded, was a bit rocky at the beginning, but I have to give kudos to that team. They've done a really good job over time of making that into a world-class grade, embeddable solution. Into our actual application.

John Hancock (01:45:47): It's part of the thing we sell, which is we go out there and people say, "Oh, what if I need to do this?" And we say, "Well, do you know, Power BI?" And they go, "Of course we know Power BI. It's completely ubiquitous. It's everywhere." So then we tell them, "Okay, well take those Power BI reports and put them into your end-to-end workflow. Hook them up to the centralized hub of all your information coming in. Voila. Now you have all the value of Power BI." You have that put into people's day-to-day workflow, as opposed to just being some separate thing that you go and query.

Rob Collie (01:46:14): If you're listening to this and you're involved in any sort of line-of-business software. I mean, I... Okay. It's, almost cheapening to call what you do, line-of-business.

John Hancock (01:46:23): It's totally line-of-business. That's exactly right. It is the day-to-day system that people use. It's just that what they use it for is, novel and unique. You've got to think about it as, that is the system. That's the line-of-business system. They use that eight hours a day, 10, 12 hours a day. Many of them, that's what they use. When you think about that process, I agree. Line-of-business systems. If you think about the need to also serve their requirements for custom ways to understand their information, you don't go build that kind of stuff yourself, because Power BI has gone rocketing way beyond anything you would do yourself.

John Hancock (01:46:59): The value is that someone can get a Power BI Desktop. It's actually, like I say, normal humans can use this thing. They can create basically a mini application, as you said, Rob. It's really their domain specific programming effectively. Then you can push that back into your Hubstream platform. Now it's part of your day-to-day. It's part of your line-of-business system. That's the approach we started taking a couple of years ago, once Power BI embedded really got going. That was a really good move. I think it's worked out so great for us and the clients.

Rob Collie (01:47:27): You completed my sentence. Exactly. You know exactly where I was headed. Don't be that line-of-business offer company that thinks you're going to build the reporting. Unless, you know that your customers all need basically the same, exact stuff. Sometimes that's the way that your industry works.

John Hancock (01:47:42): Yeah. Like consumer or these large scale companies, where you're solving a specific problem in a specific way. Line-of-business systems, in the modern era, whether you're doing Salesforce or Dynamics or whatever it is. Whatever's in front of the person, I look around and I see tons of these Vertical SaaS companies, just like mine. Vertical SaaS thing is this huge phenomenon out there for good reason, right? You have some team of people who understands a particular Vertical really well. They can assemble a team that can execute really quickly. They can pull together all these Azure cognitive services and Power BI and they put everything together and pretty quickly they can come up with a solution, which is highly tuned to that Vertical. So the trick in that world is not to develop things you don't need to develop. If you're building things you don't need to develop. If you think you're going to go and crack open D3 and do a better job of ad hoc development of custom visualizations. Good luck to you. It's not going to work.

Rob Collie (01:48:35): If I can just edge just a little bit commercial for a moment. By the way, if you happen to be working on a vertical-specific, line-of-business system, P3 would love to help you avoid that mistake, right? We would love to help you integrate Power BI into your offering and offer an actual winning capability for your customers without you spending... It's just crazy. You can spend so much time and money trying to build reporting that works for all of your customers. Just a sinkhole of money and ,never come close to succeeding.

John Hancock (01:49:06): That's exactly right. I completely agree. Now the thing is, luckily in our case, we did have that background, in-house, in the company. Since we helped build, as you said, the beating heart of Power BI. We were back there, back in the day and through some of the iterations. We had the experience to just go crack it open and go do it. I do think that anyone is building a Vertical SaaS product, should look at that as an option in their model. Wow, it makes a huge difference. You're never going to get to the point when you can hand over something like Power BI Desktop to your business analyst or whoever, whatever analytic type shows up in your, in your Vertical SaaS world, is somebody who can crack open some kind of tool like that. If you're going to go build a designer from scratch, you're crazy in this era.

Rob Collie (01:49:46): My last question, what's next for Hubstream? What's the next big frontier, the white whale that you're chasing? What's the stretch goal?

John Hancock (01:49:53): Stretch goal for us is that we have been doing this long enough to know what is going to unlock the next level for us. We know that we're doing fantastically well for big, national and global organizations. If you look at our customers today, it's all the big names that you would recognize. If you go do a search on government procurement sites, you'll see who we're working with. It's all the big ones that you would imagine. Particularly the, what they call the Five Eyes countries, New Zealand and Australia, Canada, the US and the UK. That's where we are, right? Doing a fantastic job on those big agencies. But like I said, the world is full of so smaller organizations, right? They're not going to go and stand up and do an implementation of this big line-of-business system, if you're a local police department or a small brand that's just coming up, or a company that's starting to see fraud in their transactions.

John Hancock (01:50:45): None of these people are going to go and say, "Right, we need to reorganize our whole entire universe around this hub". We are working and have been doing for a number of years now on the next generation of our technology, which we'll be launching later this year. You can invite me back to talk about it at that point. It's...

Rob Collie (01:51:00): I would love to.

John Hancock (01:51:01): Yeah, this is our V4. In the history of software, V4 is when you really start to get your game on, right? You do V1. It's limited. You do V2, which is disaster. Always. All V2's are disasters. V3, you stabilize all the mistakes you made in V2, and you realize what you're actually going to go do when you grow up and V4 is our grown up version. We're hoping to take all of those lessons, all of the scale, all of the way that people work and make it available to a much larger audience. Anybody who is touching criminal investigations or investigations at any scale, weighed all the way down to a single human, we have something fantastic and mind blowing that's coming. That the team has been working their butts off. We triple the size of the team in the last two years to go build that. That's the white whale we're chasing as you correctly discerned. We are after something here and that's it. So stay tuned.

Rob Collie (01:51:51): Indeed. We're going to simultaneously release the comic book...Dr. Data.

John Hancock (01:51:56): Great. Fantastic.

Rob Collie (01:51:59): M. Knight Shyamalan is already on board for the movie.

John Hancock (01:52:01): That's great. Like I said, John Oliver can play me. So that's good.

Rob Collie (01:52:04): Yeah. Totally. John, this has been a blast. Always loved the podcast excuse to reconnect people that I knew 10 years ago. It's so great.

John Hancock (01:52:19): Yeah. It's been tons of fun.

Announcer (01:52:20): Thanks for listening to the Raw Data by P3 podcast. Find out what the experts at P3 can do for your business. Go to powerpivotpro.com. Interested in becoming a guest on the show? Email Luke P L-U-K-E-P at powerpivotpro.com. Have a data day.

Subscribe to the podcast
  • This field is for validation purposes and should be left unchanged.

Other Episodes

Copy link
Powered by Social Snap