Interview with data drivers: Michael, part 2

Part 2 of the interview

Click here for part 1

 

What specifically should a client stipulate in an agency contract?

Clearly, the unrestricted right to export the data generated with the company's money. This sounds obvious, but in practice it is not. Many agencies resist this because it disempowers them. So: be sure to secure access to your own data! This data, it is very important, should not be aggregated, i.e. summarized, e.g. per week or so. We are talking about raw data. A company or association must be able to access this data at any time for its own analyses. This does not mean that the company's own data must necessarily be stored on its own computers. Another service provider is sufficient. The main thing is that it's not the computer of the person who has been entrusted with the media funds, i.e. Google, Facebook, agencies, by which I don't mean content or creative agencies.

Now that sounds like in-house data is needed primarily to monitor service providers whose numbers you can't believe.

No, that's just a peripheral aspect, there are enough examples of owner-managed agencies working cleanly, no question about it, we get on very well with them too. The main motive is for companies to push open the door to their own digital value creation through communication. If you consider that 70% of an automotive brand's value creation is linked to its communication and image, and not to its product, then it becomes clear what kind of potential it has. Content and marketing have an insane potential to create value from data. We want to leverage this for companies. To do this, I first have to gain sovereignty over the data, I have to be able to control it as a company, then I also gain transparency over the, sorry, thoroughly corrupt online advertising market and can enter into a data-based competition for the heads and hearts of my target group. Sovereignty and control over one's own data are the absolute prerequisite for creating value from data. This applies to the development of digitized products as well as to communication and marketing. In the case of machines, for example, which use usage data to optimize their own maintenance cycles, this seems immediately obvious to everyone. Why this is not the case in communications and marketing, and why it still takes more than 20 years to convince people to use digital media, is beyond me. The fact that digital marketing has not been able to grow a herb against naïve bullshit, fraud and socially irresponsible progress fairy tales from Twitter kings and surveillance priests for decades surely has something to do with the fact that companies and advertisers have been passive for too long when it comes to data sovereignty. That's finally changing, thankfully.

So how do you create value from data?

Well, in that people or computers use them to make decisions and carry out actions that they would not have done without data. With computers, which are known to be fools, as Peter F. Drucker once said so well, it's quite simple. WDDD - if this, then that. For computers, rules are defined and processed bluntly but highly efficiently. Whether Twitter kings call it "artificial intelligence" or not, it doesn't change the principle. Computers are stupid, but above a certain amount of data, mass turns into class.

Creating value from data with people is our job, but it's more complicated. Because people live and work in groups and they organize themselves along power and hierarchy, even in "agile structures," what used to be called "flat hierarchy." The social environment in companies, like everywhere else, is colored by very personal and very different perspectives and interests. If you want to make data-based decisions, you need few standards and a lot of practice and patience to overcome the resistance that logically comes when you want to standardize.

Surely it shouldn't be a problem to build a marketing analytics service where all the relevant numbers converge?

In rare exceptions, this is really not a problem. There is top management intention and also the necessary pressure from above associated with analytics budgets. But where is that already the case? "Tell me, do we really want to be measured?" Just last week, a customer told me about exactly those words from his boss. And my customer, who had all the wind taken out of his sails, was actually the Head of Analytics. I'm not making this up now. It is often frustrating to see how strongly the striving for power, territorial instincts and vanity dominate everyday corporate life, but that must be the case with Homo Sapiens.

How does this show?

This already starts on the lower floors, in the media silos. There is one colleague who is only responsible for Facebook. Then there's the next colleague who does Twitter and LinkedIn, then an online editor only for web content, the media person who only does branding, and so on and so forth. All these silos are small principalities, often with their own evaluation tools. In such an organization, no one is motivated to make their own area of responsibility transparent and comparable by evaluating data in the same way. Data embodies dominance and power. Why should I give that up? Every company with a local distribution network knows the problem. It took car manufacturers decades to gain access to customer data from local dealers. And whether insurers have managed it by now after 20 years of Internet, who knows, the internal battles were bloody in any case. In most cases, the head offices still do not have access to the data from the markets. If, for example, France is served by an independent national company - no chance. The resistance to persistence is only social, i.e. power-motivated, and it is gigantic. It often takes a new company of one's own to break them.

Okay, let's assume that we succeed in connecting all agency data and breaking down the internal data silos. Then I have a data lake, and I have to educate and train analysis experts. Is that realistic?

The big names manage, with a lot of money and effort, to bring these experts in-house and build up the know-how themselves. But I don't think it's a good idea to have analytics experts in-house. After all, the experts who analyze data in-house are subject to exactly the same social mechanisms as their colleagues. The same struggles for power and influence as the others. If the data experts are in-house experts, then they can't analyze and interpret data neutrally. That is really out of the question. They have to serve internal interests first and then look at the data. What value can be extracted from the data under such circumstances? How can "digital excellence" emerge?

Companies from Hochsauerland or Upper Franconia have it much better. They have no illusions. No, no data science expert will migrate to Bergisches Land to work for a hidden champion. All the data science graduates will stay in their own milieu, in Berlin, Hamburg or other cities. Companies in the provinces - I'm talking about most of the strongest and best companies in this country - hardly have a chance on the job market for data science, sometimes they don't get a single application for their data science job posting. Then there is only one solution: trustworthy external service providers who get access to the company's data, merge this data and drive evaluations in close consultation, which are understood and implemented. This can all be done remotely, via video conference, and you don't have to be on site.

Does that only apply to large medium-sized companies in the provinces?

No, .companion also works on a global level for one of the world's largest automotive suppliers, for example. They have outsourced the entire analysis area to us, even though they also have our expertise and much more in-house. Economically, however, it makes much more sense for the company to use its own data science capacities in its own digital products. That's why data science for marketing and communications should be outsourced. And then they chose us, and we've been doing it for years now(knock on wood). External services help companies to finally concentrate on their core task of content marketing, namely to become better and better in terms of content, interaction and efficiency. Our biggest client, Bosch, shows that this really works. The success of content marketing is outstanding even when you look at it objectively and soberly. This example shows that moderate centralization, intelligent outsourcing, and agile integration of external and internal staff are the right way to go when it comes to analytics. In fact, I think it is the only right way.

But if I hire a service provider to analyze my digital data, haven't I already relinquished sovereignty over my data?

Not if you.have commissioned him to bring together the relevant data for the company and keep it available at all times. That's a huge step for advertisers. For us as a service provider, this is a big responsibility and requires a lot of trust, but no one becomes dependent.

You sometimes call yourselves a data agency. What characterizes a data agency?

A service provider who assists in the evaluation of data must be a fiduciary of the customer. He must be neutral and have no conflicts of interest. You can quickly test whether this is the case by conducting a thought experiment. What happens if I terminate my service provider? Does anyone in the company even notice? If not, the service is of no value, so I get rid of it. Is my data then gone? If yes, there is negligent management and I need better contracts. Do I lose all my standards for analysis and reporting? If yes, then in reality I have no standards because they are not internally anchored. This anchoring is actually our main task, everything else is technology. It only serves to provide very simple and universal analysis standards that everyone can follow.

Data analysis in itself is all well and good, but it is not, after all, the product, not the actual value creation that should take place. How do I prevent myself from analyzing to death?

Data analysis, no matter in which area, has the sole purpose of increasing value creation in a department or a company. Otherwise, it is indeed not only worthless but counterproductive, it is then a waste of time. Data analysis must not be an isolated activity, it must absolutely be integrated: in the standard processes, in the regular meetings and in the strategy development. If you take a quick look at these processes, the different players, situations and interests, it quickly becomes clear how diverse this is, how many totally different issues and perspectives need to be supported by data. This diversity in a marketing organization is almost unmanageable, and new management models are constantly being developed to describe it. Reacting to this chaos with a bouquet of many colorful metrics and dashboards is the normal practice, but it is fundamentally wrong. The opposite is required, radical simplification.

And how do you arrive at simple solutions?

If you want to use data in your organization and start with a differentiated briefing on key performance indicators, you have already lost before you start running. Right at the beginning, it is necessary to define as few key figures as possible against which the performance of marketing and communications is to be measured in the future. As few as possible doesn't mean that you can't look deeper into the numbers if needed. But if you have communicated about a new gas turbine on 120 national landing pages in eight languages and on a dozen social networks, you first need to be able to get an overall picture. That's only possible with a very few KPIs - our three, to be precise.

Does a consistent and comprehensive analysis of the data also require different internal structures?

Quite clearly, yes. If no actions follow from data, the whole thing is worthless. That's why the organizational embedding, interpretation and discussion of data is the all-important thing. This is already happening today, but still mostly in media silos, where specialists talk first about optimizing their Facebook posts, then about improving their Twitter tweets. Website content performance on the same topic is discussed elsewhere, often with other specialists. In such a siloed world, content marketing doesn't stand a chance, and everyone knows it. But why does this fragmentation exist in the first place? From my experience, overly complex "KPI" and "analytics" have contributed to this quite decisively and now that digital media have arrived in the mainstream and maturity phase, now wrong KPI bear the significant responsibility for this fragmentation. If you want to transform communication, you need other KPIs, and that's one way of looking at it.

How can you imagine other internal structures?

At least once a month, everyone who is responsible for content and/or media should sit around the table: the internal team, data analysis, and of course the agency. It is important to talk openly here! It's not about avoiding campaigns that may have gone badly, that's what you need for executive meetings. We are here on a working level, it is about learning from tops and flops. So, for that you need trust and a really open culture of discussion. That's extremely important, it's also about experts from very different fields being able to understand each other at all; the newsroom in Cologne with the product manager from Swabia, with the data analyst from Berlin, with the content editor from Hamburg.

Can a communication strategy be derived from the data alone?

Nonsense, of course not! You also need goals to ask the right questions of the data. Content marketers in companies, often need good facilitation to identify what they want. Their goals aren't just lying around on the shelf and if there's something there, it's last year's strategy presentation, water under the bridge, no longer relevant. So, working out goals is always the first step. Yes, with the goals clarified, I can then ask the right questions of the data oracle. But what about the answers that come back, all those statistics and charts, how am I supposed to understand them?

We always say "data only speaks when you talk about it". This is where the team comes into play. Only the team knows under what conditions the measurement results came about. Only if I know the circumstances do I know whether the super engagement can be explained by targeting or by creation, for example. This is the so-called context knowledge, this is the crucial knowledge for the interpretation of data, the data itself is not enough. No data expert or artificial intelligence can explain why one post went through the roof and the other did not. Was it perhaps because the company's CEO was arrested that morning? Data provides very good clues, but explanations only come together with contextual knowledge from within the company. This expertise will always remain within the company, is reassuring, isn't it?

What might a good start look like for an ideal analysis practice?

For example, you can start with a review of one year of content marketing, with a content audit. We pull all the data into an analysis environment, across all channels mind you, and then we go into a strategy workshop with a prepared dashboard. This is exactly what happens there: we discuss and interpret the data together with the team, we derive insights, recommendations for action and goals for 2019 in open discourse. The audit has the advantage of entering the "new territory" of integrated data analysis in a completed project. At some point, this is clearly over. Later, you can still freely decide whether you want to have integrated analysis across all digital channels permanently and continuously. We recommend that, logically.