Erwin De Werd: In the race to AI, most companies are starting with flat tire, bad data. You're listening to the Data Edge. I'm Erwin de Werd from Pearstop In each episode, Stephanie Wiechers and I explore the intersection of data quality, enterprise standardization, and the real world value of information. Whether you're an asset manager or a bit specialist, we are here to give you a blueprint for a reliable data layer. that fuels smart decisions in the next generation technologies. This is the foundation for what's next. Welcome Stephanie.

Stephanie Wiechers: Thank you.

Erwin De Werd: Yeah, so in the introduction we talk about a couple of things. We talk about standardization, but also the blueprint for ⁓ next generation technology. And I would like to start with that because recently I see this interview and I think I agree with that. It becomes now more and more easy to build applications, so we will see more and more applications what can be very smart applications. However, then. It also become evident like how do we move data from one system to the other? So that that mean I think it will be more dynamically more and more to have data moving from one system into the other to, yeah, actually work on the data and then do the application. How do you say functionalities on top of that, right? What do you think?

Stephanie Wiechers: So, yeah, so with all these, well, Vibe coding opportunities, but also just working with Copilot and getting some information out of a data set, suddenly everyone has become a developer ⁓ or everyone has become a data scientist. that means basically without... any prior experience now that there's so much possible, which is really cool. It's one of the coolest developments that's been made. ⁓ with that comes an immense responsibility. because where before you'd have data scientists and when they would make their analysis, I mean, when I would make an analysis as a data scientist, what I would do is I would make my dashboards, my mockups, would also just like find my numbers. but you're super super critical and you know how to be critical on those numbers and you implement all the sense checks and then you know how and where to go check them and there's always a ton of mistakes made and then you go to the data that's behind it and then you see okay wait this project manager hasn't booked their hours this ⁓ this item is categorized like incorrectly. suddenly it seems as if, you know, one project's been very expensive or you spend like a million on fruit bowls, you know, like random things happen. Yeah.

Erwin De Werd: Yeah. Yeah, I completely agree with you. And I've seen this also several times that first of all, the challenge is of course to make these dashboards. But this now is disappearing because now it becomes very easy to make all these dashboards. But the next thing then is if you have the dashboards and you sit in this meeting, remember I had also, well, I spent a lot of time preparing a dashboard and I was so happy with this. And then I come in the meeting with salespeople and so, and I show the dashboard.

Stephanie Wiechers: Yeah.

Erwin De Werd: and they immediately find the one thing that they say, yeah, but this is not right. This number cannot be good. And then there is a well, downfall that then they say, okay, this is the whole dashboard is not good because there is mistakes in it. But my opinion is there are always these flaws, always.

Stephanie Wiechers: Yeah. Yeah.

Erwin De Werd: So that, the overall picture of the dashboard is still very powerful. However, it becomes more important then to get the data right because people will be more critical about it.

Stephanie Wiechers: Well, when there's flaws in the dashboard and in the numbers that it's showing, I actually do agree with your colleagues there. Because when you can't trust what's there, then it's really, hard to make decisions on it. So it's, it's essential that whatever shows in the dashboard is reliable. So that means either it needs to be perfectly correct or it needs to be highlighted where there are still some like possible errors, some possible data entry problems. Because you don't want to make decisions on like on a dashboard that's not reflecting the actual ground truth of the reality below it.

Erwin De Werd: Yeah, I'm with you. with you. So I normally look at the trends in these kinds of things. However, of course, if you focus on the actual more details in the data, then it must be reliable. So the data must be good quality. That's one thing.

Stephanie Wiechers: Yeah.

Erwin De Werd: And the other thing is, course, then if the data sits, one of the other challenges, for instance, ⁓ I also recognize that as an example, there's a lot of data sitting in finance systems. There's always a good source for data. ⁓ But then you won't use it not like for commercial reasons as an example. So you won't create this dashboard, but then how to get this data out of this final system, move it to your other system to work further or to combine it. So moving data from one system to the other, that becomes so more important to have more dynamics. So, and how do you see that? So what's more important to get that, let's say possible.

Stephanie Wiechers: Well. So most larger or more mature businesses now make use of some form of a data lake. So that could be ⁓ Microsoft Fabric. That's obviously a very popular choice because everyone is on the Microsoft cloud. Well, there's like a ton of them out there. Databricks. We previously also ran projects where the client was much smaller and there we would just link their ERP system to a BigQuery setup and so that would then allow them to still merge different data sources. So like sales data, POS data, any inventory to different parts of the system. It doesn't really matter. how you do it, it's more of an internal preference where there's lots of considerations, right? Like price is often a big one and what's our tech stack look like? Cause it does make sense to use Microsoft if you're already using Microsoft. How easy is this to use? How easy does it integrate with other AI tools? But in the end, the... The migration to these platforms or like using these platforms is often seen as a natural inflection point of taking some care around the data quality that's going in. And the reason why is essentially when you're going to use these data lakes, those will serve as a foundation to then build those dashboards upon. Yeah.

Erwin De Werd: Exactly, exactly. The quality of the data is also extremely important. And while the other thing what I noticed with several of the projects ⁓ running is standardization. And all industries have different standardization like in construction or in manufacturing or in healthcare, they all have their own standardization in there. And that helps of course to share data across different entities in that industry, So how do you see this and ⁓ how do you think this will develop?

Stephanie Wiechers: Well, when I link it to the AI developments, are, well, in my opinion, there are two ways to go about this. And I think standardization shouldn't be seen as the holy grail. There are a lot of advantages or there are a lot of things to say for it. So. ⁓ We of course work with the UNASBSC standard and we help companies also move into that standard without, you know, being some bureau in India that like does everything by hand with also the corresponding inaccuracies that come with that because there's no context of the business. Like having said that, we know that... In general, is expensive to move into these these type of schemes. ⁓ Well, especially when you haven't been doing it yet because there is like some learning curve. Everyone in the business also needs to adopt the new system. ⁓ Especially when you're a smaller organization, there's quite a lot to say for just having like a very straightforward, small and lean system that just works with, you know, what you are only processing in your business and what works for you. Whenever you get a bit larger though. then you can think about if you are buying what like 10,000 different items and you need to categorize those. Well, at some point you're also going to have to spend a lot of time designing a system which will then only be useful internally. So no one who works with you will have any knowledge of it before and it won't integrate into any type of standard or system. So that's like a trade-off to make I would say.

Erwin De Werd: Yeah, yeah, well, well, even though when you are a smaller company, you might want to work with or you can be a supplier, right? You can be a vendor. ⁓ You can be a system integrator, whatever. So you sit in a larger network and if these bigger organizations, they work with these standards, then yeah, you would like to adapt to that. So recently I heard about a company in the, let's say the manufacturing engineering and so. And they also said that for many of their clients, the standardization is extremely important because they need to respond and to work together with ⁓ companies that have these standards.

Stephanie Wiechers: Yeah, and then there's a bit of a challenge in translating what you already have into the new language.

Erwin De Werd: Yeah, so that is a challenge. Yeah, well, let's focus a little bit on that and that how you can do so. So because I also agree with you that if you're a smaller company, then you're not sitting waiting. Oh, let's do all this work to work in the standards. However, if you would like to serve certain clients and they have these standardization requirements, then you'll have to. So what are the latest developments to? bring you from a non-standard situation into a standardized situation.

Stephanie Wiechers: Well, all I can say is it's not the most easy exercise in the world. Sorry to bring the bad news there. No. ⁓ Well, the first step is decide like what standard are you working with, right? So yes, there's the UNSPSC. like, it's very...

Erwin De Werd: It's realistic. Mm-hmm.

Stephanie Wiechers: It's very broad, it's industry agnostic and it's really meant to serve as an international standard. But within different industries there is of course the FIRE standard. I can give you a ton of examples, but anyway. Different industries have different standards, if your industry has one, then it makes sense to focus on that. ⁓ It also depends on the goals and so sometimes there's a little bit too much strategy without having the final goal in mind. ⁓ And so then you can get stuck onto like a detail and then that becomes the highest priority. But in the end, there's a reason why you would want to move towards any other system, which often is cost saving, right? It's when you are well able to classify what you're actually doing that allows for a very, very straightforward way of identifying kind of gaps, spend leaks, but any type of inefficiency that is in the business. ⁓

Erwin De Werd: And can you give a reason why one of your clients and projects want to move into such a standard?

Stephanie Wiechers: Yeah. So that's very much around the cost saving exercise. And what happens there in procurement is obviously they do have a classification, right? They're like very well organized, very well structured. have different entities on which they buy, and then they have different cost categories in which everything they buy already goes. So there is a pretty solid baseline. The thing is that

Erwin De Werd: Mm.

Stephanie Wiechers: At the moment, the baseline is not granular enough to really do the spend analysis correctly. And in order to do a proper spend analysis, you'll need a very, very granular level of.

Erwin De Werd: Okay, yeah, so this is really tied into spent analysis exercises, yeah. Okay, great, yeah, so. Yeah, we discussed several things, right? So the dynamics of moving data from one system to the other that will only grow. And so that's an important thing. Doing so.

Stephanie Wiechers: Yeah. Yeah.

Erwin De Werd: Then you have concerns regarding the quality of the data, especially also because of the trend of building dashboards and giving insights that that's basically it's not about the dashboard, but it's more about the insights that people will have into the data. Then they will also find the flaws. Probably they were always there, but now all of a sudden you can see them more easily. And that gives challenges, of course, to. And also the requests to improve the data and the data quality. So I guess that will be a growing concern. And lastly, then what might help is standardization, especially when you work in a bigger environment where other parties are involved. So all these things are important. And yeah, I think we should dive maybe deeper in some next episodes in these different. elements there and for so now for now I want thank you. For the insights in these fields is always very valuable and yeah, I looking forward to what you can tell us and how you can bring us more in the world of data the next time.

Stephanie Wiechers: Thank you. See you next time.

Erwin De Werd: See you then, bye bye.