Forbes: Big Data - Insights or Illusions?
by Alastair Dryburgh
What is Big Data? When infused with human understanding it can lead to powerful new insights. On the other hand it can be a "truth machine that you can't challenge because it's big data," cementing and intensifying the toxic assumptions in a company. I talk to Christian Madsbjerg of ReD Associates, who gives examples of data's use and abuse.
Christian Madsbjerg: We did a really cool thing lately with computer science people and anthropologists, which are not groups that normally hang out so much together. But they found a common interest, which was science, really, but here we looked at fraud. So fraud, credit card fraud, is one of those topics that scares the hell out of every bank executive. And very little fraud is really detected. So what we did was we infiltrated fraud networks in Queens and in Boston to challenge a big bank on whether they really knew that much. And what we did was we lived with the fraudsters for awhile, and we learned how their business works. How they got into it. We understood what the practices are, who they work with, how they get rid of their stuff and so on. We found 50 hypotheses about their behavior that could then be tracked in the big data sets, but that the banks hadn't even considered. They hadn't even thought about it. I'll give you a couple of examples. We found that they would buy Nike sneakers or trainers, but they would buy the same shoe in many different sizes.
Alastair Dryburgh: Because they're stocking a shop.
Madsbjerg: Exactly, so we crossed the people that buy the same trainer or the same jersey in different sizes with fraudulent behaviour, and the predictive nature of that was massive. And the credit card companies had never thought about that before.
The other one was we found that they don't want any connection between themselves and the places that they buy things online with fraudulent credit cards. We found that what they do is they buy things on amazon or something like that, and then they send it to places that are empty. So we crossed property that's been on the market for a long time, and high volume of packages. And boom, again, big, big predictive power in that. So that was anthropology, the slowest, most high touch intense kind of research with big data analytics, which is probably the most abstract kind of research, but when you combine them that's when you'll have ... we call it contextual analytics, but it's analytics that isn't just machine learning and AI and whatever people are talking about. But it's infused with human understanding.
Dryburgh: This is really interesting, because it seems to me that big data could be a very two-edged sword. On the one hand you can use it in the way that you've described to validate hypotheses that you've arrived at by very subjective, qualitative means. I guess the other alternative is that you can use it simply to provide confirmation for what you already think.
Madsbjerg: Which is what's happening, and with the ethos that we've got a truth machine that you can't challenge because it's big data. So you'll cement and intensify the toxic assumptions you have in the company if you don't use it to challenge and explore, rather than to confirm things you already know.
Dryburgh: Do you have an example of an organisation who is using big data simply to reinforce and cement toxic assumptions?
Madsbjerg: Every single organisation I'm dealing with, some of that is happening. I was in a room with 25 chief data officers. That's a title now. And these were from some of the most sophisticated companies in the world, so financial services, healthcare, and they all said "Yeah, we have a problem with this." And there's a general sinking feeling that big data was oversold. There's nothing wrong with data. It's the way you use it.
Madsbjerg: If you use it to confirm what you already know and not do the painful thing of challenging the assumptions of a company, you can get anything out of it. That's not what you want to do, so there's some of that happening everywhere, but the good leaders and the people that are working honestly with this are the ones saying "Okay, how do we have a process with these data sets?" That's scientific, that's challenging hypotheses rather than proving hypotheses and where we use it to learn.
Dryburgh: Which means probably asking for analysis or setting up experiments actually specifically designed to disconfirm a hypothesis, as the good scientist is supposed to do.
Madsbjerg: Exactly, exactly. And ... scientists would be appalled if they saw how truth is manufactured inside of companies, because purely process-wise, it is dressed-up as science and data, but without the ethos of science, which is one of exploration, understanding, challenging and so on.
Dryburgh: How fascinating. Now I imagine a lot of this you're not able to talk about publicly, but I'm wondering do you have a good example, which you are able to share of this manufacturing of truth? Can you give us something that would help people recognise what really bad practise looks like?
Madsbjerg: Well, a very typical case would be what it looked like at Lego, back in the early 2000s. There was an assumption that kids had no attention span. Hence, we need to make play sets that are designed for that, and we need to make things that are simple and that the kids can build quickly, and that you can't necessarily use for all kinds of other things. And then a new CEO came around and said, "Are we sure about this?" And we looked at all the data, and they'd spent a lot of time on picking the data that showed that kids have short attention spans. So in that case it was levels of ADHD diagnostics in America in particular among boys, and they used those pieces of data to say that they'd been right for the last 10 years, yet it wasn't selling.
Madsbjerg: And then what we did, we went out and we looked at kids and we found that even the kids with ADHD diagnoses had, given the right circumstances, a deep interest in learning in complexity and could play with pen and paper for hours. Or one of the cases I saw in Germany was a kid doing skateboard tricks over and over and over again. That's not short attention span, that's the very opposite. And we just saw that everywhere, and we said, "Well, if that's true, then probably 70% of your portfolio needs to go, because it's built on the assumption that kids' attention span is short and will go even shorter. That was part of the turning around of Lego. Around what, eight, ten years ago. We called it the thinking pollution in the company. There were polluting assumptions about kids that came from not playing around with kids. And not engaging with kids, because it's easier to stay in the office. It's cleaner to not have to go to children's birthdays and get on your knees and play with them. Yet if you're a toy company, you have to get down on your knees and understand children.
Dryburgh: Those assumptions that kids have low attention spans, were they from bogus science, or just anecdotes?
Madsbjerg: Nobody wants to lie with self-transparency in that sense. It's not that they say "Okay, now I'm going to lie." It is layer upon layer of handpicked evidence for a particular assumption that is the easiest one to have right now, because it's the one we have.
Dryburgh: So it's intellectual laziness. Madsbjerg: It is, that's exactly what it is.
Dryburgh: I think insecurity as well, isn't it?
Madsbjerg: Fear of getting fired, fear of challenging things.
Dryburgh: Yes, I think for most people having basic assumptions challenged is really quite scary, isn't it?
This article first appeared on Forbes.com.