Confessions of a Data Gener w/ Sakiko Stickley

Rob Collie (00:00:00): Hello friends. Imagine a world that just makes sense, one in which organizations make the right decisions about how to manage their data. One in which the most efficient, highest ROI path is always chosen. The right projects are greenlit with the right methodologies, without waste, without lining someone else's pockets in the process. And then in this same world which we're imagining that makes sense, the right decisions are then always made from the data, where the success of the organization is truly priority one and divergent individual priorities and incentives don't lead things astray. That world is possible and we certainly enjoy helping our clients make that world their reality.

(00:00:47): But most people still live and work in a different world, one in which waste, inefficiency and frankly selfish enrichment drag things down. And the data gene crowd largely just has to sit there and quietly watch it all happen. Can't even speak up, because sooner or later you'll be punished for speaking up. Advocating for the right thing can be hazardous to your career.

(00:01:12): So in general, these stories of inefficiency and waste, they kind of rarely get told, but then someone like Sakiko Stickley shows up and is ready to tell all the stories. Now I learned something powerful about Japanese culture in the course of this conversation. Pop culture movies and TV shows in the U.S. have painted a picture for me of Japan as a rigid, respect-one's-superiors kind of culture, one in which dissent is rare and suppressed. And while that makes for good TV, it skips over an important detail. It's actually really hard to fire people in Japan. So dissent and speaking up for the right thing is not disincentivized in the way that it is in the States.

(00:01:54): Sakiko has seen a lot in her 25 years spent in the world of business and data. She has so many jaw-dropping stories that number one, we couldn't fit them all into one recording session. And number two, some of them were frankly just a little too hot to touch. We talked about a few things where even I became squeamish about putting them on the air. Plenty remain however, and we titled this one, Confessions of a Data Gener, even though she actually has nothing to confess. It's more like spilling the T about other people's misbehavior.

(00:02:24): The takeaway for data practitioners and business leaders I think is to use these stories not through the lens of the failure, but through the lens of how they could have been better. And since Sakiko is the kind of person who doesn't just criticize things, but instead always brings an alternate and better solution, you'll get a sense of how it can always be better. As a bonus, in our experience in particular at smaller and mid-size organizations, we find that the political inertia doesn't get in the way of the right thing in the way that it often does in enterprises. So even horror stories about colossal waste and mismanagement can be a source of positive inspiration when you turn them on their head.

(00:03:02): Let's welcome Sakiko Stickley to the podcast and let's get into it.

Announcer (00:03:08): Ladies and gentlemen, may I have your attention, please? This is The Raw Data, by P3 Adaptive Podcast, with your host Rob Collie and your co-host 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 (00:03:43): Welcome to the show, Sakiko Stickley.

Sakiko Stickley (00:03:46): Hi, how are you today?

Rob Collie (00:03:48): I'm doing well. It's daylight here, it appears to be daylight there. Where exactly are you?

Sakiko Stickley (00:03:53): Yeah, I'm based in Tokyo.

Rob Collie (00:03:55): So it's morning?

Sakiko Stickley (00:03:55): Yeah, it's morning. Yeah, 8:00 o'clock in the morning.

Rob Collie (00:03:59): Yeah, we managed to find a time that works for us and you and I have known each other's names for a very long time, like 12 years.

Sakiko Stickley (00:04:07): Yes.

Rob Collie (00:04:07): Never met face-to-face, I think first time talking live. Yeah, you're one of the original crowd, the OG that started with Power Pivot.

Sakiko Stickley (00:04:16): I first learned about Power Pivot from your blog post in 2013, and I read your book and it was really funny and I really enjoyed it. I also took your online video training course and it was the most enjoyable learning experience I had in my life.

Rob Collie (00:04:34): I love hearing that. It's so heartwarming for me. I mean, you're saying that my book, a technical book?

Sakiko Stickley (00:04:40): Yeah, yeah.

Rob Collie (00:04:41): You describing it as really funny.

Sakiko Stickley (00:04:43): Super funny.

Rob Collie (00:04:44): It's the best thing. It's the best thing I could ask for.

Sakiko Stickley (00:04:47): I actually laughed out loud while I was reading. And I actually read during a day because it was such a page turner. I sometimes try to read Marko's book and it was too technical and I couldn't finish it, but your was really a page turner.

Rob Collie (00:05:07): I appreciate that. I like to describe my writing style back in the day there as sort of like the entry to intermediate level courses. If you want to go to the graduate course, you end up with a different professor than me. Obviously really haven't written any books since then. 2015 was the last time I really seriously engaged with a book. I enjoyed it, but boy, is it a lot of work

Sakiko Stickley (00:05:31): Yeah, I can imagine.

Rob Collie (00:05:32): So what's your current job? What's your professional history? Just really briefly for the listeners.

Sakiko Stickley (00:05:38): I have just over 25 year career and I started my career in accounting firms, Deloitte & Touche, PWC, and then KPMG. I always enjoyed working with data. The real turning point in my career was in 2013 when I just joined a new company as a financial controller, I was tasked with revamping the company's reporting system and the company needed to produce 100 P&Ls every month.

Rob Collie (00:06:06): A hundred P&Ls, that's it.

Sakiko Stickley (00:06:08): By business units and product lines, and for both actuals and monthly focus. Actually legacy process involved 100 repetitive Excel worksheets and I thought there's got to be a better way than this. And the problem stemmed from the limitations of two-dimensional Excel and I must search for a solution to this multi-dimensional reporting puzzle. I came across your blog post and discovering Power Pivot was like a lightning bolt moment for me, and I became completely obsessed. And the only other time, I remember being that into learning voluntarily, was when I was practicing piano. So I became totally obsessed when I found about there's this thing called Power Pivot, which can handle multi-dimensional reporting tasks.

(00:06:56): Within one month, I became moderately comfortable with it and my learning was very steep. When I got stuck trying to allocate back office costs to product lines, I found Chris Webb's consulting business online and asked him to write the allocation tax formula. Thanks to his formula, I was able to deliver a fully allocated P&L model by business unit and product line, something that would typically cost companies millions in SAP projects. Within one month of discovering your blog post, I updated it within 10 minutes of closing the books each month, and at that moment I knew my life had changed.

Rob Collie (00:07:35): Now imagine if instead of 100 P&Ls, it had only been like four or five, you might've decided to just do it the manual way. The fact that it was so far beyond what you could ever imagine doing manually was almost like a gift. There's somewhere in between, maybe even 15 P&Ls you would've done manually and it would've sucked, but instead it was 100.

Sakiko Stickley (00:08:02): I hate repetition, so I always try to minimize the input to get the maximum output. This is my style. So in that company, they were using Cognos. And I even rang Cognos to try to find a way where we can directly connect to our ERP and then Cognos, and then do this multi-dimensional reporting in that way. And then they quoted the price which is out of my limit, so I had to find some other way, and then I searched and searched. At that time, there was no ChatGPT, so the find was by Googling.

Rob Collie (00:08:41): Old-fashioned Google.

Sakiko Stickley (00:08:42): Yeah, old-fashioned Google. And I came across your blog post with a search, maybe search term like multi-dimensional or something like that. After that, I was kind of really hooked and I was really obsessed. And it even came in my dream at that time, really, really hooked on Power Pivot.

Rob Collie (00:09:05): And I had caught that same sort of fever. It was super surprising to me because I had worked on the product, I did not expect to catch that kind of fever for my own product. That was not my experience in all my years working at Microsoft. Like, whoa, this thing's really good. What a shock. So the 100 P&Ls, did you end up with one model that essentially had all of the data in it and then it was just sliced by business unit essentially?

Sakiko Stickley (00:09:29): That's it. That's it, yeah. I just downloaded the P&L transaction list down to cost of sales goods. They had tagged the product code for all the tagging below the COGS down from the operating expense. They only had cost center and cost center wasn't tagged, all we've got. And we had to use the cost center in order to allocate to that. I got stuck, so I could not think of the way to do that. So I found on the internet, there's consulting business by Chris Webb and then I contacted and asked him to write this formula, which I got stuck with and he helped me.

Rob Collie (00:10:07): Good old cross-join consulting.

Sakiko Stickley (00:10:09): Yeah.

Rob Collie (00:10:10): We've done some behind the scenes coordination on all the different things that we can talk about today, and there's a huge theme in here. Essentially organizations can be so much more nimble, cost-effective, and just straight up better with their data, and yet there's so much inertia that has organizations still spending millions of dollars and all kinds of time doing things the wrong way. And I can just sense your frustration, like you sent a spreadsheet of topics, it's just the facts, and yet I can read this spreadsheet and I can feel your pain.

Sakiko Stickley (00:10:50): Accenture is my arch enemy. I mean they have made my life more difficult. I try to make my life easier by using Power Pivot and efficient tools and they have destroyed what I tried to do. And in this company as well, I mean they're not actually engaged with Accenture since I joined this company. I can see that this outsourcing of accounting function is also Accenture's doing. I'm leaving the consequence of it. Actually next week, I'll be going abroad for the offshore handover of the accounting task for entire one week, all day meetings. So this kind of inefficiency, embedded inefficiency in the finance operation is what Accenture's done.

Rob Collie (00:11:37): Let's talk about that. Let me just play deliberately dumb devil's advocate. Why can't you outsource bookkeeping? Why can't you offshore bookkeeping? It's just "tagging things" as certain types of accounts, certain types of expenses. What's so hard about that?

Sakiko Stickley (00:11:58): Because Accenture or the decision-maker never did the bookkeeping themselves, I can tell this. They never done their bookkeeping in their life ever, I can tell this. Because where the reporting is just mapping of the structured data, but bookkeeping is converting the unstructured data coming in all different sorts of forms, like invoices from different suppliers and contracts like this or something into a structured database format, bookkeeping in SAP or other ERP systems. So it's not as simple as mapping and then applying the logic in the reporting. But I enjoy reporting more than bookkeeping, because it's more converting the information into structured data.

Rob Collie (00:12:48): And so it's labor-intensive because of the unstructured nature of all this information coming at you. And so because it's labor-intensive, there's a very strong temptation to replace that labor with less expensive labor, and the way that you find less expensive labor is to find it at great distance from the company. But when you do that, you lose the nuance. People who are inside the business have the nuance.

Sakiko Stickley (00:13:16): The most ridiculous thing is the bookkeepers cannot even determine the GA account and the cost center. I have to specify which GA account to use, which cost center to use in the communication email or background communication, which this is just ridiculous. And then how can they claim that this is efficiency?

Rob Collie (00:13:38): We know how they can, right? It used to cost X, now it costs less than X. If you zoom in on just those little line items, used to cost X, now it cost less than X, people start asking questions like, no, no, no, no, no, no, no, no, no, no. Don't ask questions. It's cost less than X. That's all you need to know, just sit down, shut up.

Sakiko Stickley (00:13:55): That's it. And then they claim that it was a successful project.

Rob Collie (00:14:01): That's actually really interesting. When you can convert a known cost into a smaller number and then essentially externalize a lot of that cost into places that you can't measure it, if you can do that, you can pull this off, you end up in a worse place, but it still looks like a success to the people who approved it. That's the thing, there is now a stealth cost everywhere, but because that stealth cost isn't going to show up on any line item, it's just going to make everything worse. They hide it.

Sakiko Stickley (00:14:36): But I think this stealth cost is showing in the numbers.

Rob Collie (00:14:39): But you can't attribute it. If you ran the experiment, you ran the universe both ways, your organization with inshore internal bookkeeping, and you would be able to see the difference in that A-B experiment. But because we're only exploring one branch of the universe, the one where we chose to do this, you just can't really prove anything. The people who are living it know.

Sakiko Stickley (00:14:59): And the quality of bookkeeping is much better in the mid-sized company.

Rob Collie (00:15:04): I would say almost everything is better in mid-sized companies, because everything's more clearly attributable back to its original cause. There's less political inertia to hide things like this. And we really enjoy working with mid-market companies as a result, that's our favorite type of client. And we have enterprise clients and the ones we work with, we enjoy working with them, but when we go out to acquire new clients, we explicitly go looking for mid-market clients because we can run faster for them and we're less likely to run into silly politics that blocks things.

Sakiko Stickley (00:15:37): I see that a big company have lots of bureaucracy and lots of hierarchy, I mean lots of approval.

Rob Collie (00:15:45): People who see what's going on in an organization and understand that it could be much better, that's essentially a large percentage of the people who work for our company today are people who were like that at their last company and just couldn't take it anymore. So they come here and we sort of by definition only do work that matters. But in Japan, I really don't know much about Japanese culture other than what I see on TV or whatever, there seems to be from a distance, more pressure in Japan to conform than there is in the United States. Talking to you, you sound like an American, and so do you stand out in your own culture as much as I suspect that you might? Are you the only person like this that you run into?

Sakiko Stickley (00:16:37): Maybe. In Japan, generally speaking, there's more job security than in the U.S. It's more free to say anything.

Rob Collie (00:16:47): Interesting.

Sakiko Stickley (00:16:48): If you want to get rid of somebody, you have to pay a lot.

Rob Collie (00:16:52): Oh, it's almost like tenure in the American university system. So this is fascinating to me. My impression of the workforce and the workplace culture in Japan is exactly backwards. You're more free to speak up?

Sakiko Stickley (00:17:05): When I worked in a very, very Japanese company, they were really speaking up. But American company, I think tends to be more top down. So depends, yeah.

Rob Collie (00:17:16): This is one of those today I learned moments. It turns out that the most top down and authoritarian companies in Japan are the American companies.

Sakiko Stickley (00:17:25): Italian as well. I used to work for Italian companies as well.

Rob Collie (00:17:29): All right, so I'm going to be moving to Japan. Sounds like my kind of place.

Sakiko Stickley (00:17:32): Depends. Maybe the company I worked in Japan was not typical Japanese company.

Rob Collie (00:17:39): You're just being polite. You're saying Japan doesn't want me.

Sakiko Stickley (00:17:42): No, no.

Rob Collie (00:17:44): No more Americans. Okay, so you do find yourself when you're speaking up about things that could be more efficient, people aren't generally speaking, turning around and looking at you like you have three heads.

Sakiko Stickley (00:18:03): Actually, my colleagues also have the same feeling as me. I mean if they are doing the hands on work themselves. So it does depend if someone's doing the hands on work in terms of chart of accounts harmonization, when we were asked to do this manually instead of digitally, people who are at the finance using SAP weren't happy. And also in terms of having to handhold the offshore accounting functions, for everything we have to handhold and become the middleman for those people to complete their work.

Rob Collie (00:18:43): Most of these stories involve some sort of really, really, really expensive software that never works. This one is the opposite. This is a story where you and your team were forced to do something completely manually without benefit of software when there was a software solution that would work. And to me, that seems like if one of the big four accounting firms were involved in making this recommendation, it seems almost like out of character for them. They missed a chance to sell you a $10 million software package instead of having people do things manually. Why do you think such an intensive data transformation problem decided to be done manually? Why do you think that that was even the recommendation? I don't even understand.

Sakiko Stickley (00:19:30): I think it was to maximize their revenue. I mean they stayed there like a parasite almost two years I think.

Rob Collie (00:19:36): Oh, okay. So it was manual data transformation, but it was being performed by labor at an hourly rate. So how did this result in revenue for Accenture? How did getting this manual process going benefit them?

Sakiko Stickley (00:19:51): They were always attending the meeting and the alignment meeting and those kind of meetings, because the headquarter people are busy, so they outsourced the running of the project to the Accenture.

Rob Collie (00:20:06): And the longer the project runs, I see. It's just a question of a lapsed time, just a lapsed time.

Sakiko Stickley (00:20:12): Because I thought this was so ridiculous, I researched about the better way and I found this technique called sub-landscape optimization, which is done by boutique consulting firms specializing in that kind of thing with their special software. And I found a couple of those companies which can do that real digital harmonization with minimally invasive method, but they did not want to take this route.

Rob Collie (00:20:38): Okay, so we've established that speaking up and offering a dissenting opinion is actually relatively common in the Japanese work environment. It's accepted.

Sakiko Stickley (00:20:51): It's not accepted as such, but maybe it does not mean that you'll lose your job.

Rob Collie (00:20:59): I think that's the thing I'm getting at. So on the one hand it's tolerated without people getting terminated, without people losing their jobs. It's tolerated. But on the flip side, it doesn't seem like it necessarily leads to change.

Sakiko Stickley (00:21:13): No, actually it wasn't a Japanese company, it was Italian company. And I tried to convince them but they already made up their mind, so they're not going to change. Actually, I tried to convince them for one year, tried to talk them out of it, but they made their decision. And Accenture also was kind of treating me like their enemy because they were-

Rob Collie (00:21:39): Yeah, you were. You absolutely were.

Sakiko Stickley (00:21:40): Yes.

Rob Collie (00:21:41): I've done this, Microsoft Bing, before it was called Bing, Windows Live Search, they brought in McKinsey to advise them on some data implementation, blah, blah, blah.

Sakiko Stickley (00:21:51): Yeah, [inaudible 00:21:52].

Rob Collie (00:21:51): And these four consultants in their suits stood in each corner of the room and just monopolize the conversation. I asked them when are they going to actually help us implement this data pipeline they were recommending. They said, no, we don't do implementation, we just give you the recommendation. And I looked at them and says, you don't actually get involved in help? At that moment, they looked at me and I looked at them and they knew that I was going to be an enemy from that point forward. Behind the scenes, politically manipulated things so that I wouldn't be invited to any of those meetings anymore because I was going to endanger their multimillion dollar, yada yada, just talk and make slides amazing grift.

(00:22:25): It was very educational and yes, I got sidelined from that in an eye blink. I mean it hit me within hours, I was gone. I was not on that project anymore.

Sakiko Stickley (00:22:39): I wasn't sidelined because they had to depend on me.

Rob Collie (00:22:44): They needed you.

Sakiko Stickley (00:22:45): We only have a few people locally here to do the work. But the Accenture people was putting a real pressure on me. I mean they said that all the other subsidiaries accepted, only Japan is dissenting. And just kind of a group think kind of a tactic and it's only you.

Rob Collie (00:23:05): The world is made of people. People's incentives are wrong. You're in a big organization and you're in a position of power, your number one job, your number one incentive isn't to do a great job. Your number one incentive is to not get fired. Anything that looks like you made a mistake increases your chances of getting fired. So what you would never ever want to do, again if you play this out, is admit that the decision you made to do this manual process costing bajillions of dollars, if you were the person who made that decision to green light this, if you raise your hand at some point and say, hey, that was a mistake, we should do this other thing instead, you're now taking a risk of getting fired by someone as short-sighted as you.

Sakiko Stickley (00:23:48): Right.

Rob Collie (00:23:50): And so it kind of makes sense in a perverse way that a decision like that is one that gets followed through no matter what.

Sakiko Stickley (00:24:00): At that time when I started dissenting, they were only a few months into the project, so they haven't had so much sunk cost yet. So I thought that you can still correct the course.

Rob Collie (00:24:12): Totally. And that is the rational way to think about it, I completely agree with you. And then we turn around and we think about it from their perspective, whether their sunk cost is piled up or not, they have still said yes to a plan that's going to cost millions of dollars that didn't need to. So in terms of sunk cost of how dumb they might look they're already in, is the problem.

(00:24:36): But you and I don't look at it that way, I'm more like you. Why is it more important to have been correct yesterday than it is to be correct today? Which would you prefer? Would you rather be correct today or yesterday? I'm going to choose today. I'd rather be correct today, which means I have to be willing to admit that I was incorrect yesterday. It's no doubt being correct today is better, and yet a very, very, very large subset of humanity chooses to be correct yesterday over today, over and over and over again. And it's sad and it's really frustrating when they're making the decision to be correct yesterday and you want to be correct today and you have to listen to them. Ah, it's just the worst, isn't it?

Sakiko Stickley (00:25:27): Yeah. And I heard from the consultants who did the sub-landscape optimization, they can do the overriding of the old GL by the new GL within 48 hours.

Rob Collie (00:25:40): And then maybe some manual fixing, right? Some spot checking, some manual labor.

Sakiko Stickley (00:25:46): But as long as the mapping of old and new GL account sorted out, in terms of physically doing this just 48 hours and you wouldn't be losing any transactions in the balance sheet, which is not going to be deliverable.

Rob Collie (00:26:03): So there's multiple themes, one of the themes that sort of comes out of the list you sent over, it sounds like you've been pretty enthusiastic, an early adopter of a lot of different AI or at least experimenting with it. What are some of the places where you're finding AI to be very helpful and promising?

Sakiko Stickley (00:26:19): Even though there is a popular belief that AI will be automating and getting rid of the lower-end jobs, I see a future of AI as a strategic decision-making. Data-driven decision-making is inherently more strategic because it relies on evidence, trends and patterns over time, not just someone's opinion or opinion of the highest paid person in the room. A narrative-driven decision-making which is done by human on the other hand, often appeals to emotion and hierarchy or whatever story leadership wants to tell, even when the numbers don't support it. There's a widespread belief that AI will automate lower-end jobs while strategic decision-making will remain with humans, but I've seen cases where AI consistently outperforms human decision-making in terms of logic, consistency, ethics and integrity.

(00:27:16): That's why I'm skeptical of this narrative. [inaudible 00:27:20] position talk driven by those who want it to be true. It's a myth spread by the people with the loudest voices and not the strongest logic. I've seen first-hand that AI consistently is making ethical and more compliant and more sound decision-making without bias and without any cronyism like some humans make. So I think AI leadership will be the future which will make the future better for the corporations.

Rob Collie (00:27:57): When you first started saying that, it sounded like you were going to say no, AI isn't going to be able to replace lower-end workers, but what you're really saying is it's going to get all of us, like the CEOs who think they're safe.

Sakiko Stickley (00:28:14): I mean AI will be making much better decisions because they are based on logic, not to the kind of narrative they already committed themselves into and they have to bend the numbers to keeping that narrative.

Rob Collie (00:28:28): That is a fascinating concept. If the quality of your strategic decision-making at a particular organization is really poor or mediocre because of the factors you're talking about, because of cronyism, also just the incentive to not get fired, the incentive to not have there be a mistake. And by the way, if you're going to change course, if you're going to turn the wheel at all, you kind of have to admit that the heading you had before was wrong. So if all of these sort of political and real-world incentive forces are infecting your strategic decision-making to the point that it's really ineffective, why not replace it with an AI?

Sakiko Stickley (00:29:11): Yes.

Rob Collie (00:29:11): Again, I want to cross my fingers and say that if an organization has actually good decision-making, good human-driven communication that hasn't been corrupted by all these influences, it'll be harder for AI to be better than that. It'll be harder for AI to match that. But if you have a low bar, AI might already be able to clear it. I'm just going to hope that that's true, because what's left for the humans? It's just going to be the owners of these corporations and then the rest of us.

(00:29:40): But I'm very, very curious as to what your experiences with AI have been like that have led you to this conclusion. In order to get a comparison between AI strategic decision-making and human decision-making, you see the human-powered strategic decision-making all the time, plenty of examples of that. How do you get the test? How do you test AI at its strategic decision-making? You have to have been asking it questions like that and feeding it information, feeding it data so that it can answer the question. Am I correct in assuming that there have been cases where you've been trying this out saying, hey, what should we do based on this information and sort of play acting an executive with AI? Have you been testing it out like that?

Sakiko Stickley (00:30:26): Yeah, yeah. I've been asking some kind of ethical question and compliance questions which came to my way. Is this okay? They're doing this, is this okay? And AI says, no, it's not okay.

Rob Collie (00:30:39): It's so funny. The sky in that overlord is saying no, bad humans.

Sakiko Stickley (00:30:49): It's not okay. It's not ethically okay, it's not. So it put me in the right track.

Rob Collie (00:30:55): Suffice it to say that that is a sensitive topic, how to navigate that in the human organization once you have that advice from this AI, right? Careful, trade carefully. So if you ever asked one of the generative AIs, if you ever fed it a bunch of data and said based on this data, what kind of strategic decision do you think we should make, have you ever asked it a question like that?

Sakiko Stickley (00:31:18): Actually, ChatGPT is banned in the company.

Rob Collie (00:31:21): There's sometimes good reason for that, because unless you're in one of these temporary chats, the information that you're providing to it becomes part of its training data for the rest of humanity. And the next thing you know your company's internal policies are being cited in an answer to someone else's question, which is like, for example. So there's real concern with that, that needs to be figured out. No company is going to be able to effectively and permanently ban the use of these tools because of the tremendous competitive advantage that they provide.

(00:31:56): Now in a previous episode of this podcast, I took the output of a Power BI model and fed it into ChatGPT to have it do correlation analysis for me. I found that to be incredibly, incredibly useful. The idea of us as humans building data models that essentially structure and infuse the data with real world meaning and then allowing the AI to then go and do its thing with the output of that, I think that's a really powerful use case that we're going to be seeing more and more and more of. Why force a human being to look at 500 different reports, maybe the same report, the same dashboard, but with 500 different slicers selections, 500 different filter selections to reach some sort of conclusion about what's going on? Let a machine go analyze that even though it might take exactly the same effort to build that model today to hand to the AI.

(00:32:59): So it's like the AI is taking the place of part of the analyst's job that's using the dashboards and not necessarily as much of the person who was building the data model. And it can help with that part too obviously, no one's immune. AI is even going to replace ethical decision makers.

Sakiko Stickley (00:33:18): Yes, AI is very strong on ethics and it applies ethical logic consistently without any bias. That's what I've seen, yeah.

Rob Collie (00:33:28): I've been really blown away at some of the things I've been able to ask it to do lately. There were some other places where you mentioned sampling data for auditing purposes?

Sakiko Stickley (00:33:38): First time I was auditor in my first job, the thing which surprised me most was auditing was just on a sample basis and it's due to the limitation of the human limitation. The audit firms didn't have infinite resources. So I was very surprised that we only had to do sample testing. Now AI can do 100% checking and instantly and that's really is going to change the landscape. And any company, if they are comfortable exposing their SAP numbers and ask them the assurance and certificate from ChatGPT or AI that their numbers are correct and ties everything, that will send a very strong message about their transparency and their cleanliness of their books.

Rob Collie (00:34:31): And you can imagine there's all kinds of incentives to not do it that way. The audit firms don't want it that way and the companies being audited, it's an unknown risk for them.

Sakiko Stickley (00:34:41): Known risk as well. They know, they know.

Rob Collie (00:34:46): Known and unknown risk. That's another thing about large organizations, after a while, it's just guaranteed that there's something funky going on. I mean a guy I used to work with at Microsoft got caught stealing millions of dollars from Microsoft, had gotten away with it. And if he had just decided to not try to steal another million, he might not have ever been caught. It just blows my mind that someone can steal six to seven figures of money and not raise any eyebrows. And that's just one person. How many of him are out there right now even just at Microsoft and haven't been caught yet?

Sakiko Stickley (00:35:30): Okay. Even with their digital savviness, the surveillance system is more robust than-

Rob Collie (00:35:37): Well, so here's the thing. It looked like a legitimate transaction. And this is all public record, this is all in the federal court log. I mean I've read it all very carefully. He was basically using vendors whose relationship with Microsoft he controlled and he was running invoices through them and then having them route money to another vendor and that other vendor happened to be him. But he was able to requisition $750,000 of Microsoft's money for something that he made up. Microsoft didn't plan to spend that $750,000, he just got Microsoft to give $750,000 to something that he made an excuse for. Unbelievable, but you know this kind of thing happens in certain places.

Sakiko Stickley (00:36:22): Absolutely. I've seen that as well.

Rob Collie (00:36:24): Yeah. You had a manager who banned you from using Power Pivot?

Sakiko Stickley (00:36:28): I mean, absolutely. At one with the companies I worked, I had a micromanager who despite her many strengths had absolutely no data genes. And she told me not to use Power Pivot because she didn't understand the formula. At the time, I was totally obsessed with DAX and being told not to use Power Pivot felt like having lost my most important thing taken away from me, something only my worst enemy would do. I tried to convert her for three years, but it never worked. Then one day, temp staff member joined and I gave her the same explanation and within one minute she got it.

Rob Collie (00:37:08): The formula that wasn't understandable to the manager?

Sakiko Stickley (00:37:12): Yes.

Rob Collie (00:37:12): You said, okay, let me just try this temp worker, let me just try to explain the same thing to them and they understood it immediately.

Sakiko Stickley (00:37:15): Yes, immediately. And then that's when I realized you can't install the data genes. You are either born with them or not.

Rob Collie (00:37:23): This is something that I did talk about a little bit in a solo podcast as well about why isn't the data gene better represented in management? It does happen, there are managers that have the data gene. It is a somewhat rare trait, like 1 out of 16 people I think, but we would expect the data gene to be overrepresented in management because of the advantages it conveys. And especially in large organizations, we don't see that. We don't see that that's the case. If anything, it's under represented.

Sakiko Stickley (00:37:55): Absolutely underrepresented, and more narrative driven. So I have more stories to tell regarding this thing, and over time I began to feel like I was babysitting with my manager. She wasn't embarrassed to ask me to teach her bookkeeping, she was not embarrassed at all. And even though she was CFO, after yet another round of micromanagement and repeated requests to explain the same cube formula again, I I finally snapped and I told her that if she couldn't understand what temp staff understands, she shouldn't be a CFO. And she burst into tears and went straight into the CEO's office, and that same day I received a formal warning letter from the HR department. To her credit, she continued to treat me with compassion after that. She's very sweet, yeah.

Rob Collie (00:38:48): Isn't that tough, right? When the personal relationship in the business relationship are in different states. The personal relationship might be good, but the business relationship is strained. It's actually easier on us when both are strained.

Sakiko Stickley (00:39:00): Absolutely, yeah. We had irreconcilable differences and not personal, but in working style, she had a lot of energy and seems to have unlimited amount of time. And always wanted to work together with me on repetitive tasks in A1 plus B1 manner. On the other hand, I preferred working independently and disliked doing the same thing over and over again. And people have different talent in different areas and I learned the hard way that there's no point in trying to convert someone without data genes to think the same way I do.

Rob Collie (00:39:37): No, I agree. I think you're either sort of somewhat born with it. By the time you enter the workforce, you either have it or you don't. Now you might not have discovered that you have it. We've seen so many cases where it lies dormant waiting to encounter the right business experience where you take to it.

(00:39:54): I think there's another class of people in management who might not necessarily have the data gene that you and I very much carry, but they can recognize the value of it and those are our best allies. They might not be ones that are keen on DAX or want to build data models or could build data models. But if they understand, if you find yourself with a management sponsor who understands the value of you and happens to be good at all of the other things that make them a manager, the things I talk about in my podcast, like the power of persuasion is probably the single most important skill for professional success in an organization, is the ability to persuade, the ability to get people on your side. And so if they're a manager with those skills and they can recognize the value of your skills.

(00:40:53): This is a really important takeaway for people who are listening to this podcast as well, because a good number of people who listen to this podcast aren't people who are writing formulas. They're people who are making business decisions. They're business leaders. The advice to them is to lean into the things that make them effective members of the organization, lean into those powers of persuasion and bring a data gener or two along with you for the ride. As a team, you're going to be incredibly effective. Because the data gene crowd, the people who are good at building data models, we also occasionally luck into being persuasive people, but it's sort of the same lottery. If you luck into the data gene and the persuasion gene, I mean the world's your oyster. Go get them. This doesn't happen every day.

(00:41:41): The worst case scenario is when they lack the data gene and essentially fear it, because they don't understand it, they can't control it, they feel inferior around it. And here's the thing, if you're listening to this, it is perfectly human to be afraid of things that you don't understand. We are wired to be afraid of the unknown because we can't predict the unknown, we can't model it in our own heads so that we can be successful with it. So when you're uncomfortable with someone, like if you've got someone like Sakiko who is good at this stuff, it's intimidating and it's intimidating and it's frightening actually. And if you business leader just recognize that that's what's happening in your brain rather than just short-circuiting it and kind of shutting it down, if you are willing to sit with that fear just for a moment, it can open up such new possibilities, because the two of those things combined can really do amazing things.

(00:42:42): So that's kind of my plea. We try to do a little bit of both on this show. We try to talk about do things for the people who write formulas and do things for people who are managers and who are leaders. All of this can be better.

Sakiko Stickley (00:42:54): I was quite lucky the first time when I encountered Power Pivot, my manager, immediate manager, I used to report to a CEO, and he was not micromanaging me at all and he had better things to do than micromanaging me. He was CEO of three countries and he was always flying all over different countries and doing his stuff. So I was left alone to do the creative thing and do the thing in the way I thought was most efficient. And he recognized me for that and he said I solved the problem not in the surface level, but from fundamentally. And sometimes some leaders just want the big PowerPoint slide deck and they're not interested in this leader. And in that sense, I was really lucky that I had this CEO manager in my initial year of encountering Power Pivot.

Rob Collie (00:43:51): That was the 100 P&L company?

Sakiko Stickley (00:43:54): Yeah, the 100 P&L. And he just gave me this task, improve this and revamp this, and he just left me to do it on my own instead of telling me to put this in this cell, this cell, and he has no business with that.

Rob Collie (00:44:08): Make the font bigger.

Sakiko Stickley (00:44:09): Change the color. I mean that was micromanager was important, color. The first thing she commented after showing my Power Pivot was the color it should be.

Rob Collie (00:44:20): Last night I was told a story from within Microsoft. I'll protect the identities of the people and the products involved. The story was there were two products at Microsoft that were being developed in parallel that did the same thing. And Satya said, I feel like y'all should be able to figure this out and not do this twice you're currently doing. One of my friends put together a PowerPoint deck explaining the position of one of those products and gave it to his intermediate manager for review. And rather than say anything about the contents, the manager didn't say anything about the messaging in this deck. The only thing he said was, listen, this product that we're talking about here makes a billion dollars a year. This slide deck does not look like a billion dollars. I need this slide deck to look like a billion dollars. No feedback at all on anything in the message.

(00:45:14): The person giving this feedback about this slide deck was a corporate vice president. All it was was go fix the formatting, go find a professional, like spend money. If you need to spend $10,000 of Microsoft's money to make this deck look like a billion dollars, you go do that. We are not going to show this around. And it wasn't a bit of feedback about anything else.

Sakiko Stickley (00:45:38): Right.

Rob Collie (00:45:39): Happens everywhere.

Sakiko Stickley (00:45:40): One of the thing which impressed me about Microsoft is their approach to try to solve the problem fundamentally instead of each doing repetitive thing one by one. This is a thing which impressed me about their strategy. For example, like back in 2015, Microsoft Japan's website Japanese language was obviously machine translated. It was not natural Japanese flow. Why don't they spend on professional translator to do the job? They must have been working on a bigger thing. And nowadays, machine translation is, I mean more perfect human translation. So nowadays completely, some companies spend a lot of money on professional translator, but Microsoft seems to have not gone down that route and just what doing from 10 years ago the machine translation and their website language was not really natural.

Rob Collie (00:46:37): Have you heard about the Backstroke of the West?

Sakiko Stickley (00:46:40): No.

Rob Collie (00:46:41): The Backstroke of the West was the American movie The Empire Strikes Back, was translated into Chinese in China, and then someone made a bootleg copy of it and worked from the Chinese language version and translated it back into English and put subtitles. The audio track of the movie is in Chinese, but they put English subtitles. The translation round tripped from English to Chinese and back to English and The Empire Strikes Back came back translated as the Backstroke of the West. And everything in the dialogue in these movies, in fact they translated all the first six Star Wars movies like this. There's a whole series of it. And so the trip translation turns out to be really, really, really funny, even though human beings were doing it at both steps. And maybe they were doing it faithfully, but you just simply cannot do that. It does not survive the round trip at all. Think about it, the Backstroke of The West, the West is the empire to China, right? Like the occupying imperial powers of the West. That's empire< that's how that happened. And then strikes back becomes the backstroke.

Sakiko Stickley (00:47:57): Oh wow, completely changed.

Rob Collie (00:47:59): That's not the way to do a Microsoft website. They clearly don't approach it that way. Here's something, are you still exclusively using Power Pivot? Have you started to use Power BI yet?

Sakiko Stickley (00:48:07): Nowadays, I mostly use Power BI, except for the time I tried to do the bookkeeping with Power Pivot.

Rob Collie (00:48:14): Power BI is better in almost every single way. I finally relented. I was probably the last person on the planet who was stubbornly hanging onto Power Pivot. I have fully, fully transitioned over to Power BI. I haven't built anything in Power Pivot in a very, very, very long time now, it's been years. In an average week, how much DAX do you see?

Sakiko Stickley (00:48:36): When I joined the company, I automate most of the work, like bookkeeping and reporting. And in this company as well, as soon as I joined, I automated. So after that, not a lot, but I actually hung around in the Power BI community and tried to find interesting questions to answer.

Rob Collie (00:48:55): That surprises me a little bit, because every data model I've worked on, as long as it has remained relevant, I have always needed to evolve it. There's always been something new happening. Doesn't even have to be a changing business condition. It could just be like a new business question or realizing that I could have been doing something that I hadn't. If the data models and things that you built before aren't things that you need to revisit very often, that kind of tells me that you're not getting a lot of pressure from the organization to answer new questions.

Sakiko Stickley (00:49:30): Absolutely, not at all. This company I'm currently working is not so in terms of challenging, in terms of expanding my skillset, not at all. And once I automated it, and they only have one product, and so I don't have any place where I can utilize product by P&L formula. Yeah, so not a lot.

Rob Collie (00:49:54): If you were surrounded by data gene managers, you'd be getting all kinds of questions.

(00:49:59): So here's a funny story just from last night. It just tells you there is an importance of marrying narrative and context with what the data is telling you. If you lean all on one or all on the other, you're going to make mistakes. This is funny, so I made these dashboards for my hockey league that I don't even play in anymore. But every week they update them and there's a text conversation that happens about the update so that which stats should they highlight in the Facebook post when they post about the stats? It's really kind of a neat ritual and I'm really glad to still be part of this text thread.

(00:50:35): And one of the things I suggested to them was, this analysis that I had done just this past week, which was this one player, John had scored twice as many points this season as he had in the average of all of his prior seasons. He scored 2X his lifetime average in points this season. I think that's significant, right? That shows that this person had a great season. And maybe we should call that out, he should be highlighted in the Facebook post for having such a great season. And someone else chimed in and said, no, don't do that. The reason he has twice as many points this season is because he has hogged all of the time on the floor. Like in hockey, you go out and you play a shift and then you're supposed to come off and let one of your teammates go and play, and instead he's basically played twice as much This season he's played two times his fair share of the minutes on the floor. His score per minute is the same as it's always been. He just played twice as much, so we should not praise this.

(00:51:35): And the thing is, if we had statistics of how many minutes each player spent on the floor, we'd be able to call this out. But because we don't have that, we don't record how long people's shifts are, our data is inadequate to tell the whole story. I know it's a silly example from sports, but there's an example of where the real-world narrative, the data that's not in the data model is super important. We do not want to condone and reinforce what's essentially bad behavior.

(00:52:10): So I'm a believer in harmonizing narrative and I know you are too. You and I are not here to say that narrative and context are unimportant. What we're here to say is the narrative also needs to fit the data.

Sakiko Stickley (00:52:23): Sure, absolutely.

Rob Collie (00:52:24): You've got so many cool stories here. Your proof that the finance and accounting function can be so rich, and not just efficient and cost-effective, but also a strategic advantage in terms of decision-making capability. Again, only in an environment that is willing to accept it. You are one person in the early stages of your current role, eliminated the vast majority of the manual work on your plate. So many organizations are not aware that this is possible, that they could get a very small footprint of leaning into Microsoft's tool set. Not only save a tremendous amount of time and money, but become smarter in the process. You get more for less. You just almost never get that kind of opportunity and it's still kind of mind-boggling that that hasn't worked its way through the world yet.

Sakiko Stickley (00:53:26): Yeah. For example, this allocation by business unit and product lines P&L, I've seen a lot of manufacturing companies doing it within SAP. And I used the DAX recipe shared by Chris Webb to create this operating expense cost allocation model outside of SAP, keeping the SAP data clean and untouched. But traditional SAP allocations, not the system with a tsunami of micro-posted journal entries, they're hard to trace and add no analytical value and slow down the performance, and just extracting them can take 30 minutes, if your PC can even handle it. Using Power Pivot instead, allocating cost dynamically based on drivers like net sales or custom weights without any ABAP developers, no SAP projects and no ERP clutter. And when allocation logic changes, I simply adjusted the DAX month by month or dynamically over time using the relationship between the allocation table and the calendar.

Rob Collie (00:54:30): Data-driven adjustments rather than going and editing a bunch of formulas.

Sakiko Stickley (00:54:36): And also if you are going to do this within SAP, I've seen companies spend millions of dollars on this kind of project doing the allocation project within SAP. And benefits of using Power Pivot instead of doing it within SAP is that whole traceability back-to-invoice data and faster delivery and lower cost a the clean audit trail. And most importantly, there will be no junk data polluting to your SAP. Have you seen this allocation within SAP data in your work?

Rob Collie (00:55:09): I mean I haven't personally, but I'm positive that our consulting team has seen it many times.

Sakiko Stickley (00:55:13): Sure, yeah. A local of manufacturing companies do the same within SAP and they pollute. First time I've seen this, I was like, what are those journal entries? And they don't have any meaning and they just did it within SAP to allocate the cost of sales goods sold within SAP to the product lines. And allocating inside SAP is a legacy approach and it belongs to I think the 1990s. And SAP records what happened and Power BI tells you why and it's much better, cleaner and lighter weight.

Rob Collie (00:55:50): This is something that I've been really coming back to over and over again is that Microsoft's choosing to call their Power BI models, to call them semantic models again, is really smart. It's probably too smart. But the semantics of human meaning at the business level, that's what we're really doing in a Power BI model. We're bringing the underlying most fundamental data together with hardwired, unambiguous semantic meaning in the form of formulas, in the form of relationships. And also as you're saying here, in the form of auxiliary tables that don't necessarily need to be in the transactional record-keeping system. What a great saying. What was it? SAP tells you what happened?

Sakiko Stickley (00:56:39): Yeah, SAP records what happened and Power BI tells you why. So SAP is an excellent system for record-keeping, but they're not flexible. But I've seen many companies use allocation within SAP and they are polluting the SAP with junk data. And you want to keep the SAP data clean of what's relevant to what payment the company made to the suppliers and those kind of relevant cashflow impacting information rather than junk allocation data, which can be done virtually within SAP.

Rob Collie (00:57:14): That saying about SAP and Power BI, you are more eloquent in your non-native language than most people are in their native language. You're coining phrases in a language that's not your. I have to remind myself of that, because I don't speak a word of Japanese.

Sakiko Stickley (00:57:34): Thank you. Actually, my husband said that I should use, he said the ChatGPT voice for this podcast because it speaks perfect English.

Rob Collie (00:57:46): I believe that you might be multiple times more eloquent in Japanese than you are in English, and I'm kind of frightened by what that must be like. The day that I start coining phrases in another language, that day's never going to happen.

Sakiko Stickley (00:58:06): Thank you.

Rob Collie (00:58:09): Very, very impressive. What do you see is next for you? Do you expect to just kind of keep doing more of the same?

Sakiko Stickley (00:58:17): I enjoy working in a data-driven company where I can utilize my skills better. I mean, I will be most utilized in a maximum possible way in a data-driven organization rather than narrative-driven. I've worked in multiple companies and they have a varying good degree of data-drivenness. And some companies, they don't want to accept the data [inaudible 00:58:43]. I've had the experience where when I presented the accurate allocation by business unit and product lines P&L, the leadership didn't want to accept that. And there's a DAX formula that got me hooked on Power BI, not only helped me build some of the most transparent dynamic P&L models I've ever created, but it also got me golden handshakes at two different companies.

Rob Collie (00:59:10): What exactly goes into a golden handshake? They're like, look, here's some money, we'd like you to leave?

Sakiko Stickley (00:59:15): All I did was just apply the allocation logic consistently by business unit and product line, and suddenly everything became clear. At one company, the P&L showed that one division was carrying the weight of the others. And after the regional head was replaced by someone from a loss-making division, a colleague warned me that I might be pushed out soon because of the numbers I was showing.

Rob Collie (00:59:40): Too much truth.

Sakiko Stickley (00:59:42): And weeks later, my PC access was cut off and I was handed a severance package. I was with the company only for two and a half years, but they paid me one year salary and three months paid leave to disappear.

Rob Collie (00:59:55): Wow.

Sakiko Stickley (01:00:00): And another company, I was there for only two and a half months and I completed P&L automation by product lines using mapping and logic and all, and the legacy method relied on monthly emails and manual reclassification outside the mapping structure. When I pointed that out the reported numbers should be based on SAP data, consistently applied mapping and allocation logic, I was asked to leave with another golden handshake and two months of paid holiday.

Rob Collie (01:00:31): The only version of this that I'm aware of in the United States is either CEOs, and the only other type of people are college sports coaches. When they get fired, they keep getting paid, and the joke is they're paying Jimmy Bobob $5 million a year to not coach the Texas football team. I volunteer to not coach the Texas football team for $1 million a year. I think I could save them some money. It's kind of mind-boggling, right? An asset like you telling the truth and helping them see things more clearly, in certain moments you're considered a danger and they're willing to pay you to go away.

Sakiko Stickley (01:01:10): And so yes, Chris Webb's DAX formula is powerful and it doesn't just reveal numbers, it reveals the truth, and not every company is ready for that.

Rob Collie (01:01:22): DAX don't lie, unless the data lies. I really appreciate you starting your day with us. Thank you so much. I'm so glad to meet you after all these years. You're one of the ones that was there from the beginning, and I really appreciate that.

Sakiko Stickley (01:01:34): I'm really glad that I get to talk with you in person and I really enjoyed your books and training and you got me hooked on Power Pivot and DAX. And if I didn't find your blog posts, none of this would happen.

Rob Collie (01:01:50): I was very early and I was very enthusiastic. I got you started maybe a year or two earlier than you would've otherwise, but you would've found this.

Sakiko Stickley (01:01:58): Yes.

Rob Collie (01:01:58): You were on an inexorable collision course with these tools. They had your name on it.

Sakiko Stickley (01:02:04): 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.