Data from the Outside In, w/ MVP Kristyna Ferris

Rob Collie (00:00:00): Hello, friends. Today, we welcome P3 solution architect and Microsoft MVP Kristyna Ferris. Having joined us recently, she's new to P3 but she's been in data and consulting for a long time. She's a talented and deeply thoughtful person, contributes a lot to the community and that P3 continues to attract such a high caliber of talent to our team is super, super gratifying. Never gets old. When thinking of a title for this episode, the whole outside-in theme jumped out at me almost immediately. We've long advocated for something like this as a company with our themes like faucets first instead of plumbing first and working backward from business impact rather than forward from the tech. But Kristyna chose this outside-in approach as her path in life and truly came to the world of data from the outside in.

(00:00:51): Now, her thoughtful nature about her job 100% reflects that valuable perspective. But when I say that, I don't mean to imply that her life path made her thoughtful. Quite the contrary. I think her life path is the result of being thoughtful and then her experiences along the way merely amplified that core nature. In a way, she was so determined to come to the world of data from the outside in that she grew up on the inside. She grew up in the data community attending SQL Saturdays with her father. But as we've long documented, the data gene can lie dormant for a while, and that's exactly what happened with her because even though she was going to SQL Saturdays, she wasn't really into it at the time.

(00:01:32): When it was time to go to college, she decided to study anthropology instead. And what is anthropology? It's the study of humanity, human behavior, cultures. And I think anthropology is probably the science that's most closely relevant to business. Psychology tends to study humanity from the bottom up. It starts from the individual, which means it kind of like Peters out and gives up before it gets to the organizational scale of business where data is relevant.

(00:02:00): By contrast, though, anthropology studies cultures from the top down. Now take the word culture and swap it out for organization and you actually lose nothing. It's still a 100% accurate description of anthropology, but now its connotations include businesses and other organizations and not just say villages. And if that isn't enough to convince you of the relevance of anthropology to the things we do in the data world, well, this is the second time in the history of this podcast where we've had an anthropologist turn data Professional on the show.

(00:02:35): That's right. Years ago we had Geoff McNeely on in the episode title, An Anthropologist Goes Corporate. Circling back to Kristyna, she even then later subsequently started her data career in Power BI from the outside in. She started out by designing reports without knowing the internals of Power BI. Just viewing them as almost like web pages. How cool is that starting strictly from the human factors and where it meets the audience, where it meets the people who need to use it and not starting from the tech.

(00:03:07): Despite that beginning though, Kristyna now swims in the deepest ends of the technical pools. So what a journey it's been. We of course talked about that journey and her perspectives on it as well as themes like learning only through direct benefit. Something I actually very firmly believe in. A surprising revelation to me that one of the many reasons to adopt Power BI is that it's easier to hire for than other technologies. Never heard of that one before, but it totally makes sense. We talked about visualization as a priesthood and how there's a related lesson we can learn from a conspiracy theory surrounding Quentin Tarantino.

(00:03:43): We also discussed a number of perspectives on business value, the concept of "failing fast", the idea of exploring multiple paths in parallel. This episode is also just chock-full of stories ranging from a tale of a tableau to Power BI conversion on one end and why you might not want to give whiskey to a horse on the other. How's that for a lead-in? So now we bring you Kristyna Ferris.

Speaker 2 (00:04:11): Ladies and gentlemen, may I have your attention please?

Speaker 5 (00:04:15): This is the Raw Data by P3 Adaptive Podcast with your host, Rob Collie and your cohost, Justin Mannhardt. Find out what the experts at P3 Adaptive can do for your business, just go to p3adaptive.com. Raw Data by P3 Adaptive. Down to Earth conversations about data, tech and biz impact.

Rob Collie (00:04:45): Welcome to the show, Kristyna Ferris. How are you today?

Kristyna Ferris (00:04:49): Doing well. I'm happy the weather has shifted in Kentucky. We are getting some rain finally feeling good.

Justin Mannhardt (00:04:56): It's a shift from like scorching hot to rain. Is that what's happening?

Kristyna Ferris (00:05:00): Yes. I will take that any day of the week.

Rob Collie (00:05:03): Our house is hyper well insulated, and so as a result it felt okay putting in like a low powered air conditioner. It gets really, really, really hot. It won't keep up. So I've got to give our air conditioning a running start early in the day so that when we go to bed, it isn't like 80 in our bedroom. We also had an attic fan installed. Highly recommend. It's like free three degrees off of the upper floor of your house. What the results in though is days like today when it's not hot outside, it's freezing in our house at noon because the head start is not needed.

Kristyna Ferris (00:05:41): I work from the basement. It's always freezing down here and my husband had to tell me, he's like, "I know I don't see the electrical bill as much as you do, but sweetheart, it cannot be reasonable to have a space here on at the same time as your air conditioning unit." I was like, "You're probably right."

Rob Collie (00:05:58): The old Steven Wright joke. One time I put a dehumidifier and a humidifier in a room and just let them fight it out.

Kristyna Ferris (00:06:06): Do you have the hoses feeding each other because I feel like that would be...

Rob Collie (00:06:10): Yeah, that's right. That would be the perpetual motion version of this. You are today a solutions architect here at P3.

Kristyna Ferris (00:06:20): Yes, I am. Very excited to be here.

Rob Collie (00:06:23): We're incredibly excited to have you here. When did you start here?

Kristyna Ferris (00:06:26): First week of May.

Rob Collie (00:06:27): I was looking back through your blogs. On May 14th you wrote a blog about connecting Power BI to online Excel, which is way harder than it should be.

Kristyna Ferris (00:06:36): Correct.

Rob Collie (00:06:37): In a recent podcast that I recorded after that I mentioned like, "Oh my God, I can't believe it's still this hard." When I went back and saw your blog, I'm like, "Oh, did she hear that in the episode and go, 'Nope'"? Your blog about connecting Power BI to online Excel predates me whining about it.

Kristyna Ferris (00:06:53): It's amazing. I heard that and I thought, "Oh my gosh, did I write that the same week that that came out?" No.

Rob Collie (00:06:59): Nope?

Kristyna Ferris (00:06:59): I had the exact same thought.

Justin Mannhardt (00:07:03): I think everybody has that thought every time they connect to Excel online.

Kristyna Ferris (00:07:08): Yes.

Rob Collie (00:07:08): It's like if only these two products were made by the same company, if only they were both data products like Outlook and Power BI, we wouldn't necessarily expect it to be so dialed in. But no, the two most popular Microsoft data applications requires blogs written where you say things like, "Oh, and make sure to highlight the question mark web equals one or whatever from the end of the URL and delete it."

Kristyna Ferris (00:07:38): You know what's amazing is that quickly became my most viewed blog post.

Rob Collie (00:07:42): I bet. I mean, like what the hell? I've literally saved a Power BI PBIX file with the name, this is how you connect to Excel Online.

Kristyna Ferris (00:07:53): I have a little book of measures in my blog and it's just little things I've picked up throughout consulting or throughout various jobs. And whenever I want to go and write a blog post, I take one of those and then go write the blog post. So the code is already written. I just have to add the fluff and the screenshots.

Rob Collie (00:08:09): Which takes like five minutes. Right?

Kristyna Ferris (00:08:11): It's so fast.

Rob Collie (00:08:12): Instantaneous.

Kristyna Ferris (00:08:14): It's really just laziness that I haven't put these out more often. No.

Rob Collie (00:08:20): It's like those memes that say like, "Oh yeah, no problem. Five minutes in and out."

Kristyna Ferris (00:08:25): In and out of a time warp maybe

Rob Collie (00:08:27): That's how I used to feel about every time I sat down to write a blog. "This one won't take long." Days later.

Justin Mannhardt (00:08:32): A few hours later. This won't take long.

Rob Collie (00:08:38): So new to P3, but not new to data. Not even close. Can we get your data origin story? What was your first collision with this stuff and what about data spoke to you when you made that collision, you stuck?

Kristyna Ferris (00:08:51): It's kind of a weird story. So my dad has been in data forever since I was born, and so growing up would go to these SQL Saturdays and would speak at different conferences. He's helped write a couple of books. Now he has a YouTube channel about Fabric. So I grew up around him and around data culture. He was more on the SQL side of things, got really into MDX and so that's kind of his history and his path. I going to college was like, "Well, I don't want to do any of that."

(00:09:18): Instead, I went for anthropology and ended up learning a lot about human culture and about archeology and really loved that, but realized that I couldn't see a job at the end of that tunnel necessarily, so I quit the grad program I was in because I'm not paying more money to learn the same thing I just learned. So I ended up going and finding a job. A few jobs, but ended up sticking in an admin role where I was also helping their marketing team with data entry.

(00:09:47): Eventually that turned into helping them manage their logistics. From there, I realized I really liked logistics because it's a lot of numbers, so I was doing all of their logistics through Excel and I ended up getting hired by a third party logistics company, went over there and started learning Power BI on my lunch breaks because I had a connection from a place my dad worked and he was like, "Hey, I need somebody to make these reports look nice. I think you've got a good eye for it. Could you go ahead and just make it make sense? Make the numbers look better than the DBA made it."

(00:10:18): And I was like, "Yeah, I think I can do that." So I started learning Power BI. My free time, did that as a side gig for him. And then I thought, "Well, what else can I do with Power BI?" I've got all these Excel sheets of data coming in from my different brokers, from different truckers that I'm working with. I might as well throw this into Power BI. So I did that and I presented it to my boss at the time and she was like, "Well, this is great actually." I had put together this presentation about why I deserved a raise, and that was my motivation.

(00:10:46): And she was like, "Well, how about we get you in touch with the analytics team instead?" So ended up joining their team and we converted them from Qlik and Tableau over to Power BI and then the rest is history.

Rob Collie (00:10:59): Why would you do such a thing? Why would you take them out of their Qlik and Tableau comfort zone? It's obviously a joke, but you made it sound so easy. We converted them from Qlik and Tableau to Power BI as if that's what they were waiting for like, "Oh, thank God. Kristyna is here to change the tools that we use." That's not usually how it works, right?

Kristyna Ferris (00:11:19): No, and I give that manager a lot of credit. Him and I are actually still good friends. Basically, when I came to him with this new tool that he had never seen before, he was like, "You know what? You were able to spin this up on your own with no IT resources and you were able to get significant business value without any help from our team. We are struggling to do that in these other tools. We're struggling to get to a place where people can get good insights and Tableau licenses are expensive. Qlik licenses are expensive. Power BI wasn't. We can spin up this the fly and we can see how it runs and operates and it proved itself relatively quickly.

Rob Collie (00:11:59): Yeah. That's something else.

Justin Mannhardt (00:12:00): Here's a value proposition just to push the bold button, buy yourself also known as incredibly efficient, cheap, not as expensive and not a struggle.

Kristyna Ferris (00:12:15): A hundred percent. And it was easy to teach people, so it wasn't this whole big new scary thing that nobody knew how to operate and would take forever to maintain alongside build because that's always the question is you need to be able to maintain it after you've built it. Some of these tools are very difficult to do that with.

Rob Collie (00:12:34): Preaching to the choir.

Kristyna Ferris (00:12:34): Right.

Rob Collie (00:12:37): We get it. Just tremendous credit to the other people involved in that story because it's just an unknown at that point. They don't know that Power BI is more flexible. They don't know that Power BI is more versatile and faster and all those sorts of things. They can see the price tag.

Kristyna Ferris (00:12:52): Yeah, which helps.

Rob Collie (00:12:54): Absolutely helps. Even though it's a better tool. The price has been, I think, the number one reason for Power BI's explosion in adoption. But there's still people there. Their professional identity and the identity of the software that they use is blurred together. It doesn't sound like they held on with the white knuckle grip. They were open-minded, but that is the exception.

Kristyna Ferris (00:13:18): Yeah. And it wasn't everybody. It was the manager. He, I think, was more open-minded than anybody else, and he was awesome. He saw the potential and I think he saw the potential because he was trying to expand his team and was having trouble finding anybody with the right skill set or the right trainability-

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

Kristyna Ferris (00:13:36): ... to learn the things that his team had already built. So they had already built all these Qlik reports and these Tableau reports. They had also done a lot of things in Python and he was just having trouble finding anybody that he could reasonably put into essentially a entry-level BI role that knew those things. That just doesn't really exist.

Justin Mannhardt (00:13:58): It's hard. It's so hard.

Rob Collie (00:14:00): Kristyna and Justin, you're both going, "Yeah, that totally makes sense." I'm missing it. I'm missing why. I do know that Power BI is easier to learn than Python, but at the same time, Python is being taught everywhere. Python is more popular.

Kristyna Ferris (00:14:17): So it's easy to think about it this way. Python is a sea of things that you can do with it. It's basically an ocean.

Rob Collie (00:14:24): Yes.

Kristyna Ferris (00:14:25): And so when you are trying to learn Python in a class or in a school, they might show you how to navigate the Pacific Ocean near Japan and those waters and the reefs and how that works, but they're not necessarily showing you how to navigate Australia. It's the same language, it's the same composition, it's still the same ocean, but it's vastly different. If I were going to deep sea dive in those two places, there's going to be some similarities, but there's going to be some differences.

Justin Mannhardt (00:14:57): Well, this is kind of why I went, "Oh yeah, this is hard." This is meme-worthy stuff. You see on LinkedIn all the time where people are posting jobs for entry level data analysts and then they joke like 15 years of experience with Python, 12 years of experience with Tableaus. SQL expert. While you're correct, Rob, that in aggregate Python is probably more broadly known as a tool or a technique compared to even Power BI with its market dominance. That doesn't mean that those audiences are solving the same level of analytical challenge.

(00:15:30): You can do far more interesting and powerful things with Power BI knowing some relatively rudimentary DAX and drag and dropy chart-making compared to that same entry level in Python of what you'd be actually able to do business value wise. So I think that's why I was like, "Oh yeah, this is hard," because you actually need a pretty reasonably high level of skill to get to the same business value point.

Rob Collie (00:15:53): Yeah, that makes sense. You can use Python to write games if you wanted.

Kristyna Ferris (00:15:56): Yeah.

Rob Collie (00:15:57): So that speaks to its versatility and the many different flavors of oceans and all that kind of stuff, but any tool that is that versatile, it's also not specialized. I mean, it's true, right? I've even gone to developer conferences just to show them how unfair it is what I can do with data. They can't even though they're far better programmers than I. I just go in as like the unfrozen caveman lawyer. I'm just a simple man that knows how to draw lines between tables and write formulas. Look what I can do? And look at the query response time of this. You all can't do that, can you? It's confusing.

Kristyna Ferris (00:16:34): That must be hard.

Rob Collie (00:16:38): So I think that's the first time I've heard. It makes total sense to me now, but it's the first time I've heard of someone choosing Power BI because of the comparative ease at recruiting the right kind of talent to work with it. That's a new puzzle piece that just got installed in my brain and I really appreciate it. I like puzzle pieces. So the manager was in. Some other people might've been a little disgruntled.

Kristyna Ferris (00:17:01): There was one in particular like a lead engineer on the team who was a little annoyed with it mostly because he was a heavy Tableau user and so for him he couldn't do the visuals that he could do in Tableau.

Rob Collie (00:17:13): Right, yes.

Kristyna Ferris (00:17:14): The problem was that the business didn't care. They cared about the insights and how fast those could be adjusted to get answers to the questions they now have this week that are different from last week and that speed wasn't there.

Rob Collie (00:17:27): Visualization as a priesthood, is so funny. I love you said the only problem was the business didn't care.

Justin Mannhardt (00:17:34): Yes. There's a market and a need for truly avant-garde creative data visualizations that are just amazingly out there and it's like the one big thought piece you see in the big issue of The Economist and they're doing a study on something and you need this very bespoke thing. I don't want to rag on people too much, but you see the Tableau community talking about this all the time sometimes and I'm like, "Yeah, but that's not the everyday business need. Column charts."

Rob Collie (00:18:07): This is going to be a weird parallel, but that's one of the things I do. I read an article many years ago. The author of this article didn't believe this theory. His theory was that Quentin Tarantino wasn't real.

Kristyna Ferris (00:18:18): What?

Rob Collie (00:18:19): Quentin Tarantino had been invented by Hollywood.

Justin Mannhardt (00:18:23): Go on.

Rob Collie (00:18:25): As a means of telling all of the aspiring screenwriters and directors out there that they can do it. He'd worked at a video store. He'd worked at a video rental store. He was just a movie buff who turned out to be one of the great popular directors and screenwriters. The author of this article, his theory was is that Quentin Tarantino was a myth that was invented to keep the pyramid scheme going.

Justin Mannhardt (00:18:52): Wow, that's next level.

Rob Collie (00:18:53): All these people out there producing scripts and directing low-budget movies and everything so that Hollywood could come along and harvest the 0.1% of those that were any good. But they need all of them to believe.

Kristyna Ferris (00:19:06): That they can be Quentin Tarantino.

Rob Collie (00:19:08): That they can be Quentin Tarantino so that they keep doing it. When you were telling me this story, giving you that example, Justin, of the one data visualization a year that's truly original and actually does something for the world. That's Quentin Tarantino, right? The rest of them are all sitting around hoping that that gets to be them and they're trying to produce the one visual to rule them all, and it turns out we just need charts. And the numbers that power those charts are really hard to come up with and whether the chart is rounded or square, get it to our eyeballs.

(00:19:37): There are different visualizations that a hundred percent work better than others and sometimes it's not subtle. It's not to dismiss the whole thing, but you know that everyone is playing the Quentin Tarantino game when they define themselves as a visualization expert. Right? The demand for exactly that is not what you think it is.

Kristyna Ferris (00:19:54): No. And honestly, we try to... As report developers, I personally, when I look at a report, I view it like a webpage. So I'm trying to mimic an experience I would want somebody to have with a UI. And to me the difference between that and the visualization folks that we were just discussing, which is not all of them, but the ones who are chasing that perfect visual is that the UI/UX are always thinking about the end user. Everything they do has a purpose. Everything they do is to save somebody from moving their mouse an extra inch or saving them from having to do an extra click. Everything is in that beautiful form meets function reality that I don't think always happens.

Rob Collie (00:20:38): And also saving people's brains from having to make leaps probably the single biggest failing that happens with software in general by default. "Look what the numbers can tell you." It's not really what we should be thinking about. We should be thinking about your problems and your workflow and opportunities and working backwards from that. But what was ultimately the outcome for that one particular specialist? Did they make the leap to Power BI and ultimately happy or did they change jobs?

Kristyna Ferris (00:21:07): For a while, he changed roles. So part of the reason why they were trying to even hire more junior folks was so that he could be freed up to do more data engineering and even potentially some machine learning because that's where his interest lied and that's what the vision was for him was to be able to do that. But he ended up moving companies eventually within the year. So it wasn't a hard cutoff and it wasn't too painful for him. I think it was more painful when we were doing the translation between the reports. Just to see certain things die that he had worked so hard for, I think was hard.

Rob Collie (00:21:43): And to not be able to replicate them, exactly. Why are we adopting this new thing when it can't even do X?

Kristyna Ferris (00:21:49): Exactly.

Rob Collie (00:21:51): I can absolutely sympathize with that pain. It's only in the broader context that I go, "All right, no biggie." But on the human level, completely see it.

Justin Mannhardt (00:22:00): There's an interesting insight on the human level of that, and I think this is what you were saying when you empathize with this, Rob, and it's sort of even maybe a word of caution to us all now in terms of what's was going on in the world today. When you invest your time and your energy into learning a tool, any tool, Power BI, Tableau, Qlik, any of these things, and you spend a reasonable percentage of your life developing solutions in those things, you come to the reality of like, "Oh, my thing isn't the thing anymore." That's hard.

(00:22:31): There's this thing going on with AI and generative AI and not the old, "What's happening with Fabric and all these things?" that I think a lot of us, whatever the thing is, the sooner you realize it's the thing and are willing to adapt and change and be open-minded to the way I do my work is going to be a little different, that's just sort of a human level observation I had based on the story Kristyna was telling here. But who knows what's going to happen in the future. I think, Rob, you even said, "Listen, if something comes along that's like light years better than Power BI, we'll all switch. Until then..."

Rob Collie (00:22:59): We remain open-minded.

Justin Mannhardt (00:23:00): Yeah.

Kristyna Ferris (00:23:01): I think we're seeing that change a lot with SQL developers right now. SQL DBAs for a while now have been looking at the cloud like, "Oh, shoot. Everything I've fine-tuned, everything that I've honed in is now about to be out of my control." So I think that we're seeing that shift already. I've been looking at those ones who are excited about it, those ones who are terrified of it and the ones who are like, "I'm not changing. I'm close enough to retirement. I'm sticking it out with the technology I know." When I retire, it can go to somebody else who's willing to go through the process.

Justin Mannhardt (00:23:34): Kristyna, somebody told me once, this has been a few years, this is Justin, "There's really only two ways to make money in tech. Either be on the bleeding edge or be on the stuff that's stone age."

Kristyna Ferris (00:23:45): Oh, it's so true though. It's so true.

Justin Mannhardt (00:23:48): I have a connection who independent contractor essentially and all he does is consult on SQL 2000 or something.

Kristyna Ferris (00:23:54): Oh my word.

Justin Mannhardt (00:23:55): Right? Because there's just still a bunch of companies that are running applications that run on SQL 2000 and nobody is going to go invest in learning that for the first time. It's just not going to happen.

Rob Collie (00:24:05): Talk About niche marketing. I love it. Very specific, SQL expert in this one old version.

Kristyna Ferris (00:24:12): No longer Microsoft supported, but I'll support you.

Justin Mannhardt (00:24:15): Dominating my category over here.

Rob Collie (00:24:17): And keep in mind back then when we named products after years, we didn't name it that year. During that year, we named it a year in the future. SQL 2000 was not released in 2000. It was released before 2000 to sound cool and futuristic. It is older than 2000 SQL. SQL 2000 believably could have been the version of SQL that was introduced to add extra digits to the date to avoid the Y2K problem, right? I mean, it wasn't. But I mean it was released before the rollover of the century.

Kristyna Ferris (00:24:53): Wow.

Justin Mannhardt (00:24:55): And that version, the bit size of a value could only be so big. The application, you can only hold up to a hundred thousand numbers. Every so often, they have to run this big stored procedure to archive the entire data set and reset everything, and basically everybody gets around, lights a candle and praise that this is still going to work.

Kristyna Ferris (00:25:14): It's so funny. Being an anthropology student, I did read ethnographies that were all about different corporate cultures and one of them was focused around an IT team who had set up various toys on top of servers and depending on which servers were not doing well that day, they would move the toys around to try and help them out. It's just such a level of superstition.

Rob Collie (00:25:37): Wow.

Kristyna Ferris (00:25:39): Or baseball fans wearing their favorite whatever.

Rob Collie (00:25:42): Oh, you got to wear the rally cap backwards or why hockey players beards get longer, and longer, and longer during the playoffs. I mean that helps.

Justin Mannhardt (00:25:53): It's cold on the ice.

Rob Collie (00:25:53): Yeah, I mean it's not cold on the ice during the regular season. During the playoffs, it really gets chilly out there. I really like the crossover from anthropology to data. And it's funny that you're one of these crossover stories even though you grew up in a SQL Saturday attending family.

Kristyna Ferris (00:26:10): I know. Weird, right?

Rob Collie (00:26:13): I haven't encountered a whole lot of that in my travels. "I grew up going to SQL Saturdays." "Really?"

Kristyna Ferris (00:26:20): I didn't know what they were talking about, but I went.

Rob Collie (00:26:23): So naturally you landed in data except by way of anthropology. I also really like how your first experience working with Power BI was just working at the user layer, making these things more understandable. What an interesting pathway. No one learns that way. Few people do anyway. You start with the Power Query and the DAX, and the relationships and all that. And then the report is an afterthought. But to work backwards, to learn backwards like that.

Kristyna Ferris (00:26:53): It's funny because you guys talk about faucets first all the time.

Rob Collie (00:26:56): I almost want us to teach our Power BI classes in that direction. I don't know that it's an efficient curriculum for a two-day class, but I think it's the right thing.

Kristyna Ferris (00:27:06): I've realized that the fastest people learn is if you give them something they're interested in, instead of having those Contoso data sets and stuff, we used to have people bring their own data set and be like, "All right, let's do what we can do in an hour."

Rob Collie (00:27:19): A long time ago when I used to go and teach, I would often say, "Let's use your data."

Kristyna Ferris (00:27:25): Yeah.

Rob Collie (00:27:26): The problem of course was the outlier examples where the data was ugly in a way that there wasn't a lot of learning value in cleaning it up.

Kristyna Ferris (00:27:32): Yeah.

Rob Collie (00:27:33): It's like 80% of the time it works 100% of the time. Eventually stopped doing that walking in blind. I mean, the jumpstart consulting methodology that we use works well that way because you're working towards a specific result. If you're the client and you're interested, you can gain some education on how it works in the process. But using someone's arbitrary data set to teach them fundamentals right off the bat, you find that we're going to need to use the treat as DAX function on your first day to solve this problem. It's just too bad. Your business is structured that way and this class is going to be a write-off.

Kristyna Ferris (00:28:07): Oh, and by the way, you don't have any true dimensions, so we're going to try and create them. Here you go. We're going to dissect this apart. We're going to do distinct rows. You're like, "Oh my gosh, they did not need to know all this right away." But they did, and that's the hard part is they do need to know all that right away, but one thing at a time.

Rob Collie (00:28:27): The power of dimension tables. You're not the first anthropologist to be on the show.

Kristyna Ferris (00:28:34): Really? No way.

Rob Collie (00:28:36): Yeah. He's like an anthropologist goes corporate.

Kristyna Ferris (00:28:38): That's amazing.

Rob Collie (00:28:40): Was the name of one of our really, really old episodes, but the study of people of cultures seems pretty relevant to me.

Kristyna Ferris (00:28:47): That's how I spent it in my job interview for that analytics switch. So he's like, "What does that have to do with it?" He's like, "You've never even taken a stats class." And I was like, "Well, neither have any of our business users. Everybody who's going to be looking at this is going to want to know, how does this answer my question?" Anthropology is all about taking quantitative questions and making them qualitative.

Justin Mannhardt (00:29:09): Interesting. To your point, Kristyna, being asked in an interview, something as seemingly obvious as you've never taken a stats class. I'm sitting here thinking, "I mean, I've worked on hundreds of different things. When did I ever need to apply statistical rigor to the analytics I was doing?" But in the vast majority of business use cases, it is a people thing. Someone is trying to get some type of result in their role or function within a business and I'm sorry, but the five number summary usually doesn't apply.

Kristyna Ferris (00:29:42): No. It's like I got these five questions that my boss asks me all the time. I need answers to those five questions and I need to be able to change how I ask that question slightly every time. That's nine times out of 10 what a report ends up being.

Rob Collie (00:29:57): Concepts from statistics. You can't just apply them and get business value out of them again unless people actually understand what it is. So the standard deviation is a perfect example of this.

Kristyna Ferris (00:30:10): Yes.

Rob Collie (00:30:11): There is an incredibly humane and intuitive way to explain the concept of standard deviation in certain circumstances, but no one ever bothers. What is normal noise in a data set? What is considered normal noise versus like, "Whoa, something might actually be changing here." If you get that explanation out there, then you say, "Ah, all right, now we can use the stdef.e or-

Justin Mannhardt (00:30:38): I just steered clear of that whole drop-down in Excel formula as I was like, "Nope. I'm out."

Kristyna Ferris (00:30:44): Which is so funny because I think it is so valuable if you know what it's doing. We went to people and we asked them, "Do you know what this report answers for you?" And they said, "No." I said, "All right. We're throwing it in the garbage and we're going from scratch."

Rob Collie (00:30:58): Next.

Kristyna Ferris (00:30:59): Yeah.

Rob Collie (00:31:00): I wonder how many reports in the world today survive that test? There's also the test of does anyone even know why we have this report? Who made it? When was it last looked at? But there's one that we're supposedly all looking at every day, but we don't know what it does for us also real. Do you have any anonymized, real-world examples of things like this, a previous mindset around something and then when you applied this more human-focused thinking to it, you end up in a completely different place?

Kristyna Ferris (00:31:29): Yeah. The biggest one I think was that example. They did have linear regression in some of the reports, and so we asked the account managers if they knew what it meant for their accounts and they said no. They were not sure what it meant, like customer churn.

Rob Collie (00:31:44): Oh, was there an R-value column in this report?

Kristyna Ferris (00:31:48): It was something to that effect, and it would sort customers that way. The problem was they couldn't figure out how to change it.

Justin Mannhardt (00:31:54): Or what to do about it.

Kristyna Ferris (00:31:56): Yeah. So it was just so impractical because they were like, "Yeah, I know this person, this customer, or in this case it was logistics." It was like, "I know that X, Y and Z company is likely not going to broker their loads with us anymore. I've talked to them and they've told me what can I do to change it, and that was never in the reports.

Justin Mannhardt (00:32:16): The regression thing, we used to work at a company where we printed things and you would need to kit all these things together and ship them to stores. And so somebody up in the office did this regression. They were trying to prove how many people you needed to staff in the assembly line to make the kits, but the math would always come up with like, "Oh, you need 17 people," and then you'd go down the floor like, "You can't physically situate 17 people in the area. We have to do this." It is just devoid of reality of what to do about this problem, and I've seen so many numerical conclusions devoid of realism of action.

Kristyna Ferris (00:32:53): That was honestly the biggest reason why I didn't go into a mathematical major. I love math, but I could never see it directly applying to the real world in a meaningful way. Maybe I'm just not creative enough. There are a lot of people out there who've done amazing things with mathematics to impact the real world.

Rob Collie (00:33:11): Have I told you this story of Quentin Tarantino?

Kristyna Ferris (00:33:20): Yeah, right, exactly. That's exactly it. It's the same feeling.

Rob Collie (00:33:21): Most math professors are mostly just making other math professors and satisfying curriculum requirements. Dammit, Vanderbilt, you forced me through another three semesters of calculus.

Kristyna Ferris (00:33:39): Oh, man.

Rob Collie (00:33:40): That as you say, was never again relevant. I was captain of the calculus team in high school and I found those three semesters of calculus in college to be traumatic.

Kristyna Ferris (00:33:52): You know what's so funny, I was captain of the math league at my high school and I didn't take a single math class in college. I had done it all in high school. They just counted my credits.

Rob Collie (00:34:04): I thought that's what I was going to get. I thought I'd placed out of calculus. "No, you just placed out of the first semester and we're going to drop you in the second semester. It's going to be hard from the jump." It just kept getting worse from there. I retain nothing from differential equations. I can't even tell you what it's used for. I don't remember. No idea. Apparently I got through it. That's it.

Kristyna Ferris (00:34:28): And that's how I'm here today. That's my survival story.

Justin Mannhardt (00:34:32): I survived long enough to become an analyst.

Rob Collie (00:34:37): So coming from anthropology and working with Excel, but then your first experience with Power BI being at the report layer, since then, how technical would you say you've become? Everyone has got their own separate journey there where their line is of where the passion sort of stays lit. How far have you found yourself working backwards down the tech stack?

Kristyna Ferris (00:35:00): Deeper than I ever thought. I'll say that. So when I got into Power BI, I actually didn't know any SQL, and so the DBAs basically changed my password to the SQL database until I took some courses and could pass their test and could show them. I learned SQL. From that, now I feel like I'm never in the front end of Power BI. All I'm doing is SQL or data integration. I've even done a lot of work with the tabular object model using C sharp assemblies. There's a lot that you can dig into, and so I found myself nerding out about the REST APIs and with the C sharp scripting that you can do because if you can just make your job more efficient, why wouldn't you? I'd say I've gotten way deeper than I ever thought.

Justin Mannhardt (00:35:43): Sounds like it.

Kristyna Ferris (00:35:44): Yeah.

Justin Mannhardt (00:35:46): Talking about Tom over here.

Kristyna Ferris (00:35:48): Oh yeah. I love Tom. We're BFFs. And Timsel. We're becoming good friends.

Justin Mannhardt (00:35:55): That's the evolution of BIML.

Kristyna Ferris (00:35:57): Yeah. So the guy who made it, Mateus is amazing. If you hear him talk about it, he's just got such a passion for it. His idea was to make it more reader friendly, and so not only do you now have this way of viewing the backend of Power BI, that tabular object model, it's now in a format where you're like, "Oh, feasibly, I could come in here and change something feasibly. Feasibly, I can see what has changed if I've got it in a Git repo." It's no longer this crazy mess where you're like, "I don't even know what's different. I don't even know what I would edit if I was going to change something. So there's some cool stuff.

Justin Mannhardt (00:36:31): I have a question for you, Kristyna. You wandered yourself down the technical rabbit hole here, didn't you? You're talking about C sharp assemblies for Tom. We love Tom. I can relate to some extent, probably not even as deep as what you're describing. Sometimes when I talk to people they wonder like, "Justin, do I really need to go learn about all that other stuff?" Where do you think the line is? Obviously, you're a professional. You're a consultant. You work on some of the most difficult and challenging things that people could find in this arena, but when do you help people understand like, "No, you're totally fine. You're good. Basic model, basic decks, basic stuff."

(00:37:05): When is the tipping point for people, do you think? Or where's the real need and real value for you in exploring those things because you come from those roots of the human impact? Maybe just expand on that for a second.

Kristyna Ferris (00:37:15): I'll give you a hint. I'm not an independent learner. I won't go learn something just because I want to learn it. I have to have an end goal. So even the C sharp stuff, I got onto a client. It's like my first time consulting, my first consulting client, and they were using C sharp to update their Power BI reports. What they wanted to happen with end users would say, "I want these fields included in our base data model package." And then the C-sharp assembly would grab that list and it would create a data set for them in a workspace.

(00:37:47): So I got really familiar with it because of that use case. So for me, it came down to, and it still does in a lot of ways, is are you going to use it? Because if you're going to learn it just to learn it, you are never going to learn it until you actually have a reason to learn it. You are never going to get into the complexities that pop up randomly when you're trying to actually do it in real life.

Rob Collie (00:38:08): A hundred percent agree. I've never learned anything just for the heck of it. But brand new things, even if I have a need for it, I find the startup cost of it to be really daunting just getting the environment set up. I don't think that if someone sat me down in front of a C sharp code editor that I would have too much trouble. Because I've never written any C sharp, but I wouldn't have too much trouble learning to write the C sharp. But how do I even get into this editor? How do I deploy that code? How do I test it? What's its connection to the rest of the environment? All those sorts of things.

(00:38:46): Those things make the whole experience. It might as well be on Mars to me. Those things probably aren't. No, they're definitely not the most technically daunting, but for some reason those really, really bug me. If I could just get to the "hard part" with all the noise pruned away, I'd be more inclined to dive in.

Kristyna Ferris (00:39:09): Nobody writes about the setup.

Rob Collie (00:39:12): I know.

Kristyna Ferris (00:39:12): They just kind of assume that you know how to do it.

Rob Collie (00:39:15): Yes.

Kristyna Ferris (00:39:16): That's been my number one thing with Python. I've been trying to learn it for some time, and that to me was the biggest startup cost was I don't understand. What do you mean I need a notebook? What do you mean I need to do better? What is this? How do I open this? How do I connect it to things?"

Justin Mannhardt (00:39:31): What's pip?

Kristyna Ferris (00:39:33): Oh my gosh, and why can't I do it? I have local admin rights. Why is it not installing the way I want.

Justin Mannhardt (00:39:39): Pip, Tom. Let's put Tom on a pip of cheese.

Kristyna Ferris (00:39:45): Oh, no. I mean, to your point, right? Those are the kinds of things where if you are in the situation where you need to learn it, you are usually surrounded by people who can help, and I think that makes a world of difference.

Rob Collie (00:39:58): Yes. Just please show me.

Kristyna Ferris (00:40:00): There are other people invested in your success.

Justin Mannhardt (00:40:03): Yeah. That rings so true to me, and I've learned so much just like being around other consultants. Here at P3 especially that I'm sure even where you've been in your career so far, Kristyna. I wanted to take maybe some of the FOBO or FOMO out because when we talk about these things, people go like, "Oh my God. There's a hundred things they don't know how to do." That's okay.

Kristyna Ferris (00:40:22): No, it's so true. The FOBO is real, and I think that the FOBO can distract you from that business value that we were just talking about. I think even just like we were talking about with visualizations being bright and shiny and not useful, sometimes new technologies are bright and shiny and not useful. Sometimes you don't need them. I think the danger is writing them all off. Sometimes you do need them and it can be really helpful. It's this balancing act of knowing what's out there without putting pressure on yourself to have to deeply know everything about it.

Rob Collie (00:40:55): Having a properly pruned worry tree is crucial to life. Unpruned worry tree, and you collapse. Over pruned worry tree and it dies. Right? I'm still struggling to get the balance right. Over prune, under prune. It's a constant cycle with said worry tree.

Justin Mannhardt (00:41:15): Never ending, Rob.

Kristyna Ferris (00:41:17): I mean, I just think of it like a newspaper. Most of the time I read headlines. I never cancel my newspaper subscription. I'll read the headline. If I'm curious about it, maybe I read the article and then if I'm really curious about it, maybe I go talk to a journalist. This I never do in real life, but in the analogy here would be if I actually need it, I don't have to be the journalist writing the article to know enough about it if it's not useful for me.

Rob Collie (00:41:43): That training we got with newspapers is really serving us poorly in the age of misleading clickbait headlines.

Kristyna Ferris (00:41:50): So true.

Rob Collie (00:41:53): As you scroll through any social media feed, if you just read the headlines, you're going to come away believing things that are wrong.

Kristyna Ferris (00:42:00): Yeah.

Rob Collie (00:42:01): The headlines literally lie now and you have to go read the article to find out that it's lying. They've taken it so far now. Headlines are junk food now. It's so awful. Anyway, this guy is completely separate from data. So, Kristyna, you have been in the consulting game for a number of years now in the data space. You've had some reps. You've seen a fair number of business leaders that you've worked with over the years. Do you have any examples of business leaders in the ways that they approached data projects, the ways they thought about it that were either really helpful or maybe even really unhelpful?

Kristyna Ferris (00:42:38): Even right now, I'm working with a certain client who I think has a very different perspective on this than I've ever seen before. This gentleman who is in the C-suite. And so his perspective on technology is he wants the latest and greatest, but he also wants what's going to work the fastest to get his business value. With that, I think what he's done is he has set aside a large portion of the budget to essentially try out a bunch of different things at once.

(00:43:04): For example, we're working with a couple of different technologies with them right now, not just Fabric, but we're also working with some other tools from other reporting softwares. And it's interesting to see that even though they don't play very well together right now, his goal is just to get the business value as fast as possible. So he was sold speed from one company for ingestion, speed from another company as far as to the business Power BI, that kind of value, and he was also sold Fabric on the premise of, "Yeah, it's one lump sum."

(00:43:38): You know what your fixed costs are going to be. We've got these reserved pricing and it's going to be cheaper than what you've spent previously. And to me, that's interesting because typically what I see is the opposite. Typically, what I see is kind of a low and slow approach. So you get business leaders who are excited for prospects of change. They get really excited by buzzwords and really excited by kind of the analytics curve that you hear people talk about. But when they get down to it and they start talking to their teams, they get bog down in the details and aren't able to make the quick signing decisions that are actually going to get them to those insights.

(00:44:16): So it's interesting to see some business leaders who take an approach of, "You know what? I'm not going to buy into sunk cost fallacy. I'm going to take a risk. We're going to go ahead and go down this path and we're going to go down the path until we hit a wall. When we hit the wall, we have other paths we can try. Nothing is going to be sunk cost because we will have tried it and we will have known if this works or doesn't work for our company and our use case." But this whole thing is going fast. You failing fast. The faster you can fail, the faster you can succeed.

(00:44:50): And you just have to not be afraid of failing over, and over, and over again. And I think if you're a business leader and you're trying to figure out where do these different technologies play a role, start using them. Start failing. But understand that you're going to fail at the beginning.

Rob Collie (00:45:07): Yeah. I think we with Power BI, power platform and more TBD Fabric, but certainly we take something like Power BI. We're used to knowing that it's the best of breed. We're on Power BI because we know that. But if you don't know that, for example, and you're just looking at a menu like the market's offerings of all these tools and there's so many of them, and they are still inventing and funding and building brand-new BI dashboarding tools to this day, I don't understand why those are still being funded. Can't imagine trying to break in against Power BI today as a startup. Whatever. But if you don't know, it's just a bunch of noise,

Justin Mannhardt (00:45:48): Right.

Kristyna Ferris (00:45:48): Yeah.

Rob Collie (00:45:49): What happens is that you end up being talked into one of the N options on the market and you feel trapped there before you make a similar. You decide, "No, this isn't working. You didn't get the right one it turns out." Some number of years later you make a similar high-risk transition again, with the same level of confidence essentially. Oh my gosh, it'd be so much smarter to deliberately plan for a shallow investment in a lot of these. With the idea that you're not trapped, that changes the whole mentality, and you don't end up stuck with many, many years of sunk cost.

Kristyna Ferris (00:46:27): You save so much money. And not only that, but you're now building up developers on your team who have full faith in the system you're about to go forward with. You're no longer ending up in this situation where teams are told that they're switching to this new tool and they have to make it work because we've put in a five-year contract. You have to make it work, and you have to try and do this. That's a really good recipe for your team to quit. Instead, involve them in the decision process, involve them in the discovery of the tool, and they're going to be much more likely to stick around, and then you're not losing your business history that way.

Justin Mannhardt (00:47:03): It's an important lesson not only for the technology choices leaders need to make. My experience when we go into a project, even myself as a leader, craving this sort of certainty, we've got to get this right. This has to succeed. I'm certain I want to solve this problem or I want to bring this capability to market, or I want to provide this opportunity for my team, but if you go in with this no-fail mindset, you're going to fail at some ratio.

(00:47:31): And so it's the teams I've worked with, customers I've worked with that are most open to learning and pivoting during these projects, they ultimately are the ones that are the most successful. And so yeah, it's true. The technology allows us to pivot more efficiently than we could have in the past, Kristyna, but I think that's a brilliant observation that as a leader, creating the environment where curiosity and learning are ultimately held at a higher premium than just success at all costs.

Rob Collie (00:48:00): So there's something about confidence and information that's really important. It's underlying that point. I did this recent solo podcast about using these tools to help my wife, Jocelyn manage her health situations, so we'll try something. And the old mentality used to be like, what if we try this and we spend the time on it and the money on it and it doesn't work? So there used to be a paralyzing fear about that.

(00:48:26): Now though, it's like, "Oh, then we know that that doesn't work." We've gained information that we write down and we know that, A, that thing doesn't help, and B, it also helps inform us down the road as well. When we gain more information on our questions and become more precise, we go, "Oh, you know what? It turns out we've already tried treatment, which would've worked had this been true." And so this idea of being able to try things out, as long as you're trying them out with the mentality that there is the ability to back out, there is the ability to change course."

(00:48:58): Trying things out, but sticking with them to the bitter end, that's the dangerous path. Now, I still get a bit of the heebie-jeebies about this whole thing though when we're talking about whole software stacks. Robotic process automation.

Kristyna Ferris (00:49:12): Oh, yes.

Justin Mannhardt (00:49:13): Oh, yeah.

Rob Collie (00:49:13): Do they have the best tools there? I mean, they might, but I don't have that same confidence in their RPA offering for instance. When I'm saying this, it gives me the heebie-jeebies about using this sort of breadth first testing approach with software platforms. I could see where that could go sideways whereas using it as a mindset within a project is the way.

Kristyna Ferris (00:49:36): I think of it like you need an MVP. You have this whole pie that you want to put somewhere and you want to bake it a certain way, but instead, let's just take a slice of this pie from crust to toppings. I want to know, is this pie good? Is this one going to work? You don't have to build the whole thing just like a little bit.

Justin Mannhardt (00:49:53): Pie. Ooh, that's a good one. I'm hungry.

Kristyna Ferris (00:49:55): Sorry.

Justin Mannhardt (00:49:56): I come back to this idea all too infrequently. How can I cut this project in half or how can I deconstruct this to really understand if I'm going to have a false start here? And I think that's where the sunk cost fallacy that you mentioned, Kristyna, is we get so committed to this solution and forget about the problem. Oh, now we're down this road and we have to figure out how to do this. Someone is like, "Oh, this will take us eight weeks. How can we make it four? How can we make it two? How can we make it one?" What do we do today to see if we can confirm if we still like the path we're on?

(00:50:30): If we all feel like we're still on the right path, then let's keep going. But as soon as we feel like we're not, it's okay to reframe our commitments to our own choices. Every few months we realize like, "Hey, we wanted to go down this path. We're going to take a left turn now and we're going to take all the good stuff we learned going down the path with us, and we're going to leave a lot of stuff behind."

Kristyna Ferris (00:50:49): And we're going to take the questions with us that we wish we would've asked three months ago that we didn't know we needed to ask three months ago.

Justin Mannhardt (00:50:56): And nobody needs to feel bad about it. We want to fail fast ideology about almost everything.

Rob Collie (00:51:01): Or know that we failed fast.

Justin Mannhardt (00:51:03): Yeah.

Kristyna Ferris (00:51:03): Yeah. Oh my gosh. Oh, that's so true.

Rob Collie (00:51:06): That's the thing about fail fast is it seems like if you just take those two words, it's like...

Kristyna Ferris (00:51:11): You have to know you failed.

Rob Collie (00:51:13): I think I want to try to win fast, but know if I'm failing fast.

Kristyna Ferris (00:51:17): It's true.

Justin Mannhardt (00:51:18): It's funny on the know if you failed or if I know that it's not going to work, that's a win.

Rob Collie (00:51:24): Yes.

Justin Mannhardt (00:51:25): I've obtained knowledge that I didn't have before, and so we've had guests on this show before, customers, and they've talked about our jumpstart product, how it's been successful and impactful to them. It's like I've even asked people before we've done a jumpstart, I'll be like, "Hey, Kristyna, if we do this and it's three days, it costs this amount of money, we realize it's not going to work, and we learned that. Would you consider that a valuable use of resources?" And it's amazing how often people say yes. Most of the time it's not a failure. I want to be clear about that, but it's also the spirit of we're in the pursuit of improving our business. We're in the pursuit of learning what's really going to work.

(00:52:02): Sometimes you go into a project thinking, "We're going to solve it with these tactics, and you get into the data and the analytics and you learn something about your business you never would've thought to have wanted to know or learn," and you realize, "It's not this, it's that." So let's go in a different direction.

Rob Collie (00:52:17): People who are in the market to hire consulting firms. I bet they're really, really not accustomed to being asked that question. Obviously, everyone is hoping that it works.

Justin Mannhardt (00:52:27): Of course.

Rob Collie (00:52:28): But to be asked that question means that if it turns out that the tech just isn't a good fit for the problem, the consultant is going to tell you that. That's amazing because that's not what 99% of consultants do. If it turns out to not be a good fit, the answer is just to spend more. That's the advice.

Justin Mannhardt (00:52:50): Some people that have worked with me closely who know this, I can be pretty cavalier. I'm like, "Yeah, let's go after that hairy idea." But sometimes when they're really hair, I'm like, "Rob, just so we're on the same page, we may wind up in a dumpster fire. Are you cool with that?"

Rob Collie (00:53:04): And I'm like, "Right you are, Justin. I love Dumpster Fire. They're my favorite."

Justin Mannhardt (00:53:07): Let's go.

Rob Collie (00:53:09): That's our philosophy. Dumpster fire fast.

Kristyna Ferris (00:53:11): Let me go get my suit. I'm ready.

Rob Collie (00:53:15): So along those same lines, I do want to tell you, not a joke, but a story that my grandfather told me many years ago. So he and his friends went up into the mountains. It was like a camping trip, horse riding, camping trip, and apparently the guy that they'd rented the horses from hadn't properly fed one of these horses for this trip. So they get up into the mountains and they're walking along this trail and this one horse just collapses. And they've got this guy with him who had done a year of veterinary school before he quit. That guy, his name was like JD or something like that.

(00:53:52): And so JD takes charge of this situation. I stayed in the Holiday Inn last night. He says, "Okay. Help me stand the horse up." So they stand the horse up. "Okay. Now give me that fifth of whiskey." This isn't a joke. This actually happened in real life. So JD holds this fifth of whiskey and pours it down this horse's gullet. "Okay, hold him for a little bit. Now, let him go." They let him go and the horse collapses and slides down the mountain.

Kristyna Ferris (00:54:25): Oh, no.

Rob Collie (00:54:30): It's horrible. It's just gone, and they're all standing there at the edge of the mountain looking down at this tragedy and there's this long pause and then JD turns to them and says, "If we'd had another fifth, he would've made it."

Kristyna Ferris (00:54:40): No way.

Justin Mannhardt (00:54:40): Get out of here. JD, go home.

Rob Collie (00:54:49): But that is the consulting parallel, right? You spent $500,000 on this project and it's not working, but that's because you haven't spent a million. That's what you don't want.

Kristyna Ferris (00:54:59): Such a lie, yeah.

Justin Mannhardt (00:55:01): I have a question for Kristyna.

Kristyna Ferris (00:55:03): Yes.

Justin Mannhardt (00:55:04): You've been in the consulting game for a long time. You're now here with us. You've worked on lots of different projects with lots of different customers. You're popular in the community, active in the community. You've got a blog., you show up on other live events. Fabric, we've had some exchanges about doing some things in Fabric. I know you got a little party with some of your colleagues doing some things. I anchor this in the FOBO because I feel the OBO from a lot of angles from our customers. "Oh, Justin, what should we be doing about Fabric? I hear it from our team here at P3 like what should we be doing about Fabric?"

(00:55:35): What's the honest answer? What should business leaders be thinking about Fabric? What should practitioners be thinking about Fabric? Is it hype? Are we behind? What do you think?

Kristyna Ferris (00:55:44): You're not behind, but I would caution and say you're not behind yet if you look at it from a purely technical standpoint. So let's start there. That's where my brain wants to go at the moment. But if I start purely technical and I look at it, you're not behind yet. This is not something I think that is ready to go for every single use case, and there's infinite possibilities with it. I think that there are really great possibilities for it, and I do think that for some people, this is the way they're going to want to go right now, but I think now is the perfect time to start learning it.

(00:56:16): It's pretty cheap right now. There's still trial licenses going on. Now is the time to start looking for people in the community using it and watching those videos. Now is the time to start hearing people who have had successes or failures with it and start looking for those. Even if you are not using it yet, just see how other people are using it because nine times out of 10, those use cases are going to pop up in your experience. And when they do, you'll know Fabric is a good fit. But you're not going to know just avoiding it or by digging too deep into the technical differences between that and Synapse or that and Snowflake, and that in any other service because those weeds, they don't paint a full picture.

Justin Mannhardt (00:56:58): The curiosity I have there, Kristyna, is I think there's a lot of noise and nuance and confusion about should I keep working on the project I'm working on your typical Power BI project. We're building models. We're building reports. We're building analytics. If we're not using OneLake and Direct Lake and you're not missing anything that's going to dramatically change your business value proposition.

Kristyna Ferris (00:57:22): No, and I completely agree with you on that. From a Power BI perspective, you're totally okay. If your lifeblood is Power BI or your lifeblood is reporting, honestly, Fabric doesn't have to change anything you're doing. It doesn't. You can continue on with what you're doing. The biggest change I think that has come out of that is how you get your data. So it's like Power BI step zero is the only impact that Power BI users are going to feel right away is how am I getting my data?

(00:57:54): And that might not even be your team. That might be a data engineering team. And so that team then is reevaluating what they're doing. If I'm building ingestion pipelines going to Synapse, for example, that's a pretty good use case for me to stop and say, "Hey, I'm building something new. Is it a good thing to build it in Synapse or should I switch it to Fabric?"

(00:58:15): And that's a discussion that they're going to have to have and that they should have. They should look at what's different. They should look at those technical weeds we were just talking about. But they have a use case for building something new that Fabric can solve. Power BI, it's golden. We get even more benefits now with Fabric if you want to change up stage zero. If not, you're okay. You're doing great. You are rocking and rolling. You're in the winner circle, baby. You're here. You know what workspaces look like. You know how to publish things and share things. None of that is changing.

Justin Mannhardt (00:58:50): Kristyna Ferris, solution architect, P3, you're in the winner circle, baby.

Rob Collie (00:58:56): If you're in Indianapolis, you're drinking milk.

Justin Mannhardt (00:59:00): That's T-shirt worthy. That might need to be going on your personal brand. You're in the winner circle, baby.

Kristyna Ferris (00:59:07): But honestly, that's how it feels sometimes. You look at these Fabric things and you're like, "Oh, I know this. I've been in Power BI for years. I feel like I understand what's happening."

Rob Collie (00:59:17): From a data practitioner perspective versus a business leader perspective, the question, what to do about Fabric, right? It's completely different, completely different questions. But what to do about Fabric if you're a data practitioner question, there's a little bit of that whole. You might have to learn it for its own sake because you might not have a direct application today for the new XYZ thing that's in Fabric. But it wouldn't be bad to be familiarizing yourself with it.

Kristyna Ferris (00:59:45): Yeah. You might not need to know how to do it, but you might need to know what exists.

Rob Collie (00:59:49): The more I think about it from a business leader perspective, the question, what to do about Fabric in the end, it is a poorly formulated question. We've been talking about what to do about AI. And how that's a poorly formulated question.

Justin Mannhardt (01:00:02): What to do about the mosquitoes?

Rob Collie (01:00:06): Because it's not working backward from a business need. It's not working backward from a business value. It's working forward from a tech, right? Well, if you think about Fabric as being like something that is to some large percentage, not a hundred percent, but some large percentages is a wide broad tech platform that is meant to facilitate AI, it's even less sensible from business leader standpoint to be asking, "What should I do about Fabric?" than it is to be asking, "What should I be doing about AI? What are the problems that you're trying to solve?"

(01:00:35): It'd be more sensible in a way to be talking about, "What should I do about AI than what should I do about Fabric?" Now, if you're a business leader and you have two completely different kingdoms of tech at your company, you've got a big investment in data engineering and data scientists, and then you've got a big investment in BI and they're completely separate, like the Venn diagram has no overlap. There's so much that you can gain by slowly uniting those two kingdoms, unifying allowing them to benefit from each other. That in itself could be a mission.

Kristyna Ferris (01:01:07): Yeah.

Rob Collie (01:01:08): Now, you might have reasons to use Fabric, even if you don't have the data science and data engineering kingdom, and a lot of our clients probably don't, right? They probably are more just in the BI kingdom today, and this can be their first toehold in other things. But why? If you're in the situation where you already have the two kingdoms, you have those two kingdoms for reasons. They're already providing business value. It's easier to identify how allowing them to benefit from each other will provide more business value. So from a business leader standpoint, does the question what to do about Fabric? Is it already setting us up to fail?

Kristyna Ferris (01:01:43): Oh yeah, a hundred percent. All you're saying is you're assuming that Fabric solves a need for you just by asking that question. There's an assumption and you need to ask yourself, "Okay, what's my need?" It has to be what should we do about this thing that maybe Fabric can solve?

Rob Collie (01:01:57): It's like that time years ago when I was working on Excel, and this one guy asked me, "What are you going to do about Bluetooth?"

Justin Mannhardt (01:02:06): I don't know.

Rob Collie (01:02:08): It turns out nothing. I'm going to do nothing. Nothing is the right thing to do about Bluetooth with my Excel BI features. Boy, did he feel like he was rocking me on my heels when he asked me that question and he was actually [inaudible 01:02:22]. I was like, "Oh, no, I don't have an answer." Hadn't even been introduced yet. There weren't any mobile phones that used it or anything like that. And I'm sitting there going, "WTF is Bluetooth."

Justin Mannhardt (01:02:34): Like shaking my core understanding of what Bluetooth actually is. Why would I ever need Excel to connect to Bluetooth?

Rob Collie (01:02:43): But at the time, I guess it was plausible that Bluetooth could have been some sort of data transfer protocol, right? And it kind of was.

Justin Mannhardt (01:02:49): Well, it is. Yeah.

Rob Collie (01:02:50): It just wasn't known yet as the, I think, the hands-free audio.

Justin Mannhardt (01:02:55): Maybe Bluetooth will be the solution to connecting to Excel online.

Rob Collie (01:03:01): No, I think that's RFID. Don't be silly.

Kristyna Ferris (01:03:06): I've never been a business leader, so it is harder for me to envision those kind of questions. I'll just say that. But I do think even just talking with a lot of business leaders or seeing what they want in their reports, a lot of times your entire goal is to make sure things are running smoothly, running efficiently as possible, and that you are providing some value to the larger community that you're in, to the larger industry that you're in. And I think that it can get really easy to lose sight of that in the face of fear.

(01:03:37): It can be a little bit like opening a door to a brick wall where this should be giving me a pathway to be more insightful and effective, but instead I'm being slammed into this wall of fear about the new thing. It doesn't have to be like that.

Justin Mannhardt (01:03:52): Boom. Kristyna, I think you're working on one of these projects where clients have come to us and they want to get into Fabric. They want to implement Fabric. They want to do something with it. What does that mean in the context of what you're working on? And that might help some of our listeners think about what they might work on, and we've talked about how it's a lot is still new and fail fast, all that. So what are you up to these days, hands on keyboard?

Kristyna Ferris (01:04:14): I'm currently working on a project where we're taking them out of Azure Synapse and putting them into Fabric. So we're using their same source system, the same CRM, and instead of using Synapse notebooks and Synapse pipelines, we're actually converting those into Fabric. So we're doing kind of a MVP approach kind of that small piece of the pie that I talked about, and we're going to scale it up as we go along. But to start off with, we want to go soup to nuts with this one focused area.

(01:04:46): So the first thing we did obviously is ingest some data into Fabric using pipelines. And then we're actually using warehouses as our medallion architecture. So we're going from a raw zone where we're loading data in exactly how it looks in their current on-prem sources. And then inside of the silver layer, we're doing some light transformations, just some naming convention, data cleansing sort of thing. When we get into gold, what's ready for Power BI, we're spinning them into some views that are going to be more practical and have more business logic.

(01:05:19): That's what we're doing in a nutshell. We've played around with a couple of different methods for those things. Like you can use data flows to do your transformations. You can use notebooks. They're working on getting notebooks to work with warehouses. Right now they only work with lake houses. I'm sure by the time this podcast actually comes out, it probably will be available to work with warehouses. But that's the joys of Fabric is if something doesn't quite work, you just look at the roadmap and it's probably coming.

Justin Mannhardt (01:05:47): Love it. One question I have for you specifically is the synapse versus Fabric debate.

Kristyna Ferris (01:05:51): Yeah, it's a big debate.

Justin Mannhardt (01:05:54): Big debate. One of the things I've heard from a few customers and even recently from prospective customers is they have this feeling that their synapse environment creates this black box effect for their business users. And so their question is like, "Oh, if we moved over to Fabric, do we really create more transparency and understanding of where data comes from, where it goes, what happens through it along the way?" Do you feel like that problem gets solved in Fabric? If so, to what extent?

Kristyna Ferris (01:06:23): It can be solved that way? I don't think it necessarily is solved that way. Let me just put it in perspective. In workspaces, you can hide entire workspaces from people by just not giving them access. So if you had all of your ETL in a separate workspace from the end reports, which is a pretty valid model for an IT group, if you were doing that, you essentially still black boxed where the data comes from to the business. The only people who now have more visibility are your Power BI tenant admins.

(01:06:54): So in order to really solve it, you'd have to think through your security model, and you have to think through, "Okay, do I want people to be able to see these things without editing it?" If so, then yeah, I can probably give them view access to those items or to those workspaces, and it's no longer a black box. I could use things like task views and I could put in there how things move from A to B to C to their report, and they'd be able to understand where their data is coming from that way.

(01:07:24): So there are ways to do it, but there are also ways to have it still be a black box. So I think if you want data transparency, you're going to need to build it with transparency in mind.

Justin Mannhardt (01:07:35): I'm glad you brought that point up because there's insane value in having the people that are going to actually leverage solutions to the benefit of the business involved in the creation of them.

Kristyna Ferris (01:07:46): Yes.

Justin Mannhardt (01:07:47): So if you have this black box situation with Synapse, Snowflake, something else, the technology doesn't necessarily remove the black box to your point. You might have very valid reasons why people do and don't need to see certain things.

Kristyna Ferris (01:08:01): Right.

Justin Mannhardt (01:08:02): Okay. I can see a Python notebook. I don't know how to read Python. This doesn't help me. But Fabric may present the opportunity for you to bring a group of people together to collaborate and build that transparency of the business problem, build that transparency around what actually needs to get solved, and feel much more confident in the accuracy, the usefulness, sectionability of the solutions you're creating. It seemed like the classic stories of trying to validate reports and all along we just realized the old report was wrong in the first place.

Kristyna Ferris (01:08:32): Oh my gosh. That happens so often.

Justin Mannhardt (01:08:35): So I think, the data transparency really is achieved by the act of doing it together more than it is that you picked one tool. Maybe Fabric makes it easier than other tools. I don't want to discredit Fabric too much here, but it's still the human part of it and the business application part of it, alignment to the problem part of it.

Kristyna Ferris (01:08:55): And I think too, documentation is also tricky, and it always has been, and it always will be.

Justin Mannhardt (01:09:00): I hate documentation

Kristyna Ferris (01:09:02): Who doesn't? This is where I'm hoping Copilot comes in. I'm hoping Copilot can take a look at my repo because now everything can get integrated with a DevOps repo, and it can go ahead and document it, and people can ask questions on how things got from one source to another, and it can give them a human readable answer. That's my hope for the future. So I never have to write documentation again.

Justin Mannhardt (01:09:26): Well, Rob thinks we'll have a DAX Copilot in 18 months. We better have documentation in Copilot.

Rob Collie (01:09:32): If they're smart, they'll charge a thousand times more for the documentation, Copilot. Go make all the formulas.

Kristyna Ferris (01:09:43): And honestly, that would be such a huge win for accessibility too. So my dad, he's lost all ability to type. He does a lot of things with like eyegaze software and voice to text and Copilot stuff that's coming out, I am so thrilled. It's going to keep him able to engage a lot easier. He can have summaries written for him of things and he can just tweak it. That is so much easier for him than typing everything out.

Rob Collie (01:10:09): Oh, I bet. All of that volumetric grunt work that is exhausting even when the input devices are as efficiently designed for you as they possibly can be, and if suddenly the most efficient input devices are no longer available to you, give me some AI.

Kristyna Ferris (01:10:29): For real.

Rob Collie (01:10:30): Even more so. Speed is a big deal, and speed is often the difference between doing it and not doing it.

Kristyna Ferris (01:10:36): What's the point of having a quarterly report that comes out once every six months?

Rob Collie (01:10:39): Yes. Well, Kristyna, thank you so much for taking the time to record this with us, and so pleased that you're here.

Justin Mannhardt (01:10:46): Ta-da.

Rob Collie (01:10:47): Genuinely thrilled that you're part of the P3 team.

Kristyna Ferris (01:10:50): Very excited to be part of the team and so excited to be here talking to you guys.

Speaker 2 (01:10:54): Thanks for listening to the Raw Data by P3 Adaptive podcast. Let the experts at P3 Adaptive help your business. Just go to p3adaptive.com. Have a data day.