Subscribe
Share
Share
Embed
A show to explore all data matters.
From small to big, every company on the market, irrespective of their industry, is a data merchant. How they choose to keep, interrogate and understand their data is now mission critical. 30 years of spaghetti-tech, data tech debt, or rapid growth challenges are the reality in most companies.
Join Aaron Phethean, veteran intrapreneur-come-entrepreneur with hundreds of lived examples of wins and losses in the data space, as he embarques on a journey of discovering what matters most in data nowadays by speaking to other technologists and business leaders who tackle their own data challenges every day.
Learn from their mistakes and be inspired by their stories of how they've made their data make sense and work for them.
This podcast brought to you by Matatika - "Unlock the Insights in your Data"
Welcome to today's show. We're joined by Bethany Lyons, an absolute gem of the data industry. She's gone from product management, you know, on the data vendor side, through to now consultancy, and she shares her stories and her advice to be a full stack data person. Let's get into it. Hello, Bethany.
Aaron:Welcome to the show. I really appreciate to have you on.
Bethany:Thank you so much for having me on.
Aaron:It's really, really great. So, also, I'm a little bit scared to be fair. I love what you do. You you, you know, commentate on start ups and data ventures, and you've worked in the industry a long time. So, yeah, I think maybe today I'm expecting a little bit of a challenge to what on earth we do, as well as what you do.
Aaron:So I'm starting to wonder whether I'm the one on the podcast or or actually you're coming on mine.
Bethany:I think yes. Welcome to my show. I'm gonna be doing the interviewing today.
Aaron:Brilliant. So tell us a little bit there because, you know, people are always interested in what people have done and and where you've been. Tell us about you. What what's your what's your career been today?
Bethany:Sure. So, so kind of my longest, stint was at Tableau. I was there for almost 9 years, doing both presales and product management. Then, like, I had a brief stint at a hotel tech company and realized I'd rather be in data. So then I joined a data analytics startup for about, I don't know, 16 months or so.
Bethany:And now I'm working at the one of the prospects of my last company who I was unable to close, and I was very mad that I couldn't sell to them. So I decided to join their organization and figure out what is going on here that you didn't buy my product.
Aaron:Always a good experience. Yeah. So the, the the the prospect and the you actually you're in the start up you're in. So you're you would in a sales type role. I mean, everyone's selling in a start up.
Aaron:Let's let's be totally frank. I You're a basic company.
Bethany:In the start up. So I was the chief product officer, which is basically the CRO in a startup because if you your job as the CPO is to validate the market wants to buy what you're building. So I did most of the biz dev and sales and, that that kind of stuff in that role.
Aaron:Yeah. The, so my experience is that you're absolutely right. I kind of add 2 other people to every sale. You've kind of got the CEO, founder, or head of product. You've kinda got the actual sales guy who is there to make sure the deal's working, to make sure that actually, you know, we have the right things in place.
Aaron:And usually, there's a kind of technical person who Yeah. Does the driving, does does the product stuff. That is to me, like, a secret formula of of of a sale and what it looks like. And if you try to do it all yourself, doesn't really work. If you try to do it with just a sales guy, it doesn't really work.
Aaron:Yeah. Sell to. Yeah. Challenging. So now that you're on the other side, do you see what went wrong in the deal?
Bethany:Is it A little bit. Yeah. It's so so the way I I I kinda think about it is we built, like, the perfect product for this client. They they absolutely need this product. They just aren't looking for it.
Bethany:It's it's kind of a so so the way it came to us was through a consultancy. That's where I work now is the consultancy on behalf of the client. The challenge is the client doesn't know that they need the product. And then it was like on the consultancy to introduce the product. But then the consultancy was like, well, we don't wanna ask for more money for more software.
Bethany:Yeah.
Aaron:Yeah.
Bethany:We'd rather we'd rather save that budget for services It's basically what it came down to. So it's like, how do you sell something to someone who doesn't know that they need your product?
Aaron:So I would love you to find out the answer to that whole the dichotomy because, like, consultancies and product are always in some kind of tension because they want hours. They want people. They want to sell a project. Yeah. And software can take away from that.
Aaron:And, you know, worked for a enterprise software company for a long time where exactly the same tension on the other side. They want more license. They want more software. They don't want to chip away at the deal because it takes a long time to implement. Yep.
Aaron:Yeah. So if you figure that out, let us know.
Bethany:Right. It's definitely, it's a tough one. It's, yeah. Interestingly, I'm now bringing into my clients a competitive product to the one that I built. And they're like, we love this.
Bethany:We need this. And I'm like, I know. You were in my sales pipeline for, like, a year and a half. I know that you need this.
Aaron:So then without turning into too much of a segue, what is the need?
Bethany:It's it's basically a warehouse native spreadsheet. So so the product that we're bringing in is called Sigma Computing. They're like the number one recommended analytics platform by Snowflake. They're backed by the same venture capital fund as Snowflake. And, yeah, it gives, like, a credit risk team, a market risk team, equity research, finance, treasury, like all of these different roles in the company, the ability to connect to broad transactional level data and do their own analysis on it.
Bethany:Because at the moment, these teams are all working in Excel, They have a million row limit. Every time they need to drill down to details, they have to go back into the source system to try and reconcile. So, like, the amount of hours, days, weeks, months lost in this organization to reconciliation is outrageous. I'm like, if you just do all of your analysis on raw transactional data, you don't need to be doing all of this reconciliation of aggregates that have, like, aggregated away all of the detail.
Aaron:Yeah. So that that sort of takes me back to when I started Matatika. I I had this general idea that conveying the right information was kind of the problem space, but it didn't wanna be competing with Tableau or Power BI or, you know, visualization was not really dashboards were not really solving the problem. So we we built this feed of insights that kind of surface the insights that everyone was most interested in through their engagement. And, you know, think of it as like Facebook for your data.
Aaron:Awesome idea, but, really, the first learning was that as soon as we showed someone a chart, they were like, yeah. What's behind it? And, like, they were like, I really wanna drill down, like and every single conversation since has been like, that is exactly what they wanna do. And that to me is, like, the fully the answer of why Excel is necessary and so prolific is because it starts from that basis. This is what's behind it.
Aaron:Now you build a a visualization.
Bethany:You build it. Yeah. You build from the ground up instead of reverse engineering from the summary data down to the detailed data. Totally. It's it's about trust.
Bethany:It's like people don't, you know, they don't just look at a chart and immediately trust it because it's pretty. They need to see, like, what made up that number.
Aaron:Yeah. Yeah. Yeah. Exactly. Yeah.
Aaron:I I totally like, the the whole trust thing, like, if I had to summarize what we do, data professionals as a data company, data software? It is actually to maintain trust because our data is a representation of the world. Everyone wants to know that it's a fair representation. Is it actually what's happening? And the most, you know, awful moment is as soon as they see something they don't trust, they don't trust anything.
Bethany:Anything that you put in. I am so scared to put data out, like, because of that.
Aaron:Yeah. Man, it's so damaging to a whole data teams can collapse as a business function because they're, like, put out something that was absolutely, totally, obviously wrong. And then you just, you know, you well, you can't trust what you guys are telling us now because, like, you're bonkers. That was that was, like, making like a revenue or whatever it might be.
Bethany:Yeah. I'm building this it's kind of a throwback to my my math days in uni, but I'm I'm building this linear program right now to do some automated reconciliation.
Aaron:It's a big actually on LinkedIn. Yes. And lots of comments about it.
Bethany:Yeah. You've got your bank transactions. And then you've got all of the things that the bank transactions are paying for. But usually what happens is, like, a client won't pay every transaction individually. They'll say, I owe you 100, 200 and 100.
Bethany:So let me just send you 400 in one payment, and you need to then match that 400 to the 3 transactions that it paid for. Which is like, on a small scale, you can just inspect the numbers and do the matching. But on a large scale, this becomes a really complicated problem. And so right before this podcast and the reason I was slightly late is I was like so deep in the math. Try to get it to work.
Bethany:And, yeah, the the problem is that, like, the bank transactions don't actually sum up to the payments because it turns out there's, like, the the payments is only from client data because that's what we have in the booking system. But then there's all this other stuff that happens with, like, money market makers and fees and and and stuff that is in the accounting system that you need to extract to match with the client transaction data to get it to balance with the bank payments. And only then will the linear program do a correct matching.
Aaron:Yeah. And it's super challenging problem. And I suppose this is this is the other thing about start ups and looking for problems to solve. So I might just try and fill everyone in on on where we started our conversation. We were chatting about what you wanted to do and what what was some of the reason you were doing the consultancy.
Aaron:Am I right in saying you were looking for the biggest opportunity of unsolved problems? Because ideally, you'd love to have your own start up one day.
Bethany:Yeah. More or less. I think I've I've realized, like, having done a start up now, I'm like, you know, the the ideal way of doing a start up is to make the customer the product manager. And everybody says they make their customer the product manager. And I'm like, I'm gonna actually do it.
Bethany:I'm gonna go be the customer for, like, you know, until I find, yeah, some some big unsolved problem. And at the moment, I'm like, reconciliation is like an extremely painful problem in finance. And there's, like, you know, the company where where, like, my client, they have about 40 people in ops, and all they do is reconcile stuff.
Aaron:It's a lot of costs to do this. What should be a simple job. Right? For
Bethany:something that can be in principle automated. But it's the the challenge is when you speak to people about, like, what are you reconciling? They're like, you wouldn't understand. It's all in my head. You can't automate this.
Aaron:And, yeah, this this is possibly, so, this is possibly the reality. Like, some problems are not solvable with the current constraints. Like, you might need and I saw some of the answers to it. You might need some extra transaction data. You might need a reference.
Aaron:You might need something extra supplied to do a a decent match. Well, of course. But you can't necessarily change the supply chain of information to get that information. So it might remain an unsolved problem until until that that happens.
Bethany:Until then. Yeah. And then you yeah. Exactly. So then you just solve it, like, in the brute force way that I'm doing it by trying to build some math to Yeah.
Bethany:Like, create the relationships after the fact. Because, of course, the proper solution is that at the time that you get the payment, you there's a process where you allocate Yeah. Bank payment to the things that it paid for. But that requires you to completely overhaul all of your transactional systems, and no one's gonna do that.
Aaron:Yeah. And so, of course, I was in the banking technology software space for a long time. There are companies trying to do that. You know, trying to actually change the supply chain. And even counting products like Xero are trying to do the same thing so that you supply payments along with with more information and proper references.
Aaron:Yeah. It's a tough problem because you're trying to change everybody. Like, you're trying to change what everybody does, and, of course, that takes a long, long time.
Bethany:And it's also impossible to do in a startup. This is a this was one of my learnings from, like, startup world is you need to solve a problem for, like, a person. Yeah. If you're solving an organizational problem, you can't do that in a startup because you don't have the credibility to survive the enterprise sales cycle.
Aaron:Absolutely.
Bethany:And, like, as soon as your sales cycle requires you to align a buying committee of, like, 30 people, you're you will die.
Aaron:You might not survive it. So that that was actually one of my, early choices, deliberate choices, is that although I had a lot of connections in financial software of, you know, those, you know, financial companies and companies I could sell software to, there was no point because I wouldn't be able to survive the buying cycle. The sales cycle would take too long. And they also you wouldn't have enough credibility as an early stage startup to even get through the sales process. So I had a couple of, every I think every startup has, you know, in Fintech at least, has HSBC, NatWest, or some, like, you know, amazing brand, as one of their early leads, and and I did as well.
Aaron:And it was like I very quickly was like, this is going nowhere. There there is absolutely zero chance this is gonna turn into anything for them or me. And they try to use innovation departments, and they try to make this sort of possible to get access to innovations, but they can't. They they have to sort of kill them off themselves sort of by their own nature of, you know, being risk averse and, you know, startups have to get on with it sooner as as well. Yep.
Aaron:Okay. So we've been chatting about startups and Fintech and the art of the possible for a while. Maybe we should go back to data and then and how this applies to the world of data. So the problems that we see in data and whether they can be solved or not, how do you think that helps people? Or what would what would you say to people inside an organisation who are going through exactly the same challenge?
Bethany:What if I knew that, I would have been able to sell my last product.
Aaron:They haven't got the answer yet necessarily.
Bethany:Yeah. I I think the thing the thing that's difficult is, like, data is super broad. It underpins every business process in the world. Like, data is just a digital twin of the process is being executed in the in the real world. And I think this is the challenge of, like, marrying up solutions to problems is, like, people don't think about their problem as a data problem, and so they don't look for data solutions.
Bethany:They don't even look to hire data people. So so the it's been really interesting for me to see, like, who is getting hired on the inside of this fund, And like, how do those skills relate to the problems? I'm like, this is a math problem. This isn't a
Aaron:this
Bethany:isn't a finance problem. You know, you should have hired a mathematician. Yeah, even though the domain is finance. So so, yeah, I think it's just it's it's really hard because data is so horizontal and broadly applicable, and business people don't think about their problems as data problems. Yeah.
Aaron:Yeah. I did. What what I what I like about doing this, so we're doing a recording, sort of a riffing about what what challenges we come up again. I have absolutely been there where I'm like, what does data even do? Like, you just like, you know, because you're like, okay, I understand, obviously, what it is.
Aaron:I understand what the technology is. But what does it do? Like, and you're like, what you how do you then describe to people if you're sitting in a data team inside a company, what what you offer to them? Like, what what can you actually do for them? And it's it's definitely a tough moment where you're like, okay, I have to figure out actually what the business is about.
Aaron:I have to figure out what matters. It's not really about working on the data at all. It's actually, like, trying to figure out what matters and trying to move the needle on something. And, you know, that that sort of yeah. Basically, imagine a startup.
Aaron:You're like, oh, my god. What do I actually do in the world?
Bethany:Yeah. I think all the projects I've worked on, in my current role have been about materializing relationships that exist in the world in the data. Because the the data is like an impoverished representation of reality.
Aaron:And
Bethany:so it's like, how do we enrich the data so that it actually represents the reality of what's happening in the business. And it's because the systems that capture the data don't capture all of the information. And so you have to derive the information from what is there.
Aaron:Exactly. And the way I've, you know, I've noticed that I started out with this idea of it was about how we convey information, how we share information, kind of in our journey, we went away from that to solving a much more practical problem that currently, that we need to move data as part of how we access it. You know, that's kind of the current state of the art. So we've gone kind of to more that data platform requirement, how do we help people actually solve those challenges. But I do eventually get back to the point where it is about assisting a conversation.
Aaron:It is about supplying something that they couldn't necessarily see, prove. And and, you know, that that is the, you know, the user's role in the organization is to assist the conversation, not necessarily prove or or be the sole source of truth. Or, you know, they they want to have one as good as possible representation so they can make a more informed decision. And that's kind of where I get to, like, back to it is about communicating the data, not not viewing the not
Bethany:forcing. The data.
Aaron:Yeah. Yeah. Exactly.
Bethany:And that's one of the things that can be so soul destroying about data is, like, 99% of the work goes into the plumbing. And, you know, I've seen so many people just like the last 1%. They kind of just throw it on like without that much thought because they're like, Yeah, it's only 1%. That's the only thing the business sees. Exactly.
Bethany:All their effort in the last 1%.
Aaron:Yeah. And so I I'm, you know, without, you know, blowing our own trumpet, I'm actually satisfied that solving that problem, that plumbing problem is a problem worth solving. Because the value is then able to be spent on the 1%. The other thing I thought was kind of necessary and, you know, you know, beneficial to to companies is that there's a whole load of data innovations that are inaccessible. So the the plumbing and the kind of 99% problem, there's a lot of tools and a lot of things that you would like to have, but you just can't have because there's a whole lot of effort to get them implemented.
Aaron:So
Bethany:Yeah.
Aaron:I think the kind of next, phase, let's say we solve the plumbing. Let's say we solve the infrastructure. Yeah. We mature as an industry. Well, actually getting access to better tools, better way of doing analysis should be easier than it is.
Aaron:So that that's kind of also part of the 99% problem to me at least. Yeah.
Bethany:I I, I I completely endorse this, like, solving the plumbing problems. I'm I'm highly skeptical of startups that are like, we're automating insights. And I'm like, how? You can't like, most companies today can't report on their sales. And you're gonna automate the decision making process on the reports that don't exist.
Bethany:Good luck.
Aaron:And you know what? Until you get into it, that's a that's a very hard thing for someone on the outside to understand. So may maybe to give a flavor to anyone who's watching who's not in the data space or hasn't experienced it, like, the idea of revenue and, you know, income should be quite straightforward. But it's absolutely not. Like, when is it income?
Aaron:You know, when has the work been done? When has it been invoiced? They're all really important things that need to be understood and then, you know, reported through finance. So it's not just like one little number. And It's not something that's
Bethany:not some of a column.
Aaron:Yeah. Yeah.
Bethany:And and they're all coming from different sources as well. Because your your invoicing system is different to your, like, when was, you know, if you're a services company, it's not revenue until it's been delivered. And like, that's in a different system.
Aaron:Exactly.
Bethany:And you need to bring all of this together. And then there's a 1,000 timestamps, and you need to pick the right ones.
Aaron:Yeah,
Bethany:it's complicated. It's like getting basic company level metrics is an insanely complicated problem.
Aaron:Which, you know, if I look at companies and their sort of maturity, I think, actually, you can measure them pretty accurately on whether they have a data team or not. So it's almost like to me whether they have realized they have a problem yet or not. So if you're really tiny, and you've got, you know, a pretty simple business, let's say you've got your, ice cream van, you know, you have customers, you have suppliers, you have, you know, you take some money, you've got most parts of it. In fact, some people said it's all parts of a business in in one little van. But you don't have a big enough problem that you can't just see what's going on.
Aaron:So as companies maturing get big enough, like, I think a a sort of leading indicator is they've got a data team. They've got enough of a problem. They can't tell what's going on just by looking at. So it's, you know, and as your data teams get bigger, you sort of bigger companies. Therefore, you have bigger problems.
Bethany:Bigger problems. Yeah. Yeah. It's it's and it's hard to know as well. Like, what should you have your data team deliver versus what should your business deliver?
Bethany:Because the data team might do the corporate level metrics, but then that doesn't reconcile with the department level metrics that each department is building on their own.
Aaron:Yeah.
Bethany:So that's a really complicated problem. I see
Aaron:a little trend at the moment. I wonder whether you you see this as well, that a company is attending to split data supply and data analytics. So the way I see it happening more and more is that analytics is closer to the business. And that makes total sense, because they understand the business. And they're not necessarily fully fledged data engineering professionals doing that, that they're more analogous, they're more mathematicians or statisticians.
Aaron:And the other supply is handled by more IT type developer type solving infrastructure is the bulk of their work. And they just have to make sure that it's good supply that it's good data. Do do you think that's more broad than just the co customers I'm speaking to? Is that is that gonna become, like, the norm in the in the future?
Bethany:So I read this blog post from Tristan Handy, the founder of DBT yesterday that really resonated. And it it was I think the title was something like analytics is not an assembly line. And so when you try to divide it up like an assembly line across multiple people, you get really bad outcomes because they, you know, the supply side and the demand side don't meet.
Aaron:Yeah.
Bethany:It's very it's very rare for them to, like, actually meet in the middle. And part of it is because analytics is such an iterative process of, like, you get some information, you may have an insight that leads to another question, then you need to add in new information. And so you're constantly having to go back across the assembly line, if you will, from like, insight back to like, now let's get a new source system and integrate that. And that's really hard to do across people. So the thing I've been trying to do is be the entire assembly line for my clients, like I do everything from the fetching of data from the source system, like finding it in the hell that is the 10,000 tables of the accounting system all the way through to like, here's the insight.
Bethany:I can't imagine splitting that process across multiple people. It's, like, unfathomable.
Aaron:It is it is definitely an operational over can be a bit of a nightmare. So the so 2 two things there. So it made me laugh, actually, because there's been a few posts since Tristan's post about data is definitely an assembly line. So it always, like, the counterargument, like, it this is Yeah. Yeah.
Aaron:The case when he and he's saying it's not. But I wonder if, have you read a book called The Phoenix Project?
Bethany:I have not. No.
Aaron:So it is talking about a DevOps pipeline. And if I grossly summarize what it's about, your processes of change, rather than get rid of them. So you're getting rid of it by trying to be the complete end to end, which is definitely a solution because you can just go and do both parts of it. Mhmm. Solution that is working in IT infrastructure in general is to make changes flow through much, much faster.
Aaron:So if you want an extra column or you need an extra data point, well, you need to get it added and captured and out of that source system much, much, much faster than currently happens. Because, let's say, order, processing doesn't collect the date of delivery, but you really need it to make an important decision. Well, that takes forever because now you have to add it to the system. Now you have to input it. Now you have to have it.
Aaron:What about all the old orders? And then it has to flow all the way through. So that's taking months, maybe years, right, to get this one extra piece of information. So the solution is to compress that down to, like, you know, let's add a field and then let's have it through. And and, yeah, if we can get that happening faster, that's a sort of solution instead of just trying to, like, go and ask the customers, did they get their orders?
Aaron:You know, you can't really can't do everything.
Bethany:Mhmm. Mhmm. Just reminded me of a thing that I'm working on right now, which is it's very similar to like, add the customer delivery date. What I'm trying to do is add the bank transaction date to the payment items. And it's like, yeah, we could do it through like a system and whatever, or like, we could just hack together this matching algorithm that then ports the date from the bank transactions into the payments.
Bethany:There's the pragmatic side of things as well. Like it's nice to be like, oh, in the ideal world, you would have this as an input field in your system. And it's like, yeah, that's a, it's an expensive it project and a huge change management, like transformation like, yeah, that's kind of the way I always see data is like shadow it in a way.
Aaron:Yeah, I definitely agree with that.
Bethany:There's so much you can fix with broken IT systems in a kind of, you know, not an ideal way, but like in a good enough way by just hacking together data behind the scenes from multiple systems. So I actually was like speaking to the head of Treasury yesterday, and I was like, today, your shadow IT building all these access databases, like, let us, the central data team, be your shadow IT department. So you you be the business.
Aaron:You'll be
Bethany:the shadow IT team, and then the IT team can be the IT team.
Aaron:Yeah. That that resonates with me so much. Yep. I think the one one place my mind goes to when you work on transaction, record keeping systems, you often completely forget about the downstream reporting. And you're even a product manager of a data processing system.
Aaron:You think your job is to move an item through its lifecycle. But your job doesn't end there. It has to be the reporting output as well. So you kind of shadow IT nature is that the poor old data team is coming along afterwards and trying to collect together what happened and the information they need. And, you know, that that's just repeated over and over because people don't think about the ad.
Bethany:They don't think about it at all. Yeah. And and this happens, like, when I was in Hotel Tech, we built a property management system, and I worked on the reports. And it was always, like, just fighting fires because feature teams would wanna do stuff like enable you to uncancel a reservation. And I'm like, oh, my God, don't implement that.
Bethany:You will break my pickup report. It will be nonsensical.
Aaron:Yeah.
Bethany:And you have to, like, anticipate every possible eventuality. It's, yeah, it's bonkers.
Aaron:Yeah. Well, I think that is an awesome point to leave it on. Data is an unsolved problem. We're still absolutely loads of problems to solve. And, yeah, a great space if anyone wants to have a career for Thanks, Bethany.
Aaron:I've absolutely loved our conversation. As a parting note, I wonder if there's one thing, one piece of advice you'd give to data people to make a difference in the world?
Bethany:Yeah. Definitely. So I would say try to broaden your skill set. I think before there was, you know, cloud engineers and data engineers and analytics engineers and analysts and then business users. And, you know, everybody did their little piece in their little silo of this supply chain.
Bethany:I would I would recommend trying to be a general data professional where you can kind of do the jobs of, like, all of those different roles. I think AI is at a point where you can radically expand your skill set. So if you've always worked on building dashboards, you can now build ETL pipelines, and and vice versa. I think you can have a much bigger impact on your client when you own the end to end kind of life cycle of data as opposed to just a piece in the supply chain. Yeah.
Aaron:And would you extend that into the business, like, even beyond, like, what people think of as technology?
Bethany:Into the business. That's a it's a good question. I mean, I I don't know how well that would go down if I was like, guys, I'm taking over treasury.
Aaron:I'd like to see that.
Bethany:I don't think that would be good. There would be, like, bankruptcy the next day. So
Aaron:Okay. That's
Bethany:not something I'm it's not advice I would take for myself.
Aaron:So you definitely need professionals. Okay. Let me summarize it a different way. In in software, need professionals. Okay.
Aaron:Let me summarize in a different way. In in software development, they often call them, like, full stack developers. It sounds like you're saying full stack data Stack data person. Yeah. Exactly.
Aaron:Yeah. Cool. Right. Well, thanks very much for coming on the show. It's been an absolute pleasure.
Bethany:Thank you so much for having me, Erin.