Audience Development Deep Dive with "We Like Mags"

Christian Kallenberg [00:00:02]:

Welcome, dear listeners, to the first episode of the Audience Development Deep Dive by SPRYLAB Technologies and We Like Mags. Very briefly, what is this all about? You may know the We Like Mags podcast by me, Christian Kallenberg, as a B2B format where decision makers and decision makers talk about themselves and their work in the publishing industry. One of my first interviewees in this podcast was Benjamin Kolb, CEO of SPRYLAB Technologies, with whom I had just talked about audience development, among other things, and who is sitting across from me now. Hello, Benni.

Benjamin Kolb [00:00:39]:

Hello, Christian. That's right. Back then, at the end of our conversation, we both had the feeling that by no means everything had been said here. That's why we decided to dedicate a series special to the topic of audience development, in which we would interview guests together on the various aspects of these increasingly important topics.

Christian Kallenberg [00:01:01]:

Yes, and before we get started: Benni, please tell us again in one or two sentences how we actually define Audience Development, so that all listeners of course know what to expect here in the next few minutes.

Benjamin Kolb [00:01:16]:

I would like to try. Audience development in the context of publishing houses can be explained in the same way as audience development for other products and brands, in principle, in terms of conversion to the actual business model, i.e., from awareness, acquisition of new customers via familiar digital paths such as search engine optimization and marketing, to the topic of customer retention, engagement, and generating loyalty in the conversion of the business model, i.e., paid content at publishing houses, advertising at publishing houses, or also the participation of third-party revenues via the brand, for example, own products or cooperations. Audience development in the publishing environment is therefore quite wide-ranging and diverse, a large complex of topics from which we want to pick out individual topics with our guests in which they are specialists and discuss individual topics with them in appealing depth. Of course, we also want to introduce our guests personally and learn a little about their lives and interests outside the publishing world.

Christian Kallenberg [00:02:18]:

Oha, I can already see that this is going to be a big program. Let's just get started now. Here we go. Our guest today: Peter Dyllick-Brenzinger, Head of Product Editorial at Axel Springer NNT. Hello, Peter.

Peter Dyllick-Brenzinger [00:02:33]:

Hello.

Benjamin Kolb [00:02:34]:

Hello Peter.

Christian Kallenberg [00:02:35]:

Peter, that's a great-sounding title, of course, the job title. What do you do all day?

Peter Dyllick-Brenzinger [00:02:41]:

I take care of all the products that we buy in and build ourselves for our digital editorial team, primarily BILD and WELT as the big brands, of course, but also the smaller ones, specifically the UX. These are content verticals, or our colleagues in Hamburg, with whom we don't yet collaborate much, but that is still to come. And at the core of these tools that we make, it's actually a CMS and then a lot of tools around it and that also includes something like analytics tools or smart little bots or assistance systems. Exactly, and perhaps very briefly: Our goal is to give our editorial teams superpowers in the best case scenario. Of course, there's also a lot of brown bread, i.e., a lot of things that simply depict everyday life, but we actually want to offer our colleagues in the editorial offices a real competitive advantage with our work.

Christian Kallenberg [00:03:36]:

Okay. You've just mentioned the colleagues in the editorial offices. I would now be interested to know: How important or how closely do you work with them? How important is the exchange with the journalists, I'll call them now?

Peter Dyllick-Brenzinger [00:03:48]:

Quasi crucial. So that's actually also the decisive advantage for us, that we have precisely this access. In the new building, where we've just moved - if we were in the office, if we were in the new building - we can simply take the elevator to the WELT colleagues, we can walk across the street to the BILD colleagues and talk to them. We also have many contacts on an ongoing basis, i.e. we also have many regular appointments and we also constantly try with our work to test this in a modern product development process, i.e. to test our problems and hypotheses, to test our solution hypotheses accordingly and we are in close exchange.

Christian Kallenberg [00:04:25]:

Okay, what KPIs do you use to measure your work? So when do you say: "Today was a really successful day for me"?

Peter Dyllick-Brenzinger [00:04:32]:

Yes, that's actually not so easy. We are still discussing exactly how to do this, because at the heart of the matter, we are of course like many others, i.e. IT activities in a corporate context. Are we guided by productivity? And productivity is always a bit difficult to measure, especially in such a creative environment as journalism, because of course you could just bluntly say: "We simply look at how many articles a day or how many videos a day were somehow produced by us with the tools", and then everything is great. So we could now look at, actually, "Today they wrote 500 articles." That's the big challenge, after all, is that most of the time we always don't know what the right ones were until afterwards, so at least often it's like that. So what we definitely look at are efficiency metrics, so to speak. We already see them, that is, that we look at: How quickly do people get through the processes we set up? In the broadest sense, you can compare an online publication process with a checkout process. There are simply certain steps, and each step that we take out of it is better, so it really is a step-by-step process. That's something we're definitely looking at, and then we're also working together with our colleagues, then more in the analytics context on the classic online publishing metrics, i.e. reach, subscriptions, purchased subscriptions, and topics like that.

Christian Kallenberg [00:05:49]:

Peter, are there any international role models or international references in your work? Do you look at: What are the New York Times or the Washington Post doing in those areas, or are you focused on your own work and not so crazy looking at what's going on in other countries?  

Peter Dyllick-Brenzinger [00:06:05]:

No, we're always looking at what's happening and who's doing what. Almost five years ago, we decided to build our own system, and it wasn't without reason that we were inspired by people like the Washington Post with Arc or Vox Media with Chorus or Business Insider, who had already built something of their own. We have looked at the market very, very much and are still looking at the market, even if it is of course always difficult to really look behind the scenes. But we also have a bit of a network, i.e. people with whom we are in good exchange, and we always look at what is possible and makes sense.

Benjamin Kolb [00:06:48]:

Exactly, our podcast episode here is also about the topic of audience development, so I'd be interested to hear more about that: In all your tools that you now offer to the editorial team, how much does it revolve around the topic of engagement and loyalty? What tools do you offer the editorial team to bind the target group more closely?

Peter Dyllick-Brenzinger [00:07:06]:

So we have a whole range of tools. We have a team that deals exclusively with intelligence and analytics topics. We like to call it "Actionable Insights" because that's something that's really important to me. Or what has also turned out to be totally valuable is that we are moving away from classic dashboard views and towards actually intelligent assistants or quasi dashboards that not only display the figures, but also offer a lot of context, so that you don't have to take such a big step in your head to really draw something directly from them. But that means that we also have real-time analytics based on Adobe data and then a whole series of Slackbots based on this data. This is something that has proven enormously successful at BILD and that we are now also using very successfully at WELT, where we then use this data to make little tips, as it were, that then somehow come directly to the editorial team via the existing communication channels, in this case Slack, and then happen via our own channels, as it were. And that's also something that relates to a relatively large number of reach aspects and, in the plus version - we always want to sell subscriptions - also has a lot to do with commitment.

Benjamin Kolb [00:08:25]:

Yes, now BILD is probably also one of the fastest (rotating) editorial departments in Germany. So it's probably also of central importance that real-time data also leads to real-time actions in the editorial department, I suppose. Are there really mechanisms in place to react to what your analytics say directly, or is that still a thing of the future?

Peter Dyllick-Brenzinger [00:08:43]:

Well, we have various bots, so you could call them "AI" in a fancy way, but in fact they are just relatively solid statistical models. So that is somehow machine learning, but it is not deep learning in the sense that we predict things. So predictions are then again - that's sort of technically correct, but it's maybe a bit misleading. So we have models that explain to us or that tell us how we should interpret certain things, in other words, that can tell us how we should best react to data that we can measure quickly, that we have immediately. And we actually do that very successfully. So that's something that is seen as an enormous help. We do that both with reach and with subscriptions. And that is something that has proven itself completely and also just - and I found this to be a really great learning for us - just not the wacky, latest, greatest models now in the data science context, but really just as solid statistics - well, I'm a social scientist, I also heard all this at some point, somehow learned it at university - because they are just explainable. And if you then calculate such a model and have a prediction at the end and can then say at the end: "Yes, these are the numbers. If you screw something to them, then you screw the larger number, and we can show you that and it really works that way," and then the editorial team also experiences that moving this number works well and that really works just as we say it does, because there really is a strong correlation. And in part, especially with the reach things, we see enormous correlations in the models, which is also not surprising, but if you then somehow recalculate it and then show it like that and you really have such a mechanism, then that is also really motivating. That's what we experience with our colleagues, that they really enjoy turning these screws, somehow with their work.

Benjamin Kolb [00:10:31]:

Yes, of course that's a super exciting topic. Now, of course, the question for me is: If you turn certain screws and can show the editorial team that you're moving certain numbers, do you also have comparative values? So do you do A/B tests where you say, "When you turn the screws, you see that it actually works better here than on the other side, where you didn't turn the screws"? So you're to the point where you're doing comparative tests that prove exactly that your prediction models deliver better results in principle?

Peter Dyllick-Brenzinger [00:10:58]:

No. But actually also because it simply doesn't work at all in some cases. For example, in the case of reach topics, we actually talk relatively quickly about search and there is simply no A/B test. So you can forget that. Google is no longer interested in that. And then, on the other hand, with the Plus topics, we are in the process of creating an appropriate infrastructure, but that's not quite so trivial when you have as much traffic as BILD. And there, too, I would be skeptical. So there you have another problem in the Plus context, because even if we sell really good subscriptions with BILD and also WELT and can also gain many new customers every day, the absolute numbers are of course not such that you would really get super-significant results in such an A/B test. And then you're more or less back in an area where you actually need proxy metrics, and then it gets really shaky again. So we are working on that. That's also an important aspect, but it's not really that trivial.

Benjamin Kolb [00:12:02]:

Yes, and while we're on the subject of big data, in this age of data privacy, the question is always: How does this whole issue of consent actually affect the tools and methods and the analytics that you actually use as a basis? Does it change anything for you? Do you still have enough data that customers leave to you? Has it even improved because everyone is just clicking "Agree," or what's the impact of that? So how does this work in practice?

Peter Dyllick-Brenzinger [00:12:27]:

Yes, we definitely see that and unfortunately it's not positive. On the other hand, speaking as a citizen, it is of course more data protection, so as a German citizen I don't think it's so bad, but now at work it's actually already an issue that we simply have considerable losses in "normal tracking". So the problem is at this point: We simply continue to work with what we have. We have no other choice, but of course we don't really know whether this is not a distortion, or rather we know relatively well that the number as such behaves differently now. We can see that in any case. On the other hand, which is of course an advantage again: At the moment we sell a subscription, we of course know that we are selling a subscription, so the number is fixed, because at the moment the purchase is concluded, we somehow have a booking in our merchandise management systems. So it is then also clear, because this is a conclusion of a contract, that we can also record this, insofar as this is one of the most important currencies for us, of course, fortunately now quasi: Clean measurement.

Benjamin Kolb [00:13:31]:

Perhaps one more detailed question on this: Especially when you think in terms of cohorts and segments, you can often get the feeling from analytics tools like this that you're somehow looking the truth in the eye, but if you think about it now, it might also be segment-dependent whether people give their consent to such a consent at all, and that maybe some target groups do things differently than others. How do you deal with such a bias? Not at all, or do you try to factor it in somehow?

Peter Dyllick-Brenzinger [00:13:53]:

No, not at all. So you also have to say that we actually only do the real-time analysis and the historical analysis is done by other colleagues, also in another department. And we don't stand a chance in the real-time view anyway. And in fact, it also has to be said that in the real-time view, we hardly ever look at visits; instead, we are actually in a PI world and then calculate the models accordingly on the basis of the PIs. So we're also working on personalization topics and we're already in the process of exploring that, so to speak, but actually we don't have a chance there either, of course. So if someone comes in who doesn't give consent, so to speak, and doesn't want their data to be processed in any way, then we can't play them any personalized content, and that's okay. I mean, it's the decision of the respective reader.

‍

Christian Kallenberg [00:14:48]:

Peter, I have to take another quick look and go back to your own KPIs against the backdrop of this non-consent traffic: How much weighting do you give? What is important? Reach or actually the plus model, i.e. writing subscriptions? That's kind of a balancing act that you have to manage, isn't it?

Peter Dyllick-Brenzinger [00:15:07]:

Yes, totally. The goals actually contradict each other. There's no need to deny that, because every non-subscriber I send to a conversion page, i.e. to a Plus article, regardless of the brand, may be one who drops out and won't read the second or third article again. So of course, if he buys it, then hopefully yes, then it is of course also an effect that I then have there, but that is of course a clear risk. And the same applies to Search, for example. You always have to weigh up between: Is the article actually something we can use to generate real traffic on Google? Then I'll just leave the subscriptions. Or the other way around. And we're in the process of measuring that, too, and then making a recommendation accordingly.

Christian Kallenberg [00:15:54]:

Okay. I would still be interested: The decision, which article is Plus and which is Free, that is also played out differently from user to user with you, right?  

Peter Dyllick-Brenzinger [00:16:06]:

No, not yet actually. That's something we're aiming for as a dynamic paywall. The Wall Street Journal has been very successful with this and is always the big, shining example, but as of now everything is either Plus or Free.

Christian Kallenberg [00:16:24]:

How important is visualization for the journalist in this decision-making process? So the one question is: What tools do you have? What can they do? But the other question is: How quickly can they be understood and implemented as analytics for the journalist? The journalist is your customer, so to speak, isn't he?

Peter Dyllick-Brenzinger [00:16:47]:

Yes, exactly. Yeah, sure, totally. So we are like toolmakers for them. That's what our customers really are, or users actually are, because fortunately they don't pay anything, or rather they are charged in a different way. Exactly, so the visualization - I think two aspects are important. One is that we very often do without visualizations and basically only convey the insight. And these are then actually just simple text messages where we simply write that, for example, an article has lost its ranking, which we find out using a relatively simple time series model. And that's enough. Then you know: "Okay, this is now the article", and then you can look again in a real-time analytics tool in the visualization somehow, you can also look again, how it stands now somehow in the corresponding analysis tools and in other commercial analysis tools. That would be one aspect. And the other thing is that where we visualize, that is, where we have overview dashboards and such, we are actually very visual and show few numbers. So that's also something that we've learned. There are individual people in the editorial offices who have the leisure, so to speak, to spend so much of their time on numbers in their function that they appreciate really having tables, really numbers, but that actually the majority of colleagues really need it quite strikingly.

Christian Kallenberg [00:18:19]:

Pie chart or what?

Peter Dyllick-Brenzinger [00:18:20]: Yes, bar charts for example is something very important. We have sparklines for example. That plays a huge role for us, because they simply allow us to see at a glance, really at the line level, at the individual article level, where things are going. And that's where we've learned a lot, for example, with scales and things like that, so that you can actually achieve comparability across the lines somehow. And then it's a matter of really quickly grasping things visually and getting away from pure numbers. So what I've actually learned is that the ideal state is usually even without numbers. Of course, that doesn't always work, because not all patterns that an experienced editor can recognize quickly can be mapped really well by machine, but there are a surprising number of patterns that can be mapped relatively well.

Christian Kallenberg [00:19:07]:

Okay, and on the patterns to be mapped, that's where we come back to the tools that you're using, and that brings me to the question of buy and build. You talked about this earlier. You guys also do a lot of your own work, but of course you also buy in solutions. I would be interested to know: What is the ratio of software that you integrate externally and how much do you do yourselves?

‍

Peter Dyllick-Brenzinger [00:19:30]:

Well, I think we do a lot ourselves, I have to say, and especially in the context of analytics, where it's not so obvious at first glance, but that's where we really had the issue - before we decided to actually build our own real-time analytics solution, so at least the processing and display part, that is, data collection, we do via analytics, but virtually everything that comes after that, we built ourselves. We actually had the problem that what we needed was not available on the market. So the combination of real-time and then with a strong focus on conversions, i.e. on the paid content model, which we needed, was not available in this form at the time, at least with one exception, which then did not work strategically, so to speak, and that's why we had to build it ourselves, which has also proven itself to a certain extent, I think, because we somehow have a very good grip on this data and can then put these quasi advanced, higher-quality solutions on top of it. In fact, it's just a bit of an issue, because where we work quasi statically or now with data science means, are actually now geared to traffic, to reach, to the now perhaps also audience, we actually don't see any solutions on the market that serve that as precisely as we actually need. At least, that's not what we've discovered now. There are other areas where this is more suitable, i.e. actually in the processing of articles, where things can be simplified. And we have also purchased something like that. But, as I said, in the analytics context on the one hand - we built the solution ourselves, but of course also with many cloud services. That's something you can't ignore here, because it's something that would not have been possible five or seven years ago for such small teams as we have here to somehow set up such a relatively substantial infrastructure. And we really do use a lot of very strong, powerful components from ABS. And that is actually also a kind of shopping.

Benjamin Kolb [00:21:38]:

Yes, you said earlier that you also want to get users into subscription models, so to speak, because you want to make yourselves a bit independent, but advertising still plays a role. As Head of Editorial, what I'd like to know now is: How do you actually see the editorial department of the future? So what does an editorial office of the future actually do? And is this whole technical apparatus, let's say, that lies underneath it as a basis, an essential component for you in your vision of the future? Perhaps also in the context of the fact that nowadays you can reach a huge audience as an individual via social media and, of course, also spread news through them. What do you see as the newsroom of the future?

‍

Peter Dyllick-Brenzinger [00:22:21]:

So I think that's a very concentrated editorial team on the content. So that's also my vision for our work now, so to speak. In concrete terms for our department, we are actually disappearing more and more, but disappearing in the best sense of the word, so that we can get out of the way, so to speak, so that input of non-creative work is somehow automated more and more and is then actually no longer visible. In other words, that the expression of what the journalist experiences and wants to express is immediately publishable. That would be the vision, and I'm super-optimistic that it's possible. So if you look at what is happening with the newer language models, i.e. GPD-3, and what is possible in terms of text editing in the broadest sense, text generation, text smoothing, text work in that sense, then I find it totally exciting. And the same applies to all other media, be it photos or videos. There's a lot that's possible nowadays, and most of what's part of online publishing nowadays or what an online journalist has in his or her job description doesn't necessarily create value for the reader, but a lot of it is almost archival work, assigning keywords, writing a description for the service, writing another description for the service, writing three different versions of the headline, just because the teasers are somehow different in size on the different platforms or something. So a lot of that is not the core of what a journalist does or wants to do or should do. And I think that this is actually the great opportunity of our time, and it's also kind of the mission that we have, namely to actually take away this work, which doesn't really have to be, because it doesn't really represent the actual value, and to basically free them and give them time to do what they actually want to do.

Benjamin Kolb [00:24:42]:

Exactly, so for me it looks a bit like this, the vision that in principle you will ensure through a massive underpinning of technology, which in the end may no longer be so perceptible or noticeable for the journalist, that everything that has just been built up through technology hurdles, which a journalist has to do in addition, disappears again and he basically works the way he worked before the Internet was basically there, namely that he concentrates on research, good stories and that becomes the core of his work again, so to speak. Have I understood that correctly?

Peter Dyllick-Brenzinger [00:25:14]:

Yes, exactly. Yes, although maybe even better then. Of course, it's also nice to be able to just sit in front of the typewriter again and hand the three or four pages over somewhere or put them in someone's basket. If you look at how many steps you need nowadays to go through a real story completely and so on, that's also quite crass. That would certainly be a good thing, but I think you can almost go further. I believe that you can reduce the number of coordination loops, for example, because you can automatically achieve a higher textual quality when creating the text, for example, by using assistance systems and even - but I think that's still very much in the future, and of course that always leads to a lot of unrest - but also formulation aids. In the meantime, there are exciting products in the American context for more transactional marketing texts or logos, and I think you really have to keep an eye on that, because what is currently possible with GBT-3, i.e., this open-air model, is already extremely striking. And whether you don't perhaps at some point also move over to the fact that you actually only create a research or a content somehow as an outline and then quasi get the first draft generated somehow, which you then revise again or something. So I could imagine something like that. Maybe not for everything. Certainly not for big feature pieces, but for news. So there are simply different text genres where I don't think anyone would be so sad. Maybe even there, but I could imagine that something could be done in that direction.

Christian Kallenberg [00:27:03]:

But I would actually still like to know how important - you've already talked about this - speed actually is in the delivery of your work. So how quickly does the journalist have to get the data and be able to penetrate it? Are there any guidelines where you say that it has to be clear within ten seconds? So that's also a question about the measurability of your own work.

Peter Dyllick-Brenzinger [00:27:31]:

Yes, that's actually something that's important to us, but where, to be honest, we still need to catch up. And that's actually totally exciting, because speed is an inherent journalistic value. So two identical stories and one is out faster - the faster one is the better story. And the same must then apply accordingly to all our systems. So, as I said, we're not where we should be, I think. For example, I find super inspiring what Superhuman is doing. It's an email client that's crazy expensive and is really aimed exclusively at the Silicon Valley elite. And right from the start, they've managed to get everything they do down to less than 100 reading seconds. That means you have the feeling that you have wings when you use this tool, because it simply reacts so quickly. And that, for example, is something that totally inspires me, because then again, of course, aspects like that come in, that you also bring people into a flow like that, that you simply also allow them to glide through their work processes without stagnation, without somehow stopping them. As I said, that's something that's really important for us, which is unfortunately even more of an aspiration than a reality. So, now that's more generalized to editorial tooling. With the actual analytics or insights, it is of course totally important to be fast, but then you also have to honestly say that of course we can't react immediately. At least for us in the editorial offices with which we work, Axel Springer, no one sits around and waits, but they all have crazy full days. And then you have to look. But that's also the task of our models and of the things we find out, that you then only send the ones that are really important. And then, of course, it's also a matter of being fast, but precisely because you can't react immediately, it's not so bad if we need a minute longer. So in all these places it would actually be ideal to have that immediately. So again, maybe to your question, because you also mentioned such concrete numbers: We don't have any fixed thresholds now where we say, "Okay, this has to get out there so and so quickly now." We have target values for the performance of our tools. As I said, these are simply goals that, like good goals, are always a bit demanding. Exactly, and we also notice when things get stuck. Our colleagues in the newsroom notice when things get a bit sluggish or so. Then they get relatively quick feedback: "Is there an error somewhere? Is there something going on?" But we're also getting better at recognizing that ourselves.

Christian Kallenberg [00:30:07]:

Okay, that sounds so Peter, as if that's actually a job that keeps you busy 24/7 more or less, so if you have to think about it also maybe then at home, not only at Corona home office times. But now I found out that you - and I hope I'm allowed to ask and say this - so not only do you have a common professional background with Benni, but you also have a common hobby. You both love to cook. Do you actually know that, both of you? Have you ever stood together at the pot?

‍

Peter Dyllick-Brenzinger [00:30:36]:

No.

Benjamin Kolb [00:30:37]:

Unfortunately not yet, no, but the exciting question is, of course, Peter: In which cuisine do you feel so at home? Is it the Mediterranean cuisine or is it more Asian or what is your thing?

Peter Dyllick-Brenzinger [00:30:50]:

No, actually I cook a lot of pasta, of course, like somehow everyone and accordingly so really probably then somehow Mediterranean, but what I also like to do extremely, is somehow everything that is a bit more French, even if not so the really sophisticated things. So I have now once recently somehow afforded the Escoffier and looked in there and then quickly closed it again. So for those who don't know: That's just the classic French cuisine from the 19th century and it's crazy elaborate, the stuff that's in there. And because I not only like to cook, but also have two small children, I don't really have the time for it, but it's actually my great role model and the one thing I'd like to be able to do more of would actually be French cuisine, but, as I said, I don't have the time.

Benjamin Kolb [00:31:37]:

Yes, I can only recommend the standard work by Paul Bocuse. Completely without pictures and the recipes always start with where he buys the animal at all on the market, and then also how it actually comes to the individual parts that he cooks. So quite exciting actually, what one makes there in the kitchen actually so everything with the different parts of an animal. I can only recommend it.

Peter Dyllick-Brenzinger [00:32:00]:

Yes.

Benjamin Kolb [00:32:00]:

Paul Bocuse, the standard work.

Peter Dyllick-Brenzinger [00:32:02]:

Yes, that's just cool. I also find it totally desirable to always have a broth on the stove, for example, but I think that if you don't cook at Michelin or restaurant level, it's also difficult to always have a broth there. And it's also exciting to have these recipes that start by preparing a stock for half a day or a day. Yes.

Christian Kallenberg [00:32:26]:

Peter, last question from my side. What I was still wondering when I looked a little bit: Who are you anyway? What do you do? I read on your Twitter profile that you love the desert. And now I asked myself: Is that meant in a figurative sense? So does Peter like to send people into the desert, or is that -

Peter Dyllick-Brenzinger [00:32:42]:

No, for God's sake. No, specifically the desert. So I'm super happy to be in the desert, unfortunately too rarely, so now Corona times. I like that there is so little going on, actually. So it's quiet and there's a lot of space and vastness and usually really great light, because deserts are classically relatively warm places. So I've never actually been in an ice desert, but I've been in all kinds of other deserts. No, and I just really like that somehow. So I think that has a lot to do with the light, actually, and also with the reduced - let's say it euphemistically - vegetation.

Benjamin Kolb [00:33:24]:

Yes, you probably need that as a counterbalance, that sometimes there's nothing going on at all, if you otherwise spend the day working with the BILD editorial team, don't you?

Peter Dyllick-Brenzinger [00:33:31]:

You say that now. But it has to be said that most of them are better than their reputation.

Benjamin Kolb [00:33:38]:

Yes. Do you do other things as a hobby as a counterbalance to the rather exciting work, I guess?

Peter Dyllick-Brenzinger [00:33:47]:

Yes, the classics. Of course I like to read, I like to drink wine - well, not too much, don't worry, but the things are a bit too expensive for that. No, I like that kind of thing. And actually I like to meet with friends, but that's very difficult at the moment. And I also find it - as I can see right now - incredibly exhausting. Well, I have some friends who regularly do something with videocall in the evenings, but when I've been on videocall for seven or eight hours a day - I don't know about you - then, to be honest, I don't want to play any games on Zoom or anything like that.  

Christian Kallenberg [00:34:32]:

Great. Peter, so we really appreciate that despite all these video calls and conferences, you still took the time to talk to us here now. Thank you very, very much for that. I found it very enlightening, learned a lot about actionable analytics. Benni probably already knew everything. But, I don't know, Benni, what did you think of it?

Benjamin Kolb [00:34:55]:

I found it very, very exciting. Of course, I didn't know everything and it's actually good to hear first-hand from those who provide these tools to such editorial departments what's actually behind it. So many, many thanks for the insights. I had a lot of fun and also many thanks that you have actually now also taken the time after a probably exhausting working day already at the computer.

Peter Dyllick-Brenzinger [00:35:16]:

Yes, thank you very much for your interest in this in any case and for the invitation. In any case, I was also very pleased.

Christian Kallenberg [00:35:21]:

Great and see you soon.

Benjamin Kolb [00:35:24]:

See you soon.

Peter Dyllick-Brenzinger [00:35:24]:

That's right, see you soon.

Benjamin Kolb [00:35:26]:

Exactly.

‍

‍