I have looked very much forward to bringing this post. Florian was my early teacher and the first person to open my eyes to how sophisticated digital marketing can be and he is still one of the persons, if not the person, who has influenced my thinking on digital the most.

To those unfamiliar with Florian, his impressive background spans back to the internet heydays of the late 90s. He’s the co-founder and managing director of the Berlin-based early-stage investor and company builder Project A; an angel investor in over 40 companies; a founding Managing Director of Rocket Internet (and through that, he was an instrumental co-CMO at Zalando); and not least, the co-founder of JustBooks/AbeBooks (acquired by Amazon). If I had to describe Florian in a single line, it would be that he’s a guru when it comes to automation, CRM and BI, among other things.

This is the first part of a two-part interview. Read the second part on the future of digital marketing and brand-building here.

Stefan: When you look back at your 15 years in the internet sphere, how would you describe what you have been doing?

Florian: Whether, it’s been on the entrepreneurial and operational side or more on the investor side, the common denominator has always somehow been related to customer acquisition, retention, customer behaviour, and trying to make that as good and systematically replicable as possible. It has always been somehow consumer-related businesses. That’s always been my interest.

I have always mainly worked in an entrepreneurial setting either by funding start-ups myself or by trying to help people to excel in this space.  So far, it has worked more often than not. It has been quite exciting. That’s what I do.

Of all the projects, you’ve been involved in, what would you consider the most successful from a marketing perspective and what was your role?

Zalando is by far the most successful one. That’s fair to say.

I would call myself a Co-CMO. There has always been somebody responsible organizationally or hierarchically for the marketing field, but I always spent roughly three days a week there trying to help that person find the right kind of marketing or CRM approach for the business. I think Co-CMO fits quite nicely.

The beginning of marketing automation at Zalando

How would you describe the campaigns and processes you set up to promote Zalando?

What I am the proudest of and also the unique part about Zalando was the industrialized and very data-informed or data-driven cross-channel approach. We were trying to encompass the customer journey across a wide range of channels and boil them down to the same metrics and optimization logic. To me, designing this machine has been a much bigger success than just setting up a specific campaign.

We’ve not been so creative on the actual content side of things. I mean, they have good content and TV commercials etcetera, but I did not come up with that. It’s not my kind of thing.

While it was not a very creative exercise in the sense of the message, the creativity was in how we designed the processes and the way we organized how people worked within these processes. I think that has been the biggest success and probably what we have been particularly good at. That’s not only Zalando, but it also applies to Rocket Internet and what we do today.

Zalando has been a prime example of that thinking: about how IT and data can actually make marketing more efficient and better. Often times, people in the marketing organization are not very data-driven and not very IT-driven. That has been the core achievement at Zalando for marketing and CRM. I wouldn’t say we had maximal usage of IT and data, but a solid usage of the possibilities that IT systems and data offered.

Where did that creativity come from? Back in those days, online marketing was not as developed as it is today, and most marketers would probably start thinking of creative messages in a traditional sense. Why do you think you thought differently back then?

It was because of the scarcities at Rocket very early on. The real scarcity is the experts on doing things like building a data warehouse or making excellent SEO campaigns. Those people are very scarce and, unfortunately, they mostly don’t get the credit they actually deserve.

Back then, what drove me was how we could scale the few very good people we had. It’s like in IT where you have an IT architect. You build a system and everybody uses the same architecture. I would also use the term “architect” in marketing. Very few people can design a good marketing machine or system and have good enough understanding of the creative, tech and data part involved to do this.

ProjectAInsightsBlogThe whole idea was really that the only way we could scale the Rocket type of companies was by scaling the few architects we had. You can essentially run it with very few people who know what they are doing. If you look at a Zalando type organization, they probably have 400 people in the marketing team. They probably still do. But they likely have only maximum 10 or 15 people in this organization with superior prior experience.

Essentially, the basis for getting people up to speed at these organizations has been a very solid set of processes and systems to work by. Rocket Internet could be a lot better, and they have got a lot better, by valuing people a lot more.

How has this insight changed the way you engage with businesses now compared to the past?

I think we give a lot more credit to experts. At Project A, the internal power balance of experts and the McKinseys and business-focussed people, or what you’d call “triple-A guys”, is much more shifted towards the experts.

Another outcome of this thinking is to say, “well, infrastructure is not just a tool; it’s a very attractive field of investment that can be a lot more capital efficient compared to investing in other parts of the business”. We now invest quite a lot in infrastructure. When you look at the different fields of investment, infrastructure is the single biggest one.

How do you then find these experts?

It’s incredibly difficult to find these people, so we have to teach them ourselves.

The most difficult part is getting the first 3-4. You don’t need more than that. The biggest privilege of a company like yours or ours is that if you have a few people who know how to do this, then you can focus on finding people with the potential and on teaching them. It is a lot easier than finding somebody who works this way with 10 years of experience because there are very few organizations that teach you this way of working and thinking. Two, three or four is enough and you can teach the rest.

Designing the user journey

You touched upon that the creativity was in the process rather than the message. What has been some of the processes that made the biggest difference in the past?

2015-09-10_1759Understanding TV advertising on performance-like metrics has been a big step. It enables you to buy TV and have an, at least somehow, a similar set of KPIs to your online marketing. That has been instrumental.

We’re pretty good now at assessing the direct and indirect effects of TV in terms of how many customers come directly from TV and the later effect. However, we don’t really measure the quality yet and we don’t really understand what’s happening inside people’s heads and why certain things are successful. Comparing it to medical diagnosis, it’s like working on the symptoms but not really understanding the cause and effect relationships. That’s especially the case when it comes to branding.

People always say, “Offline advertising doesn’t work. It’s crap blah-blah-blah…” From today’s thinking, you might say that the old offline companies throw money out the window because they’re doing so much offline. In the end, many offline companies have often successfully built a brand before online was an option. They probably did not do it in the most efficient way, but they were nevertheless successful.

Having a better understanding of how brand-building works and how brands are created on a less fluffy level, is something that I would see as a great breakthrough. I don’t really know how it would work but that is something I would definitely appreciate or see as a great benefit from.

Another big area that has been instrumental in being able to predict lifetime values by introducing predictive analytics. We can now say that a campaign makes sense on a prospective lifetime basis early in the process or campaign. We have been quite good at that at Zalando. However, it’s also a scale game because you need quite a lot of resources to do it. You need somebody just focusing on this. It’s not that easy conceptually and there’s still a lot of stuff that we could do.

Finally, the understanding of the customer journey has been instrumental. We now have a good feel for it and have a much better understanding of how many contact points people actually need and things like that. But there is a part that we haven’t managed to understand yet, and that’s still a challenge for us. The next big step is going from understanding the customer journey to designing the customer journey. If you understand that the average shoe customer needs four visits, then you should know the optimal kind of mixture of messages, creatives and type of contact points that you need to show to a person across different channels to make the ROI, on the overall journey, the highest.

Even though we have an attribution- and lifetime-type thinking, we haven’t yet really implemented an optimization effort across the entire customer journey. We attribute, but we still optimize every channel individually. You take the effect of every channel into account, but you consider it as a given, rather than an object that you can actively manage.

Today you have one coherent logic across the customer journey from the first contact to first purchase and can target people at different stages. However, if you see a journey that continues with repeat purchases etcetera and if you are able to manage this in a better way, then that would be a concept that could be an equal breakthrough to some of the breakthroughs I mentioned earlier like attribution and customer lifetime value.

That will be another big step in marketing efficiency and effectiveness. Once you are able to design customer journeys, real-time advertising will be an elementary pillar. You need though a Zalando type venture to do this as the infrastructure and intelligence that you’ll need for this has a high fixed cost. It doesn’t make sense for businesses where you plan to spend a million dollars a year on marketing. There are some businesses now where it makes sense to implement this thinking and integrate the CRM and customer acquisition perspective into one.

We’ve spoken a lot already about automation. What are some of the general trends that you think will dominate the next months and years?

The whole social advertising area is becoming incredibly interesting. We see most innovation right now around Facebook, Pinterest and Twitter, not only on the social media part of it but also in the advertising products. It wasn’t clear to me a few years ago that in essence, Facebook is in, at least, as good a position as Google to do marketing. You’ll see some business actually shifting budget away from Google to Facebook.

I’m sure Pinterest will be the next thing because traffic coming from Pinterest is great for e-commerce or for any type of business that relates somehow to Pinterest. That’s an entirely new field. You can position yourself again as a very innovative advertiser by simply understanding that the possibilities you have on a Facebook-type tech stack. I’m sure the same will go for Pinterest.

The other thing I find very interesting and that very few people have yet understood is the whole usage of video advertising. A big shift will be away from TV towards YouTube or other video platforms and types of video advertising. That is where the customer journey aspect in the sense of designing customer journeys becomes much more relevant. Video advertising will give you the possibility to be much more sophisticated on the messaging you can get across. This allows you to design customer journeys actively. For me, it’s still a relatively early stage compared to the marketing dollars that potentially will go in there 5 to 10 years from now.

These are probably the two things I find interesting and then certainly mobile. Mobile advertising outside of Google and Facebook is still quite nascent and there’s lots of traffic around. Being able to benefit from this as an advertiser is still a big opportunity. I’m always looking for the new fields where I can get a competitive advantage if I do it more cleverly or more systematically.

Connecting the dots: Towards a better application of data

What are some of the trends you see inside data application, data analytics, and BI in general at the moment?

For me, the key challenge is how you can connect data to action. Trying to figure out what’s really the data I should look at. It’s often not so much about the big data stuff. Many people are talking about big data. My notion is that most people still have many problems identifying what is the relevant data out of the structured data like customer data and transaction data. There’s so much stuff around that’s not big data – structured data by definition is not big data – so, in essence, they’re not even at the big data stage. The first step and a big trend right now are to get all the data that you have—all the structured data – and making it available in one data repository that’s truly connected and usable. Analysing data, making it available, and then making it actionable. Then, after doing this, you can think about things like how you can integrate weather data into something like that. Well, then you have real big data, but not so much right now.

What are the challenges in data application and BI right now in your perspective?

For the companies that are successfully more sophisticated than others, it’s really about how they apply data. Just one example in the context of Zalando was saying, “Now we have attribution sorted out, but then it’s really about managing the customer journey.” That’s an application question. For the more sophisticated people, the problem is more about making use of it. Most people still have problems with collection and aggregation and not so much of a problem with analysis and making it actionable. That’s at least my feeling.

The challenge is that you have so few people who can architect this. People don’t really know where to start, because so few people have an overview of how to approach this problem.

Over time, more and more people will figure it out, but right now, people are still talking about their Hadoop clusters and talking about all the things you can do with it. But if you have 18 data warehouses, then a Hadoop cluster won’t solve that problem. The first problem to solve is how to reduce the number of data reports. Most people are not even at the stage of making it usable, so there is no need to talk about Hadoop. Hadoop is not your problem. It’s a nice buzzword, but it is not solving your lowest hanging fruit-type of problem.


Interesting. If we go to the acquisition part and talk about performance marketing –is that a commoditized product by now or is there still an opportunity to create an edge?

Yes. A year or two or ago I would have said it’s getting commoditized, but now with all the new platforms coming in like Pinterest and also still Facebook, I’m not sure any more. Very few companies use the full potential of Facebook, Instagram, Pinterest, et cetera. My feeling is that the possibility to make an edge there has improved over the last year or so with new platforms getting more reach or significant reach beyond Google.

Sometimes it takes some time to mature. If you compare the performance that Facebook generated for advertising clients 2 years ago compared today, the platform today is more scalable but it’s also more complex. You shouldn’t underestimate the time it takes to reach a Google-like level of sophistication to produce performance. The whole progress that Facebook has made through the custom audience and things like that just take some time. Because of the data they have, they’re probably one of the few platforms that can actually deliver performance in a scalable way.

CRM is the new SEO

In the context of acquisition and retention, I would like you to talk a bit about a passion of yours –CRM. Without risking guiding you too much in one direction or another, I would like to hear your take on where you see CRM going and what the future looks like for CRM.

I have posted something recently where I said that CRM is the new SEO. I think that’s true, in a way.

When we started at Rocket in 2007-2008, the idea was “okay we do performance marketing and then, if it doesn’t produce any ROI, we just keep generating growth. Then SEO kicks in after 1 or 2 years, and then the whole model has a chance to become profitable”. At least it has a chance to become profitable.

That sounds like a very cynical approach, betting it all on SEO to kick in?

It was okay in those days. I mean; you are on your way, right? It’s a process and not just that you sit there. That was the original approach and in a similar, but much more scalable and systematic way, we applied it to the CRM side. Today, SEO is very hard. It has become increasingly difficult and can be taken away at any point with new updates. The underlying risk profile is horrible. But if you think about it in terms of CRM, it’s really cool because it’s not visible from the outside. Nobody knows what data you have exactly, why you are giving what message to which kind of customer and at what point in time. It’s very hard to reverse engineer unlike SEO, SEM, or retargeting. All of that stuff is, not necessarily easy to reverse engineer, but much of what you do is visible to the outside world at least. That’s not the case with CRM. It’s very hard to say which kind of clusters you build and why you do that. It’s very hard to reverse engineer.

I like CRM so much because there’s a lot of room to build a competitive edge. That’s what it takes and that’s why we invest. That’s a similar thinking as we said earlier with data collection part, preparing CRM as early as possible. One should start early because if you do it right then it can give you a competitive edge and shoot you into profitability.

What are some of the trends that you think will affect the future of CRM?

A trend that we are working on and I would say probably only a few people are doing yet is the Facebook custom audience and, with mobile apps around, push messages. You have more ways to reach customers and you can theoretically do CRM via real time advertising as well.

By targeting existing customers via RTA, not just retargeting, I think you have completely new possibilities for CRM. It’s really about finding the best way to reactivate or keep customers active and orchestrate this: which kind of message through which kind of channel. We’re currently working on a tool that will help you orchestrate the CRM approach on existing customers or depending on which kind of customer segment they’re in, based on the behaviour they’ve shown. I think that will be a great way to become more profitable and better in terms of your business model. Again, the cool thing about is that it’s very hard to reverse engineer it from the outside.

This is the first of a two-part interview. You can read the second part about the future of digital marketing and brand building, here.

Photo: Vimeo.com


What have been the most successful ways you have automated your marketing?

34 thoughts on “Florian Heinemann on Lessons from Building the Zalando Marketing Machine and Marketing Automation

Comments are closed.