insights! #97: How well do you really know your customers? The role of data-driven decision-making in e-commerce

Sonja Fuhrmann:

Today we're talking about data-driven decision making in e-commerce, Joubin. Why is it more important than ever for companies not only to have data, but also to understand it?

Joubin Rahimi:

Yes, and to make decisions on that basis too.

Sonja Fuhrmann:

Of course, makes sense.

Joubin Rahimi:

Exactly, hugely, hugely important. For one thing, we now have data at all. We didn't have that before. So, so much used to be a gut decision. And in the VUCA world we're in, so very volatile and lots of change, that gut decision is often shaped by experience gathered over years and decades. But data can illuminate much shorter timeframes, and that lets you make far better decisions. That's why this focus on data-driven decision making is hugely, hugely valuable, because companies can simply make better operational decisions. Strategically it's always difficult, but operational decisions are much, much better.

Sonja Fuhrmann:

But doesn't that overvalue numbers and data, at the expense of creativity and human judgement?

Joubin Rahimi:

Now that's the really hard question: where are you actually still creative? We all work for one another in this world, and you have to fit into that gear mechanism. And the more efficiently I do that, the faster I do that, the more and higher the value I generate. And based on data, I can do that much faster. And by the way, experience shows that younger people in particular are much more numbers-savvy than older people, who tend to rely more on experience. And with a lot of operational decisions, it turns out the numbers prove you right.

Sonja Fuhrmann:

Example?

Joubin Rahimi:

For example, which advert performs better or which advertising campaigns should I run? That's certainly a topic in e-commerce. Although you shouldn't be fooled here either — if I run adverts, say on LinkedIn, Instagram plus Adverts, and I look at how the funnel went, how the prospect became a customer, and I measure it incorrectly, for example last-click wins, that means the last click is credited as the advertising measure that worked. That's of course wrong, because often it's an interplay of many. But that's exactly what it shows: if I no longer have that, I can steer these advertising measures much better and perhaps know that a trade fair visit is still important for me as a company, or lots of videos, like we're doing today. But hardly anyone clicks on them, yet people respond to other topics. But it only works together and for one another, not on its own.

Sonja Fuhrmann:

Okay, but doesn't that lead to an endless arms race for data?

Joubin Rahimi:

Absolutely.

Sonja Fuhrmann:

Are we already right in the middle of it?

Joubin Rahimi:

We're right in the middle of it, and we collect a huge amount of data. And then there's the question of consent, for example whether you can allow these marketing cookies. So how do you handle it, do you say no? I always say yes as a rule in that context, because I think: okay, I also want to be served properly. I spend a lot too, and it really doesn't cost me anything serious in that context. It's the case that of course we collect data at a great many points and aggregate things from it. That should be done in a good sense. You can of course also view and present it negatively, but de facto it's an arms race of data and data collection.

Sonja Fuhrmann:

But you also have to understand the data, you just said. Data analytics is the keyword. Why is that so important for success in e-commerce?

Joubin Rahimi:

If I want a good question answered, I need a good analysis. The nice thing is that analyses are now much faster, and I have tools that quickly deliver certain analyses or even spot patterns I never asked about. That's why it's mega valuable to have a good data scientist and good data analyst in your own organisation, someone who can properly interpret the data for these domains — there are, after all, very different data spaces — and who can also ask the right questions and develop the right models.

Sonja Fuhrmann:

What exactly should you capture and analyse?

Joubin Rahimi:

Ultimately, capture as much as possible. Everything that's digitally capturable should be captured. What about data protection? I don't know if that question is still coming, but straight off: if you anonymise it properly, in other words store it in a way that can't be reconstructed, that of course helps enormously in recognising general behavioural patterns. Plus, in the places where I'd say individual interactions make total sense. For example, I'm someone who quite likes buying Scandinavian furniture, so simply in the Scandinavian style. There I know I don't need to suggest certain things quite so often. Or, on the other hand, thinking further ahead, if the person says: I'd quite like to try a different style, and I know this person has always bought Scandinavian furniture, then you don't even need to ask about that any more and can suggest other things that might, purely statistically speaking, be the next stage of evolution in that taste-related area. So on the one hand I want to go in the direction of: what can you do with it, and how much data should you capture? I'd say as much as possible, but the evaluation is the question, and that's complex. It's complex and probably also very expensive.

Sonja Fuhrmann:

Who does it pay off for?

Joubin Rahimi:

I think by now it's actually not that expensive any more. So everyone should engage with it, whether small or large. This topic of data-driven analytics, data-driven decision-making is interesting for every one of you. The question is just how deep I go with it, and how much of my own technology I use for it, and how much data I have. Because the cost factor is the amount of data. Do I have gigabytes or petabytes? That's a big cost factor, simply in terms of storage, but also in processing the data. But if I'm small, I can also start small, and in that sense it's a mindset question from my point of view, whether I open myself up to it or whether I say I trust my gut feeling.

Sonja Fuhrmann:

Where do you see the challenges for companies?

Joubin Rahimi:

They differ depending on the level. Small companies of course have the challenge that only a few people need to get very broadly into this topic, and often need to be statistically versed too, or rather able to read this data. If you master that, that's already great, but it is their main challenge. The larger the organisation gets, the more important the question of data governance becomes. So who has access to the data? Not everyone should be able to see all of a customer's data. So who has access to the data? Who can model insights? Do you, for example, have one data hub, or do you have several data hubs? And getting that straightened out once, so that you can say: I have a strategy for how I handle data, how I collect it, what I collect, and what is particularly worth protecting, especially in relation to personalisation and to safeguarding personal data. And how can I create a service where many people have access to this data without having to call IT? That's really important. Even though we do a lot with IT, IT then mustn't become the bottleneck for providing dashboards and so on. The user has to be able to do that within the framework of data governance. So having data and understanding it — there's no getting around that any more for anyone.

Sonja Fuhrmann:

Thank you very much, Joubin, and thank you all for your interest.

Joubin Rahimi:

Thank you, Sonja. And I'd like to make a proper call to action. Write in and tell us how far along you are with your data-driven decision-making strategy. Are you already doing it? What do you think of it? Or do you even have tips for others and for us? Please leave them in the comments.