Value Driven Data Science: Boost your impact. Earn what you’re worth. Rewrite your career algorithm.

[00:00:00] Dr Genevieve Hayes: Hello and welcome to Value Driven Data Science, the podcast that helps data scientists transform their technical expertise into tangible business value, career autonomy, and financial reward. I'm Dr. Genevieve Hayes, and today I'm joined by Dr. Peter Prevos. Peter is a water engineer and manages the data science function at a water utility in Regional Victoria. He runs leading courses in data science for water professionals, holds an MBA and a PhD in business, and is the author of numerous books about data science and magic. In this episode, you'll discover how data scientists are adopting marketing principles to position themselves as value creators within their organizations.
[00:00:51] So get ready to boost your impact, earn what you're worth, and rewrite your career algorithm. Peter, welcome to the show.
[00:00:59] Dr Peter Prevos: Thanks Genevieve. Good to see you again.
[00:01:02] Dr Genevieve Hayes: Data science is a unique profession. In that it's the only profession I know of where practitioners are effectively required to market their services within their own organization to be able to do their job.
[00:01:16] Nobody ever questions why the accounting or legal teams within a business exists. The value of these well-established functions is already well understood. However, due to its newness, many businesses have yet to experience the value of data science firsthand. So it's left to the data scientists themselves to demonstrate its potential if they wanna succeed in their jobs.
[00:01:41] This creates a fundamental problem for data scientists. Most data scientists were never trained as marketers. They were trained as technicians and asking them to be internal marketers in addition to building models and analyzing data is setting them up to fail, which is why I'm excited to have you back as a guest.
[00:02:02] Peter, I. With a PhD in this topic, you probably understand the principles of marketing better than most data scientists throughout the world, and you've taken those principles that you learned in your PhD research. And applied them to marketing your own data science team services internally with great success.
[00:02:26] To put all of this in context, can you tell us a bit about what the situation was like for your data science team before you started actively marketing the team's services?
[00:02:39] Dr Peter Prevos: Great question. I'll give a bit of a context on how I got into this idea of marketing, 'cause I am a water engineer and sort of rolled into data science because analyzing data is what engineers do, right? When I did my MBA, my very first subject was marketing and the lecturer, I. The late Professor Walker asked us, what's your definition of marketing?
[00:02:56] And my facetious answer was selling people things they don't need. 'cause I never thought that marketing is something that is actually relevant to what I do and Rhet Walker said, I'll change your mind. You change your mind. And I did a PhD about the relationship between employee behavior and customer perception and also lecture that topic at the Latrobe MBA.
[00:03:16] And I started implementing this in my engineering job at the time and. Created some success, but were smoothed out a lot of that relationships with external parties. Then I sort of rolled into this data science gig, and yes, now I've managed the data and insights team at my work.
[00:03:30] And the issue that we had is that we've had a lot of reports, a lot of really what I believe to be great insights that we deliver to the business. And the people who use those reports were very happy with them and they used them, there were two problems. One was that some reports, people look at 'em and they go, oh, that's interesting.
[00:03:49] And then they end up in the proverbial draw. So they don't get acted on. And the other part was that although we were really working very closely with the people who do the analysis, who push the button, switch on the pumps, et cetera.
[00:04:03] But. Upper management didn't really understand what the value was we were providing as a team. And it only talked upon me like two years ago, like, Hey this marketing also applies that what we do. And started thinking through that. Stumbled on the concept insights as a service.
[00:04:19] So now applying the theory of services marketing to what my team does, and I developed a conceptual model that can be used by people in my same situation. Either teams that do it internally, so that's business to employee, but also in a business to business context.
[00:04:36] And say if you're a consultant offering services, those principles all apply to that.
[00:04:42] Dr Genevieve Hayes: Was there a single tipping point event that made you think, yeah, this just isn't working? Or was it just the general vibe over time?
[00:04:52] Dr Peter Prevos: Well it was basically conversations with my manager or a new manager and tried to explain to her really what it is that we do. But then trying to explain it from a data science perspective doesn't really work. Right. Talking about models and data governance. 'cause that's all.
[00:05:07] The invisible stuff. And as a data team within award utility, we are like award utility within the utility. No, you, in the morning you open a tap, water comes out. You don't care about how it works. You just want to enjoy the benefits of that water. You want to have a green garden, you want to be clean, et cetera.
[00:05:24] Now, I believe that for a data team, the same principle applies that what we do is a lot of invisible stuff happening, just like water till we have data pipelines and data lakes. There's a lot of similarities there, but all that plumbing is invisible to that user who might just see their Power BI report or they get a weekly email.
[00:05:42] And then , well, how do you communicate that value? If only sees that tiny little slice on the top.
[00:05:47] Dr Genevieve Hayes: Yeah, I can understand. 'cause I know when I turn on the tap, I don't. Think about the people who go into getting that water to me, and if I look at data science from that perspective, yeah. The data scientists are just invisible people behind the scene who only get noticed when things go wrong.
[00:06:04] Dr Peter Prevos: I think the important part of that is swallow a little bit of our own pride and think that actually this is a good thing, that if people don't realize, if it's so effortless to get the insights, then we're doing a really good job. But then there's a paradox in that, that how do you communicate that value?
[00:06:21] Dr Genevieve Hayes: Yeah. Going back to when I first started doing data science. My boss told me that we needed to go out to the organization and market what the team could do for pretty much the same reasons that you just said. Unfortunately, at that time, none of my team, including myself, had any marketing training whatsoever, so we didn't do the best job of it.
[00:06:46] We set up an information booth in the lunchroom and had a day where people could come and ask us questions and we had flyers and all that, it worked in the short term, but it didn't work in the long term. So I think marketing is important, but you have to do it the right way.
[00:07:04] And clearly if you've actually done a PhD in it, you're gonna do it a lot better than someone who's only done. Data science training. So you mentioned that you had a approach that you'd developed in your PhD for marketing your services that you applied to your own teams internal marketing.
[00:07:26] What was that system?
[00:07:28] Dr Peter Prevos: Sure. And what you just told me is a perfect example of, I guess a slight misunderstanding of what the craft and the science of marketing is about. Often when we say marketing, we think advertising, promotion but there's more, wait, there's more. There's only one of the seven aspects that I'm going to, uh. When we try to find the definitions of marketing and , there are hundreds of definitions of marketing, just like all these other sort of social things. But for me, the core concept of it is it the process of creating value and promotion is a part of that. Sure. But it's also the whole, process.
[00:08:03] And I'll go through the other elements, but a really interesting definition I found in the paper that was written a long time ago. And for me that was the aha moment. Oh, I forgot the name of the author, but he wrote marketing is Customer Satisfaction Engineering. And being an engineer, I thought, ah, I get it now.
[00:08:19] And what is engineering? Engineering is problem solving. If I do engineering and I just promote, are also I am, that's not going to stick. So for those of you who have study, mark, you might remember that there is this traditional thing called the marketing mix.
[00:08:33] The four P's. It's not so much a model of reality, it's a heuristic device because why would anything in reality start with four P's? But it helps you think through things. And this is for traditional products. So you have the product and from a data science perspective, that is the data product that we deliver.
[00:08:50] The end product, we have a price. It's not just money, price. Promotion, of course, we need to keep that in mind. Then there's a component called place, but it'd have to start with a P, but it's really about distribution. So this is how does the data product get to the end user?
[00:09:06] These are all the data pipelines and the plumbing. This is for products, but we are data scientists provide a service and services are fascinating because they have very different characteristics than products. If I buy a widget, a car, or a laptop, I can see it, I can feel it, I can smell it, I can assess whether it's a good thing.
[00:09:27] Services are intangible. They're harder to assess for customers, and especially with data science because the thing we deliver requires such deep expertise. So how do our customers know that it is valuable? So data literacy comes into play here. The problem of black box machine learning models.
[00:09:47] How do we deal with that? The problem with the service delivery also, it's heterogeneous. So what that means that the quality's not always the same because we're dealing with people here, and the problems are complex. If I have a cookie factory, every cookie's gonna be of the same quality.
[00:10:02] So when you manage a service, you need to. Think that through, what is your process of making sure that you always deliver the same quality? Then try to understand why do people want to service? Well, they do that because a lack of capacity. They don't have the time. I hire a cleaner because I don't have the time to clean or can't be bought, but it buys me free time.
[00:10:21] And that's the same reason why I was an engineer and I used to do my own data science, but because things are getting more complex. We now have teams to provide that as a service to the organization. I. The other part is capability, again, leading to the data literacy, which will be a bit of a running theme of what I'm talking about today and what it's about in services.
[00:10:39] It's a model called service dominant logic, which is a fascinating philosophical concept of what marketing is about, and it's about value co-creation because never ever has it been the case that a data scientist just gave us away and then said, Presto, here's a product. Right? You, you do that in collaboration with your end user.
[00:10:56] 'cause they have the knowledge.
[00:10:58] Dr Genevieve Hayes: I'll interrupt. I'd say it's frequently the case that the data scientist bes away in the back room. Unfortunately, that doesn't tend to end up with good results.
[00:11:07] Dr Peter Prevos: Exactly, and this is why we need to start understanding what we do as a service from a services model, that co-creation is so crucially important. Then for services, there's two more things that are very important.
[00:11:18] the process of service provision. But if you purchase a product, that's it. You purchase a product, you go to the shop, that's done. But in data science, it's very important given that co-creation, that you have this process of service delivery, that the process has to be efficient, obviously, but it also has to give the client, confidence that we're doing the right thing. That the end product will be what they need. And the last part I'd like to mention is the people aspect. So we have six Ps now. Data scientists are people. Believe it or not. But what that means is that we need to communicate or the end user needs to have trust that the people they're working with have this capability.
[00:11:58] Dr Genevieve Hayes: I read something on the internet the other day where it was comparing products with services, and it was saying. If you're buying a product, the person who's selling you that product, doesn't matter because when you get that product home, you never have to deal with them ever again.
[00:12:15] So people aren't really that important with products. But when you're dealing with services, the person who's selling you the service is usually the person who is also going to be delivering it. So if that person is someone who you don't trust or who has a bad personality or something, then you are not gonna buy that service because you are effectively buying that person at the same time.
[00:12:40] Dr Peter Prevos: That's exactly the point I'm making. That the person is so important within services, overall, but a data science as a service. Specifically. And let's say you are an external provider, the people that you are offering the service for, then that's part of your promotional package as well.
[00:12:55] Like, look, we wheel out this genius who's developed all these models. But he also has communication skills. They can also explain what they do for you.
[00:13:02] Dr Genevieve Hayes: so you've got the basic model of marketing services. How did you apply that to marketing your internal data science team?
[00:13:12] Dr Peter Prevos: So some examples of wow. Implemented this, in the products dimension. One of the problems we had is that we have so many products, 430 formal reports that tell us everything about our business at different levels of granularity, et cetera. They all have a purpose 'cause people have asked for them in the past, but we still have the problem that a heuristic problem, that people weren't able to find them or they didn't understand what actually it was doing.
[00:13:38] So what we decided to do is. To create a SharePoint list with all the reports that we have. Part of our specific issue was that we have both Power BI and reporting services. So for the average user, I don't think they should have to think about, oh, do I need to go to reporting services or do I need to go to Power bi?
[00:13:55] So we thought we need a central register, and then we categorize the issue reports. So I can now go to that register and say, give me all the reports that are about water quality. Here's a list so it makes it much easier now for people to find what it is that they need.
[00:14:10] Dr Genevieve Hayes: So that's place in your model.
[00:14:12] Dr Peter Prevos: It could be place Yeah.
[00:14:13] These things overlap a little bit. Yes. For the place component, we are moving from SSRS to being almost exclusively Power bi. And the reason for that is one thing I haven't mentioned when we talk about the price component, yes. Money is of course, part of that. Not much needs to be said about it, but there's also a time price and a psychological price that a consumer of a service pays.
[00:14:36] So the time price relates to the convenience of the product. How easy is it to drive this report that you're right. How many clicks before I get my insight and the psychological cost is, gee, these graphs are so uh, I have no idea what this all means. So the end user then has to expand.
[00:14:52] Intellectual energy to really understand what's going on, and we want to minimize those costs. And I believe that a tool like Power BI helps us do that. 'cause it's a little bit more intuitive to drive for people.
[00:15:04] Dr Genevieve Hayes: Sounds like from listening to that example, you've taken something, which pretty much every organization of above a certain size would be producing, so Power BI Reports. But rather than just having your data scientists go off and creating these reports in a vacuum based on what they think the end users would want, you've.
[00:15:28] Being very much focused on maximizing the customer satisfaction with these reports from the get go.
[00:15:36] Dr Peter Prevos: Correct. And this relates to the process component where I started to formalize the process in which we develop end products. And starts with a needs workshop. And I love doing these because this is where I can go back to my engineering days and talk with my colleagues and say, okay,
[00:15:51] what are the problems that you want to solve? And I never go to these workshops saying, what graphs do you need? What I. Yeah, what analysis do you need now? What are the problems you wanna solve? What is the data that you want to have at your fingertips? Really, rather than going through all these different spreadsheets and case in point is we recently changed the way we operate a bunch of wastewater treatment plants.
[00:16:10] And previously all the information was scatter gone everywhere. People have their own little spreadsheets. So start wiping all that together. And now we've created and easy to use, graphically pleasing power BI report where everything is at their fingertips. And also management can see in like a summary page to say how overall things are going.
[00:16:28] But we set up this formal process. So there's a needs workshop that I'll write, a functional specification. So without going into technicalities, just writing a document and say, Hey, colleague or client, this is what you're going to get. Is that what you want? Sign that off please. So we have an agreement in place and then we develop it.
[00:16:45] I could say we do agile development, but it's an iterative development process where we get continuous feedback, have people have a play with what we are producing.
[00:16:52] Dr Genevieve Hayes: How often do you get feedback from your clients?
[00:16:55] Dr Peter Prevos: As soon as we have something to show, really, so I'll give you example for wastewater treatment plant report. There are various elements. For example, there's laboratory data. Now we do the laboratory data. We say, okay, what do you think of this? Then we might have data from the Scala system. We add that data, we do the analysis.
[00:17:11] Think of that and then because we're in the same organization if it's business to business, you probably need to be a bit more formal about it, but I'll just send a team's message with a hyperlink. Check this up, right?
[00:17:20] Dr Genevieve Hayes: It's a small enough office. You can just walk across the room.
[00:17:23] Dr Peter Prevos: Yes. Yeah.
[00:17:25] Dr Genevieve Hayes: So you've touched on product place process and price. How do promotion and people fit into all this?
[00:17:35] Dr Peter Prevos: So for promotion, one of the things we did a little while ago is in the induction process, when new staff come on board, we created a an induction package, which is a little video of me talking as a talking head, one of my colleagues doing a screencast explaining what data warehouses and really try to focus on what they can get out of it without going into too much of the technicalities.
[00:17:57] And that's been well received. And the report register as well was also part of promotion. So a few weeks ago I launched that in our all staff meeting, which is a face-to-face and it's like a hybrid meeting. And part of that I created an insights treasure hunt. Because we had 400 reports and what I really wanted to do is said, yes, I can tell people, Hey, iss the hyperlink to the list.
[00:18:17] They go, oh yeah, cool. There's a list of reports. So what I said is ask some questions and I asked catchy PT to convert 'em into pirate language for example how much water did Bendigo consume in February, 2023, whatever, with a little hint on where you might find that information. And that then motivated people to.
[00:18:37] Find that report, have a look at it, and filled it so , that was very well received. And I wanna make that actually part of the induction in the new process to have people actively actually play with the data and hopefully give us some feedback like, Hey, I couldn't find it so well, this doesn't make sense.
[00:18:53] Dr Genevieve Hayes: And with people. How did you build the trust in your people?
[00:18:57] Dr Peter Prevos: Two things sort of formal and informal. Our previous manager and I decided that we should all be Power BI certified because we learn Power BI with YouTube and um. ity or whatever and stack exchange, but going through a formal process is useful. I've done the certification.
[00:19:14] My team is now working towards it, and although I didn't learn a whole lot because sort of knew how to use the tool, it was good to study it in a systematic way. And also I think that certification is this external sign, just like we might say. I'm Dr. Prevos That's an external sign.
[00:19:33] That I have some competencies, so Power BI certification, and same for my data engineering team. They're working towards certification for MuleSoft. So that's a tool that I use. The informal part is that I'll make sure that the people that develop the reports are also in touch with what happens on the ground.
[00:19:50] And I've organized some data excursions. So we all went with the whole team, went out to a treating plant. We went out to one of the big reservoirs and we just had a yell with the guys about what do you do? What are your problems? And for one of those excursions, a massive project emerged, which my colleague has now presented all over Australia, which is a tool that we developed data collection app, and reporting to manage dam safety.
[00:20:13] And also another example is recently we developed an app which we call Hydro, which is power apps data collection tool. And Matthew who developed this, he actually did onsite user training at the wastewater plants. 'cause I wanted Matthew to actually smell the wastewater plant. You haven't worked for water utility if you haven't actually been to a wastewater.
[00:20:33] Dr Genevieve Hayes: Does it smell as bad as I'm imagining?
[00:20:35] Dr Peter Prevos: No, not at all. No. It's just a bit musky,
[00:20:37] Dr Genevieve Hayes: Oh, okay.
[00:20:37] Dr Peter Prevos: smell. Yeah. But it's the idea that this abstract app that Matthews created, that it has a real life. Outside the computer. These operators, they walk around with their mobile buffer phone and they need to do things and they need to collect that information.
[00:20:51] So that's the people part to also create those relationships with my team who, developers who would otherwise be sitting in a dark corner in the office somewhere. But I want to get 'em out there where the data is actually created.
[00:21:02] Dr Genevieve Hayes: And you said that this was all very well received by the organization.
[00:21:06] Dr Peter Prevos: So far so good. Yes. It's makes the team more visible. It helps the business understand who the team actually is and what they do. We're not just a bunch of it nos, we understand what their needs are. We are here to help them. This is my mantra. We are tool builders.
[00:21:20] We are here to build tools for you to make your life easier.
[00:21:23] Dr Genevieve Hayes: So if any of our listeners are contemplating replicating what you've done in their own organizations, what advice would you give them on where to begin?
[00:21:36] Dr Peter Prevos: The key to a marketing approach always is to understand the needs of your clients. And there's a big discussion about needs and wants and the common sense. Idea of needs and wants is very different to the marketing sense of needs and wants. So needs are not just things that you need to survive
[00:21:55] there are other needs. So you might talk to somebody in your organization and they have a need to solve a problem 'cause they spend so much time copying, pasting data from one spreadsheet to the other. It still happens in so many places. And then you can say, Hey, maybe I have a solution for you. So have conversations around the business and start thinking about problems.
[00:22:16] The biggest mistake we can make in technology is to think about solutions before we have the problems.
[00:22:21] Dr Genevieve Hayes: And what is the single most important change our listeners could make tomorrow to accelerate their data science impact and results?
[00:22:29] Dr Peter Prevos: Don't try to impress your clients or your colleagues with machine learning and ai. I have a little mantra. 99% of business problems don't need machine learning. There's often very simple things that we can do to make people's lives really easier. And then once you do those simple things, then also the data is available for more complex analysis.
[00:22:46] So I'm probably repeating myself, is really talk to your clients, to your customers, to your colleagues, whoever it's that you do your work for and understand their needs.
[00:22:55] Dr Genevieve Hayes: I've been in jobs where I've desperately wanted to do, your fancy. Learning models, but all the customers want is something that can be solved with Excel. And if you give them that Excel solution, they are thrilled. And if you were to do the big, fancy, deep learning solution, they probably wouldn't know what to do with it.
[00:23:14] Dr Peter Prevos: Part of the issue there is that these models are so complex that it requires the data science to understand them. And this is why I started teaching that. Data science to water professionals to get people to understand what machine learning is, and for us as an industry to be able to have meaningful conversations with the providers of these services.
[00:23:32] Because often I speak to companies who deliver those types of solutions, machine learning solutions, and they just say, oh, it's done with ai, it's done with machine learning. But take your client serious and take 'em on a journey and try to explain what the benefits are.
[00:23:46] Dr Genevieve Hayes: Yeah. So for listeners who wanna get in contact with you, Peter, what can they do?
[00:23:52] Dr Peter Prevos: I'm on LinkedIn, and at the moment I'm working on a book on this topic. So I'm still gathering my thoughts on what this model looks like. So I'm planning to really develop this model of insights as a service and what that means. And I would love to hear from people. With experiences, questions, or ideas.
[00:24:10] And maybe we can help each other and flesh this out to something that can become really useful.
[00:24:14] Dr Genevieve Hayes: When are you aiming for the book to be released?
[00:24:18] Dr Peter Prevos: It will be self-published or there's no real deadline. I'm hoping to haven't done this year. Yeah.
[00:24:23] Dr Genevieve Hayes: so hopefully by Christmas 2025.
[00:24:26] Dr Peter Prevos: The ideal Christmas present for every data scientist.
[00:24:28] Dr Genevieve Hayes: Yes, and there you have it. Another value packed episode to help turn your data skills into serious clout, cash, and career freedom. If you enjoyed this episode, why not make it a double next week? Catch Peter's value boost a five minute episode where he shares one powerful tip for getting real results real fast.
[00:24:53] Make sure you're subscribed so you don't miss it. Thanks for joining me today, Peter,
[00:24:58] Dr Peter Prevos: Actually always.
[00:25:00] Dr Genevieve Hayes: and for those in the audience, thanks for listening. I'm Dr. Genevieve Hayes, and this has been Value-Driven Data Science.