How can we make sure our analyses lead to action?

I’m joined by Dr. Annika Bergbauer, economist, IT management consultant, and host of Datenaffaire, to explore how we can put people at the centre of data-driven decisions. We discuss stakeholder engagement, clear communication, and the human skills that help us turn technical work into practical impact.

Why you should listen

We can produce excellent analyses and build powerful tools, yet still struggle to change how decisions are made. In this conversation, we explore how we can close that gap.

Together, we discuss how we can:

  • Involve stakeholders early so we build solutions that meet their needs.
  • Understand the fears, incentives, and pressures behind resistance to change.
  • Use personas to make our analyses, dashboards, and presentations more relevant.
  • Tell clear stories that help our findings lead to action.
  • Strengthen our listening, business understanding, and leadership skills as AI changes our work.

Episode highlights with timestamps

  • 01:30 — How we turn data into better decisions
  • 02:22 — What we can learn from Annika’s journey
  • 03:48 — Why we cannot always access the data we need
  • 05:09 — How we move from describing outcomes to understanding causes
  • 06:59 — What we face in randomized trials and observational research
  • 08:20 — Why our successful pilots do not always become useful tools
  • 11:35 — How we bring the right stakeholders on board
  • 12:57 — How we make our message clear
  • 13:50 — How we understand incentives and everyday pressures
  • 15:17 — How we create and use stakeholder personas
  • 17:41 — How we develop our skills as AI changes our work
  • 19:34 — Why we need to listen before we present
  • 20:33 — How we build trust through human conversations
  • 21:59 — What we bring to the table as humans

Links and Resources:

🔗 Connect with my guest, Dr. Annika Bergbauer

🔗 Listen to Annika’s podcast, Datenaffaire

🔗 Explore the Datenaffaire episode archive

🔗 Learn more about PSI and become a member

🔗 The Effective Statistician Academy – I offer free and premium resources to help you become a more effective statistician.

🔗 My New Book: How to Be an Effective Statistician – Volume 1 – It’s packed with insights to help statisticians, data scientists, and quantitative professionals excel as leaders, collaborators, and change-makers in healthcare and medicine.

Join the Conversation:
Did you find this episode helpful? Share it with your colleagues and let me know your thoughts! Connect with me on LinkedIn and be part of the discussion.

Subscribe & Stay Updated:
Never miss an episode! Subscribe to The Effective Statistician on your favorite podcast platform and continue growing your influence as a statistician.

Never miss an episode!

Join thousends of your peers and subscribe to get our latest updates by email!

Get the shownotes of our podcast episodes plus tips and tricks to increase your impact at work to boost your career!

We won’t send you spam. Unsubscribe at any time. Powered by Kit

Learn on demand

Click on the button to see our Teachble Inc. cources.

Load content

Dr. Annika Bergbauer

AI & Data Transformation Leader | Evidence-based decisions for management and users | #DrOecMama: Economics, AI & Parenthood

Dr. Annika Bergbauer connects AI transformation in industry with empirical evidence, sharing her insights with IT clients, university students, and listeners of her podcast. She translates complex technical concepts into clear, accessible language and drives AI culture within organizations. Her academic background as a PhD economist shapes her broader perspective as a leader and manager.

Transcript

00:00
You are listening to the Effective Statistician Podcast. The weekly podcast with Alexander Schacht and Benjamin Piske designed to help you reach your potential, great science and serve patients while having a great work-life balance.

00:21
In addition to our premium courses on the Effective Statistician Academy, we also have lots of free resources for you across all kind of different topics within that academy. Head over to theeffectivestatistician.com and find the academy and much more for you to become an effective statistician.

00:49
I’m producing this podcast in association with PSI, a community dedicated to leading and promoting the use of statistics within the healthcare industry as benefit of patients. Join PSI today and further develop your statistical capabilities with access to the ever-growing video on demand content library, free registration to all PSI webinars, and more. Head over to the PSI website at PSI.gov.

01:18
to learn more about PSI activities, if can’t to sign up for the 2 page.

01:30
Welcome to a new episode of the Effective Statistician. And today we are looking into a very important thing that is making good decisions based on data. And everybody in clinical research knows about it, but today I want to look at it from a little bit of a different perspective. And for that, I have a fellow podcaster here. Hi Annika, how are you doing? Hi Alexander. I’m very well. Thank you. How are you?

01:59
Very good. We are recording this in the middle of the summer and Anika has a new baby beside her. So maybe there will be some noise in the background, but who knows? Let’s see. We’ll be charming for sure. Maybe before we go into the topic, Anika, you can introduce yourself and what brought you to starting a podcast yourself.

02:22
Yeah, so I’m working as a IT management consultant. I’ve been in consulting for the past seven years. And before I’m a trained economist, I did my PhD in economics as well. I’ve worked a lot with big data sets. I’m originally from the field of economics of education. So I looked at data sets like the PISA study, which many of us know in Germany and it exists internationally. So roughly a hundred countries around the world take the test every few years.

02:52
And you get huge data sets of around 3 million students with 300 background variables. And then try to determine what is actually decisive for their learning outcomes. So I come from working with big data from coding and through my work in IT management, I’ve actually become a manager and I’m rather concerned with how do people act with data and algorithms than the algorithm itself working properly nowadays.

03:19
And I started my podcast when I actually also started to work in companies and I found out, hey, the data is not as cleaned up and available as I imagined coming from academia. There’s still a long way to go in companies until we can really take data-based automated decisions. Yeah, very good. As you mentioned, you work now more in the IT department and IT area. And of course, IT systems generate data all the time.

03:48
Every minute, every second, every reaction that we take, think the best example is probably any kind of social media data, but also in companies we leverage all kinds of different data. Can you give a couple of examples of what kind of typical data sets you would look into? Yeah, I would say this whole idea of companies use their data very much depends on their IT architecture, so how their systems are connected with each other.

04:16
Usually we don’t have a company which was set up two years ago in the ideal design to work with data and AI, but we have companies who grew for at least the last 20 or 50 or a hundred years. And maybe they have seen mergers or companies were bored and all those different IT systems just grew. And oftentimes they are not connected or connected in a poor or in a historic way. So data is not flowing as freely as we would like to.

04:46
And there’s also often restrictions. So I worked a lot in insurance and finance. So there’s often regulation, which has to be met that not all data can be connected to each other. So this is the basic setup. I want to say where we’re working and then on how data is used in companies. I would see on, so to say a four-step staircase.

05:09
where we have really basic analysis of looking back what actually happened and which groups and which teams and which countries trying to see the data there. That’s usually the easiest analysis. And still I’ve seen when I worked as a data scientist in consulting for companies, that data has not been readily available. You could not just enter your account credentials and just look at the data. At least you were allowed to.

05:37
But I oftentimes really had to go and collect data and talk to those people who entered the data because they were the only ones who understand how the data came into being and how it’s to be read. It was not self-explaining, nor was there any metadata to explain it, so any documentary. And then I would see the second step on the staircase would be forecasting. So what will happen next? The very standard example is predictive maintenance. So when do I have to look at my machines again? When do I have to repair it?

06:07
Then the third step would be correlation. What moves together is there really oil prices go up and this moves together with higher energy prices in Germany, for example. And the fourth question I would like to ask, is that really causal or is energy prices going up in Germany driven by another effect, which is not revealed yet or by the analysis I’ve been doing.

06:29
And I don’t really see that in companies being done yet, except maybe for the classic A-B testing and user design where we want to say, okay, on the website, is the button black or blue, or is it on the left or on the right-hand side? But yeah, as an economist, I’m very much interested in this causal interpretation. And I would really love to run an RCT or randomized controlled file once in a company. Maybe I’ll see that someday.

06:59
Many of the listeners work with RCTs all day because it’s a requirement for clinical research or nearly a requirement. Very rarely you can get something moving forward without an RCT. But of course not always a RCT can be run. And so you need to work with observational data and then still you need to do something. Or if you run and randomize something.

07:25
Really completely randomized is the RCT only very much at the beginning. The longer you follow up, more all kinds of different things happen. And so your randomization gets distorted. Let’s say it’s that way. And then you need to have causal inference. There’s attrition when you have long-term interactions. And also, as far as I know, medical studies are not really well known for big treatment groups, right?

07:51
It’s not like you get a thousand or a few thousand people in there and have really heavy statistical validity. Yeah. So it depends on the treatment effect that you want to see. Yeah. There’s certain cardiovascular studies that maybe have tens of thousands of patients. I would say that is more the exceptions having these mega studies. Most studies are smaller, but nevertheless can still be very costly. Yeah, absolutely.

08:20
especially when you have to observe patients over years, for example. When now companies have these internal data or external data, whatsoever data for making better decisions, what are typical challenges for making better decisions based on data? So what I’ve seen in the past years in companies is not that the algorithm was wrong. Of course, you have to be careful with the data and quality and so on. But the most difficult

08:49
step I saw for companies was when actually putting those nicely run MVP pilot algorithms and data works into production and really making the users, the end users work with it and convincing them they should change their old habits of how they’ve always done their business and now work with data and AI and really create a need at those users that…

09:16
They really want to change their way of working and they see an advantage in it. And not only there’s always a handful enthusiasts for technology who would like to use it, but especially convince those who’ve been working for the past 20 years with paper and pen and probably are also not trained in a quantitative analytical field, but have been working a lot with texts or with people and to convince those and not force them.

09:42
that will create even more resistance, but really convince them and understand their point of view, their incentive structure, their fears and needs. And that’s ideally done before you’ve created the data analytics and the AI that you don’t waste money on changing your AI product after you’ve talked to the people. But it’s really the user opinion and need is already built into the design. That’s interesting because I see similar things.

10:12
in the medical research space as well. Just because you have a new way of looking at the data, of interpreting the data, analyzing the data, that doesn’t mean that people will act differently on it. Or that they will even look at it or accept it or let alone use it. And as you said, there’s sometimes kind of this avant-garde of people that’s…

10:37
Usually always interested in the new just for because it’s a new and they will try it out. But the vast majority will be very pessimistic about it. It’s uncomfortable, right? I have to use so many technology tools already. Why should I use even one more? Yeah. And what’s the value for it? Why is the old way of doing things not at least as good as the new way of doing things? And maybe it’s frightening. Maybe I don’t understand it.

11:05
That’s a different topic, think. A couple of years ago, now it’s already quite a couple of years ago, in the clinical research world, the whole thing around causal inference as demands, things like this has been introduced and it’s still difficult. Also in the observational research space, like what you’ve just mentioned, these more educational aspects, that’s not a new thing. Yeah. People have used propensity scores and all different other methods for decades.

11:35
But in clinical research, it’s a relatively new area and people struggle with it quite a lot. So what are your techniques to get people to embrace these changes? So I would come with the basic tools of being a consultant, right? I would start with a stakeholder analysis, like who are the people I have to consider for my data and AI product? Who is involved in decision-making? Whom’s support do I need?

12:04
And who in the end will use it. And then I would also group those stakeholders into groups. Do they make decisions? Do I need to ask them? Do they need to use it? And then also create personas around those people, groups, and to be very precise, how old are they? What educational background do they have? Which training? What incentives do they have? What fears? And then really try to address those.

12:31
prioritized by how I group the stakeholders beforehand. Do I need to consult them? Do they make decisions and so on? So take them into the boat early on. And I know that something people who work quantitatively and really love their data and their programming do not really like to get into those management tasks. And then I can only recommend to get help of people who can really bridge those two worlds of.

12:57
doing management tasks and understanding how valuable and important quantitative work is. And also another step then further down the road is when you’ve done all the characterizations of your stakeholder groups and the personas, then to really tell a story with your data and really try to keep it as simple as possible. I know that in academia, for example,

13:20
It’s very fancy to put a lot of numbers and graphs into your slides and presentation, but those people you will, as a data scientist, then talk to and present your results to. They have little work, they have little attention. They’re really very much interested in the result and not in the how you got there. And yeah, try to make sure that they get the message and that their needs are met. Because if you’ve created a wonderful data analysis and no one’s going to use it.

13:50
You will not feel very happy afterwards because you feel like your work may be useless or not appreciated in a way it actually deserves. Yeah, completely agree. And I love your point about the personas that it is a lot about their incentives and their pain points. So you can talk about kind of titles and gender, age, whatsoever, but these two things, these, so to say, psychometric variables have a huge influence.

14:18
And if you don’t know what these are, yeah, it becomes really difficult because you need to understand how you can convince them to use it. And that is by addressing an incentive. Maybe they’re incentivized by doing something cheaper or faster. Yeah. That could be something, or maybe they are incentivized by being innovative. Could be another thing, or maybe they have a pain point that certain things take very long.

14:48
And then some kind of dashboard helps to get through the data much faster than hundreds of PDF tables. Yeah, I’ve worked with users who are not really incentivized that they get paid more or something when they work faster or better or more innovative, but they just feel they have such a feeling of responsibility. They say they really want to cope with their work and do their work, but they are at the same time so overwhelmed because there’s so much work waiting for them.

15:17
And they really have the feeling I need to push on and I don’t have the time to learn a new tool that will just cost me time. And it’s really then about showing them they will save time and they will be more productive and work better after they have gone through the training for the new tool and they will be able to do the work they feel pressure from in a better way afterwards. Now we talked a little bit about the word persona and we actually didn’t explain what that is. So for anyone.

15:46
who is listening and doesn’t directly have a concept in mind of what his persona is. How would you describe or explain a persona? Yeah, I would really work with pictures or at least comic pictures of a person. And for every relevant stakeholder group, I would create a persona. this could be even among the users, this could be two personas when I have two distinct groups among my users.

16:13
And that’s why I talked about age and gender and so on, to really have a concrete person in mind, to have it always in my mind when I design the algorithm or the output of the algorithm or the presentation of my analysis or the dashboard. And then I would always think in those pictures of those people of that certain group and then also design the picture of that person in the age, in the gender, maybe also with the clothes.

16:42
of that persona, of that target group, and then put around those other characteristics as incentives, as tasks they also even have to also make it clearer for my own working, like what do they actually do all day and which data maybe do they need and which they don’t need. And there it’s also very helpful to talk to those people, to look over their shoulders. You can do that on place in an office, but you can also do that

17:11
virtually online, they can share their desktop and show you what they actually do in which systems, which data they work with. Yes, these personas basically resemble a typical segment subgroup of the people that you want to talk to. And having that in mind makes it easier to communicate to them. When I do my communication as well, I have sometimes even a specific person in mind that I want to talk to.

17:41
like Peter or Claire or whoever. Yeah. And then I talk to that person because I know that person stands for a specific segment of my listeners. If we as more quantitative people want to convince others and not just stay in the purely technical area, what kind of typical skills would you see, what you need to be successful?

18:08
That’s a big question. A few podcast episodes recently on the skills for the future. And I’ve also been just on a career talk last week where I had young data scientists, analytical people telling me there’s the LLMs now and they actually do the work I like, which is fun for me. What job am I going to take now that they’ve taken away my dream job? What am I going to do now? And I also of course, cannot look into the future and…

18:36
I can just tell you, try to find out what other passions you have and where you’re strong in. And of course we will always need people to check the code, but not as many as we had in the past few decades and try to train a bit more your people skills. Exactly what we’ve talked about. On the one hand side, preparing your data and AI projects to have the right people on board and the right setup for your analysis. And on the other hand side.

19:05
Also to be able to understand the business or the underlying questions you’re analyzing. And then also to take roles like working with other people, becoming a team leader. Those are things like being in the physical world, working in the physical world and working with people who for now will not be taken away by AI. But we certainly have to react to this change in our labor market. We shouldn’t ignore it. I completely agree.

19:34
You mentioned a couple of important points, understanding the business, how it changes. And that doesn’t not only mean how theoretically your company should work, but how it actually works. Who are the decision makers? How are decisions made? What are pain points within your company? Where is innovation going? How can you drive innovation forward? The listening part is also really important. We talked about the personas.

20:04
How do you get to these personas by listening to people? Hugely underrated part. Yeah. Because without that, you go nowhere. You can be the best presenter. If you’re not a good listener first, your presentation will fall flat because you actually don’t know who you’re going to talk to. Absolutely. Yeah. And there we are with the needs again. What does your counterpart actually need? And then you can configure what you have to that certain audience. Yeah.

20:33
I once saw this meme about LLMs and programmers and it was something like, people don’t know what they actually need. We are safe. And there’s something true about it. As a technical expert, you know what the needs are, but also what are the opportunities, what are the possibilities. And bringing these together is a very unique skill. And to get this out of the people, right?

21:00
You could also, of course, send around an AI generated questionnaire, but we know the respondents rate is not so very high with online questionnaires. It’s higher when you do it face to face, person to person. And then maybe there are new questions coming to mind. You’re going to interfere there and go deeper into a topic which you couldn’t anticipate. Or people only feel comfortable sharing what their true beliefs and values and needs are when you sit in a

21:29
comfortable office, maybe over a coffee and you talk in a more informal way and then only you will get to the ground of their true feelings and what’s important to them and how they make their decisions and not from having a formal business meeting and filling out those questions. Yeah, I completely agree because one has a lot of trust in it and the other doesn’t. And I think people are far away from completely trusting an AI counterpart, which I think is their point.

21:59
And that’s good. That’s good. We should keep checking on what they are doing and always take it to the grain of salt. Because if we would completely trust them, then I think we would lose our competitive advantage. What else can we humans bring to the table? Thanks so much for this discussion, Annika. I really enjoyed it. Thank you for the invitation. I need to say I have a huge respect for you recording a podcast episode. Well, at the same time.

22:27
Being a young mother and taking care of your little one. Shout out to you for this one. And I can recommend people to listen to Annika’s podcast as well. Just remind me what was, it has a pretty nice name. Datenaffäre. Datenaffäre. Have a look into it. We’ll put the links to it in the show notes. Thanks Annika.

22:55
This show was created in association with PSI. Thanks to Reine for help with the show in the background and thank you for listening. Reach your potential, leave great rooms and so be it. Just be an effective statistician.

Join The Effective Statistician LinkedIn group

I want to help the community of statisticians, data scientists, programmers and other quantitative scientists to be more influential, innovative, and effective. I believe that as a community we can help our research, our regulatory and payer systems, and ultimately physicians and patients take better decisions based on better evidence.

I work to achieve a future in which everyone can access the right evidence in the right format at the right time to make sound decisions.

When my kids are sick, I want to have good evidence to discuss with the physician about the different therapy choices.

When my mother is sick, I want her to understand the evidence and being able to understand it.

When I get sick, I want to find evidence that I can trust and that helps me to have meaningful discussions with my healthcare professionals.

I want to live in a world, where the media reports correctly about medical evidence and in which society distinguishes between fake evidence and real evidence.

Let’s work together to achieve this.