In the last week, I’ve received a hatful of emails, all personally addressed, all brimming with enthusiasm, and all utterly irrelevant. An invitation to an obesity conference in Scotland. An exhortation to "learn to become a gladiator in Rome". A jaunty request for me to join a webinar where I can "swap HTML migration notes" with fellow programmers. I’ve even been invited to an "early risers networking event for under-35s".
What unites these emails isn’t just their irrelevance to me. Paradoxically, it is also the fact that they will all have been designed to be hyper-relevant to me and me alone. Indeed, they will have been very specifically "personalised" on my behalf.
Allow me to unravel this apparent contradiction.
Being nearly right is much worse than not being right at all.
When each of the big-data machines selected me, I know what the client will have been told. The argument will have been something like this: "What we do is mine your data to produce contextually relevant, hyper-targeted communication that is personalised to the interests and profile of your audience."
And, to an extent, the promise was kept. I have researched the subject of obesity, but I have no intention of going to a Scottish conference on the subject. I have recently been on holiday to Italy, but I do not want to become a gladiator. I do try to keep up with online developments, but I am not a programmer. And I was once under 35, but that was a while ago.
When in doubt, leave the detail out
The point is, being nearly right is much worse than not being right at all. It’s like calling a Yorkshireman a Lancastrian, just because the counties are adjacent. It’s like the bloke in front of me in Starbucks who got "Pies" written on his cup when it should have been "Piers" (he clearly wasn’t from either Yorkshire or Lancashire).
When in doubt, leave the detail out. But big data doesn’t. It can’t help itself. It assumes that, because it has a match on a number of variables, then this will be sufficient. But the other variables matter, especially elusive variables like Yorkshireness, and it is often these nuanced variables that discriminate most.
So even if the big data has personalised to 90% accuracy, it may still be 100% off. It’s worth reminding ourselves that the difference between a man and a woman is determined by only half of
one of our 23 chromosomes.
Big data can undoubtedly help build relationships, but it is the servant to this cause, not the master.
The trick with big data, therefore, is to make it small. It is not a god, so don’t worship it falsely. Distil and compress it until it tells you something concrete, substantial and discriminating.
In all likelihood, it won’t see into my soul. It won’t have "personalised" anything. It won’t really know a single thing about my "passion points" or my real interests. It may, however, be able to tell programmatically whether it is serving an ad to someone on Android or iOS, and you can immediately appreciate how valuable that data might be.
But it isn’t big. It’s tiny and it’s binary. And it’s all the better for it. It could make the difference between a wasted piece of copy and a relevant piece of copy. But that piece of copy is not "personalised" in the sense that we sometimes get gulled into believing. It just features a Samsung rather than an iPhone.
The risk of personal
The reason I think this distinction is important is highlighted in the title above. There seems to me a danger that we will see data-driven targeting as "personalised", and that is only a short hop from thinking it’s "personal". And from here, we might then see it as the best way to build stronger customer relationships. This would be a mistake.
Customer relationships are built from a genuine personal touch combined with creative communication and the ability of technology to develop better, more convenient, more intuitive customer experiences. The technological leaps that big data makes possible are seldom about understanding me better, they are more often about serving me better.
They don’t personalise, they customise; for instance, the easyJet app that not only tells me my flight gate as quickly as the departure board, but also gives me a map and the time it will take me to cover the distance.
Big data can undoubtedly help build relationships, but it is the servant to this cause, not the master. Indeed, too often, big data can be used mechanistically to depersonalise the communications process and reduce the role of the advertiser to that of a shill.
Moreover – and pardon the pun – if you’re a big dater, you’re probably not that good at relationships in the first place.