“From gut feeling to decision intelligence”—with Tobi Nehren (areto)

Recorded in German—English translation of the original conversation.

Intro

Maria-Lena Matysik: Hello and welcome to a new episode of Analytics Sprechstunde—yay, another episode!

I’ve fallen a bit behind with the podcast. I actually wanted to release an episode every two weeks, but at the moment it’s more like one episode a month :S There’s just so much going on at our end right now—the team keeps growing, we’ve improved a number of internal processes—and somehow I always plan more than I can realistically get done.

But … that’s not the topic of this episode, of course 😉

My guest today is Tobi Nehren. Tobi is managing director of areto, a business intelligence consultancy, where he is responsible for the strategy division. That means he works at the intersection of strategy, transformation, organization and data.

His career path is anything but the classic route into data—at least from my perspective 😀 Tobi originally studied political science and art history and spent many years working in digital communications and political campaigning. Among other things, he headed the newsroom at the SPD party executive, where he dealt with digital communications, content production, campaign strategy, larger budgets, reporting and data-based optimization.

Today, his main focus is the question of how companies can turn their data not just into reports and dashboards, but actually into better decisions. areto calls this process of data-based decision-making Decision Intelligence—so not just BI, business intelligence, but Decision Intelligence.

And that’s exactly what we talk about in this episode: What is needed for Decision Intelligence, in other words, for data-driven thinking and decision-making? Tobi also explains Dilts’s model of logical levels. It shows how integrating data into a team’s everyday thinking unfolds—namely as a typical change process.

And of course—as I set out to do for this podcast—we look at how companies can get started in very concrete terms: How do you bring business strategy, departments such as marketing, and technology together at one table? How do you identify the few decisions where better data would really make a big difference? And how can initial successful use cases help build positive momentum for the entire organization?

Okay, let’s go!

Interview

Maria-Lena Matysik: Hi Tobi, I’m really happy to welcome you to the podcast today. Thank you so much for your time.

Tobi Nehren: Hello! Thanks for having me, I’m really happy to be here too.

Maria-Lena Matysik: Then let’s jump right in. From your perspective, what is the state of decision-making—or Decision Intelligence, as you call it—in companies? What has your experience been?

Tobi Nehren: That’s a very broad question, of course. And we only ever see a certain slice: when everything is great, nobody looks for consulting. Consulting is requested when things need to change or aren’t the way people concretely or intuitively imagine they should be. And in the situations we encounter, things are often planned in different rooms that, in my view, should be discussed and planned in one room.

Executive management has certain requirements, ideas and goals. Department heads have a functional perspective, and that’s where expectations are formed. AI, of course, acts like a steroid on top of that, because everything moves even faster and runs even more at a kind of half-implementation level—think vibe coding. And on the other side are the technologists who then have to implement the whole thing, all the way down to deletion concepts, GDPR, interface integration, maintainability and so on.

These are often things that aren’t discussed and planned together. That frequently leads to frustration on both sides: management says, “We finally want to hear some solutions,” and the technologists think, “Well, look what they’ve come up with this time.”

That’s what we try to bring together. We create a space where this can be talked about and then also planned in a way that allows it to be implemented. And then it’s actually a lot of cultural work, too, which in transformation projects—as I’ve always experienced it in my career—is imagined, perceived and planned as easier than it really is. It simply takes a long time for people to grow into a new role and adopt new ways of working. And in our experience, data-driven work is a mindset that very strongly changes identity and ways of working. It simply takes time until it has become ingrained and has taken hold.

Maria-Lena Matysik: Can I sum it up like this, that you’d say: Working with data is shifted to the technical side too quickly, and not enough is done to clarify which questions we actually want to answer and what the whole thing is supposed to achieve—and then to connect that with the technology?

Tobi Nehren: Let me illustrate this with the classic purchasing process. The people who approach areto are typically technologists, and there’s a technical purchasing process: RFP, proposal and so on. We advise on architecture diagrams, transformation—that’s all priced in. But when we then say, “Now we also need to talk to the people”—I’m deliberately putting it bluntly here—the technologists sometimes say, “But that’s not what this project is about at all.” And then I have to justify why it is what the project is about. Sometimes I then have difficulties in purchasing processes, because transformation is actually HR. Then I’m dealing with different budgets and different buying centers and have to pitch myself all over again elsewhere as a consultant and, let’s say, as a salesperson.

That’s an expression of something I see very often: data projects are scoped and planned through a technical lens and not through an organizational lens. There are currently many studies on the success of AI projects—which are often data projects or the tip of the iceberg of data projects. According to them, success is 70 percent transformation and 30 percent technology and model work. And this 70 percent transformation is often not sufficiently priced in. You do see it in good RFPs; it happens. But very often we’re in projects where we have a clear technological roadmap and clear points of contact. Everyone more or less knows what needs to be done, and that’s also easy, because it can be planned by engineers on the drawing board.

Humans in this whole construct, however strangely they always behave, are harder to plan for, by contrast. It’s harder to scope and describe how the organizational change happens and what the path looks like. And that often gets neglected.

Maria-Lena Matysik: In your experience, why is that? What are the thought processes in organizations that lead to the topic of “making decisions with data” being driven not by the decision-makers but by the data people?

Tobi Nehren: I wouldn’t even say that it isn’t driven by the decision-makers. I think the same game keeps repeating itself. I experienced that firsthand in politics: it was “Social media, that’s your job now” or “Digital transformation, that’s his job now.” Then titles like CDO or CTO were handed out, and the topic was delegated to someone.

But I’m firmly convinced: if we really want to respond appropriately to data and AI and their transformative power on organizations and business models, that’s only possible if the know-how and the responsibility also reach the leadership level and are carried there—and if real interest and real competence are built there. Because this can’t be delegated entirely to someone you put the magic hat of “AI” or “data” on and tell, “Have a nice trip, go handle it.” It only works if it’s carried by the leadership level, by the C-level, and the change starts there too.

I often use Dilts’s model of logical levels. In it, data-driven work is a change that has to reach the identity of the company: “We now work with data, and we want to make decisions in a data-driven way.” Then it cascades through all the levels down to the dashboards—but those are the lowest level.

I just can’t put a hat on someone, say “Go handle it,” and then hope that somehow magical things happen in the organization. If I want a lasting change in how we work with data, then it’s about roles changing, it’s about maintaining data and about working on the system. And working on the system is always a management task, not the task of those who work in the system. You simply have to get involved.

RFPs like these, made by technology for technology, are an expression of that. That’s not a criticism of the technologists, either—it can only work if such things are thought through holistically. To do that, you first have to understand, and be willing to understand, how much working with data and data-driven work with AI really changes and will change organizations—and how great the pressure to act is. I’m firmly convinced of this: we will all get to work differently with data and with AI. And it will come down to which companies manage that and which don’t.

Maria-Lena Matysik: Mhm.

Tobi Nehren: And that also depends on what competence exists for it at the leadership level.

Maria-Lena Matysik: Could you perhaps put your finger once more on what exactly the problem is? What are the negative consequences for organizations that don’t manage to think about this holistically and from the leadership level down?

Tobi Nehren: That totally depends on a company’s business model. But I’m firmly convinced that data is the fuel for what’s coming now with artificial intelligence, that is, with intelligent automation. And this fuel has to be refined, produced and sourced within the organization. If an organization doesn’t manage that sufficiently and therefore can’t use this AI turbo boost, then that’s simply a competitive disadvantage. There will be companies that manage it. And there will be companies that reinvent business models from the ground up, enter the market and simply skip certain organizational steps.

The question will be which companies succeed—similar to the second wave of digitalization in the B2C market. There were companies that didn’t bother transferring their folding maps to digital because they said, “We have the best folding maps, and our folding maps fold better than any others.” Google Maps and Apple Maps and everything else that came along didn’t care about that. These providers were simply swept away because the new business model or the new service was so much better. And that will happen with AI in certain areas too—including in B2B.

Of course, I don’t know what the best solutions are. The people we advise, who know their business model and their customers, know that much better than we do. But from my point of view, it’s about understanding the principle: that a transformative force is coming that you have to understand in its dynamics and in its paradigmatic logics of change. Only if you really understand that, look at it and then examine where it hits your own business model do you have a chance to come up with something.

If I still say today, “We have the best scores for orchestras, they’re really great, and we’re adding another licensing model and another licensing model on top,” then that delays the development. But at some point—I have no idea about the market for scores—someone comes along and says, “We’ve solved this for iPads.” And suddenly there are iPads in front of the violins and strings everywhere in the orchestra and no more scores. Then things get tight pretty quickly.

Maria-Lena Matysik: I see. While you were talking, another example came to mind, on a much more down-to-earth level, from my day-to-day work. If it isn’t aligned and isn’t thought through holistically from the management level to the data people, things often get implemented that nobody ends up working with. Along the lines of: “Hey, why don’t you build a dashboard?” Then someone builds a dashboard, but in the end management doesn’t work with it—simply because, to connect it with what you said, it wasn’t thought through together. Management didn’t come and say, “We have these problems, we have these questions we want to answer. Now we need to work out together which data will help us—and then we’ll build the dashboard.”

Then, of course, an enormous amount of resources goes into projects that in the end went totally down the drain. And that creates frustration on all sides: on the management side, because it doesn’t help them make their decisions. And on our side too, when we build the dashboard, because our work went down the drain and nobody works with it. So it’s also a question of resource efficiency: Where do we as an organization deploy our resources to ultimately solve our problems?

Tobi Nehren: Absolutely. That’s exactly the step we take first every time we come into engagements that involve use cases. We say: We first need to go through a round in which we understand your business model and find out where the biggest pain point is right now. Because your management is currently investing tens of thousands of euros, possibly more if it’s a complex data platform, in building infrastructure. And that infrastructure then has to help achieve the business goals. If we don’t know where they lie, the probability that everything will drift apart is enormously higher.

When we talk about strategic consulting, my intention isn’t at all to show up with 15 people in light-blue shirts, brown belts and brown shoes, run an analysis through the entire organization and then build a 250-page PowerPoint presentation. That’s what you hear about how strategy consulting works. With us, it’s more that we stand hands-on at the flip chart with a marker and ask: What are the three main problems you’re dealing with where you don’t have your high beams on right now, because the data isn’t there or the information base isn’t clean? Where would you say: If we understood this and had better data at certain decision points, at decision gates, it would make us dramatically better?

Starting there obviously helps much more than with any random problems I can collect somewhere and where I might be able to help a highly valued clerk. I have the greatest impact with the questions senior management is asking itself—and then starting there and looking at how we can work in a data-driven way. And that’s a particular problem for mid-sized companies right now. In large corporations, reporting and dashboards and all these things have existed for a long time. Among larger mid-sized companies in Germany, that’s sometimes still not the case, because tool number one is still Excel.

Maria-Lena Matysik: I’m with you. Let’s move from the problems to the solutions now: What are your—or your team’s—strategies for tackling and solving these problems?

Tobi Nehren: I already hinted at it: We ask people to take this step first. Of course, we can’t force anyone. If someone says, “Here’s the RFP, we want to build, replace, transform or migrate a data platform from A to B,” then of course we do that. That’s totally legitimate, too. There are good people and good teams for that who know exactly what needs to be done and how to get from one platform to the other with all the technical requirements. That’s where we come from; we do that and support it. And if higher-level consulting is needed on top of that, there might be conversations about it.

The case we’re talking about now is when this shared space doesn’t exist yet. And we create it by bringing business strategy together with the data level and the technology level and making it possible to talk about the whole thing. We build an understanding of where the current business and strategic challenges lie and then analyze: How can we gain leverage on them? What data is there? What data might be needed? How do we get the data, and in what quality? Does that quality already exist in the company? And then we build a holistic path at the organizational and technical level: What needs to be done to gain leverage on certain things?

That doesn’t mean we have to change every job in the organization over the next five years because everything is one big transformation. Instead, we take a use-case-driven approach: For the one, two, three decisions that would mean a big change right now, how do we bring the data together so that these decisions can be made better, faster and more competently? How do we consolidate this data on a platform that also scales so that use cases 10, 15 and 30 can still run on it?

And then we start working, bring the relevant people along—the people whose work is changing—and tell a story about the transformation. A success story, too, where management can say, “We’ve gotten better here because this happened.” That creates momentum, a positive narrative—and out of a positive narrative, momentum that is, in quotation marks, fun, insofar as that’s a factor you can name in a company. Momentum develops where people come along because a pull is created.

My experience is: Where a pull develops, it’s five times better than pressure along the lines of “You have to do this now,” because then people get interested. If you then establish office hours—“You can come here and get advice on data, whatever it is”—then people want to participate in these new things. They want better insights themselves and have ideas for use cases. And then a holistic transformation can actually take place—if the torch of enlightenment is also carried by management, shining into certain corners and showing, “Look, the following is possible here, which wasn’t possible for us before.” That creates a positive story and a positive narrative.

Maria-Lena Matysik: Could you tell us a bit more about how, in your experience, data culture develops or is built—in the very broadest sense? You mentioned Dilts’s pyramid earlier.

Tobi Nehren: Yes. The first point is the realization and the will at the level I call creating purpose. Those who create purpose are always senior executives who say, “This is who we are, this is how we work, and this is what sets us apart.” And working with data has to become part of that.

I had a very enlightening meeting on this in 2014/15, when I was working at the SPD, with Jim Messina, who had managed Barack Obama’s reelection campaign—one of the pioneers of data-driven work. He said, “There’s no longer any decision about resource allocation here that isn’t made based on data. If you come to me and say we want to hold the following event, introduce the following platform, launch the following initiative—why should we do that? Bring me the three to five data points that prove this is a good decision measured against our goals.”

This change at the top leadership level naturally cascades down. Because by day five or six, or by meeting seven at the latest, it’s clear: If I go into this room and want something, I have to be able to justify it with data. Then I have to figure out where I get that data. Because decisions there aren’t made in one direction or another on principle, but because someone looks at the data and the defined KPIs every morning and says: We’re going in this direction—or this one, or this one.

And resource allocation is the same thing in a company. I always talk about it: Where do I put labor, money and time? When these decisions in management are made based on data and no longer based on gut feeling or unwritten house rules—and many companies did very well with that for a long time, it’s just that the world has also become more complex—so when data is drawn on and KPIs are defined, then that carries through to the lower levels of Dilts’s model, into values and capabilities. Then people develop behaviors because they suddenly know how to read dashboards, how to pull data together, how to build it and where to get it.

And at the very bottom there actually are dashboards, flip charts with certain curves—things you can see in the room. But these are the results of cultural changes that are invisible at first, because people at the leadership level and in decision-making bodies think and talk differently.

It doesn’t start at the level of the environment and the dashboards. The best dashboard in the world doesn’t lead to better decisions if nobody looks at it and nobody asks intellectually: What are we trying to decide here, and what actually are good key performance indicators, KPIs, by which we decide whether to do one thing or the other?

Maria-Lena Matysik: To sum that up very briefly: Leadership, meaning the people who make decisions, has to demand data—otherwise nothing at all happens with data in the organization.

Tobi Nehren: Exactly. And of course the executives don’t have to be able to do it all on their own. It’s about actively asking the question: “We understand that data changes things—who can I talk to about what’s changing?”

Our role also isn’t meant to be that of a whisperer standing next to the organization and whispering some secrets into the managing director’s ear. Rather, it’s about creating a space where you consider: What does data change in our business model, in our company? What does AI change? And how do we bring that into the organization in a lasting way? And then transparency is built.

If management understands that, questions itself and goes in with an inquiring attitude—also in a safe space where you can ask questions you might not ask in front of employees or in front of the level you’re actually supposed to lead—then from my perspective that’s the first step toward achieving a holistic transformation in the organization. Holistic means actually working with data. And working with data is then the first step toward getting the work with AI right as well.

Maria-Lena Matysik: I’d like to go straight to my last question, because you just touched on it. Could you sum up once more what, from your perspective, the first step should be for an organization, a team or a department that realizes: We’re not yet using data the way we actually want to? What would be the very first, most important step?

Tobi Nehren: Asking yourself the question: Where do we actually want to go?

If I can’t answer that consistently—if, as a leader, I can’t provide sufficient direction about where I want to lead people … That’s the principle of leadership, after all: I have to be able to say which way we’re going. If I can’t answer that sufficiently, then it makes sense to bring in external consulting and answer this question consistently—also in light of the question: What does data do to our business model or to our work?

We like to work with scenarios to simply sketch out how this might develop—including with several scenarios—and then build an idea from that of where things could go. This idea is then always the latest best guess. So it can always shift. It’s not as if it’s set in stone and you say: This is how it’s going to be now. Times are simply far too fast-moving for that. But asking this question—something has to change here; where do we want to change to; and what does that mean for our work, our business model and our department?—and, derived from that, saying: We have a target vision here that we may also need to refine, maybe every six months. But we can work with this target vision and orient people toward it. And then we consider what we need technically and organizationally to align ourselves with it. That, I believe, is the first step toward addressing the issue of competence.

The second step is willingness. And with AI and in today’s world, that’s more important than ever, because simply nobody knows what will actually happen—everyone is making educated guesses. It takes the willingness to fail, too, and, I believe, a different culture of certainty, or a different narrative of certainty, in the organization. Nobody knows where our society and the established systems will be in five years. An enormous amount can change, if you look at where we were in AI two years ago and where we are today. It takes the willingness to try things out, discard them and always work toward a target vision where you ask: Was that good or bad? And you can adjust that, after all.

Those are the two factors in a nutshell: First, understanding where you want to go and being able to answer that for yourself. And second, the willingness to adjust it and to see it as the latest best guess or the best current state of knowledge—with an approach that is, in quotation marks, scientific, where you say: We can revise this at any time if the normative power of the factual means we have to revise it.

Maria-Lena Matysik: Super exciting. Thank you so, so much for all your insights and for sharing your experiences. I personally learned a ton—already in our preliminary conversation. Thank you for being here today.

Tobi Nehren: Thank you very much. I hope it was interesting and entertaining.

Maria-Lena Matysik: A hundred percent. Thank you!

Outro

Maria-Lena Matysik: To wrap up, I’d like to leave you with two thoughts from the conversation that particularly stuck with me.

The first thought: Data-driven working and thinking—and above all data-based decision-making—doesn’t start with a tool or a dashboard, but with the question: What question do I actually want to answer?

We often see this in our day-to-day work. Marketing managers who aren’t that data-savvy or used to data yet come to us and say something like, “Could you please build us a website performance dashboard based on Google Analytics data?” And then we have to dig deeper and ask: Who specifically is going to look at this dashboard? What decisions does this person want to make? And which data points do they consider relevant for these decisions?

For many people, it’s not that easy at all to make this really concrete. Often the first answer is more like, “Well, I want to see our user numbers and the conversion rate by channel” or something like that. But why? What is that supposed to tell you?

I found it super interesting to hear from Tobi’s perspective and experience how he deals with this and what his solutions to these challenges are.

My second key takeaway: Data-driven work is a leadership task. Team leads, department heads and CMOs can’t simply delegate the topic to the people responsible for data or IT—along the lines of: “Our head of data takes care of everything, and in the end the whole company makes data-based decisions.” That won’t work.

If I may put Tobi’s message into a somewhat lofty saying—sorry, Tobi, and sorry, Gandhi, ChatGPT says the line supposedly comes from Gandhi ;)—“Be the change you want to see in the world.”

If a whole team is supposed to actively work with data and think in a data-based way, it’s essential that the team lead models and demands it. If team leads or the C-level obviously go with their gut and don’t factor in data, then their team won’t do it either.

And with that, I’ve reached the end of my notes—and the end of the episode. I’m always happy to get feedback and thoughts on the podcast, whether on LinkedIn, by email or via the contact form on our website.

Until next time—bye, bye!