Did you know that we humans have 50% of our DNA in common with bananas? I don’t remember where I first stumbled across this little factoid, but it was, as it was probably designed to be, just wacky enough to catch my attention. As far as advancing scientific knowledge, does it do much more us? I don’t think so. This little factoid is interesting, but not valuable.

What makes an insight valuable? My answer: when it can be acted upon. Retailers don’t care much about genetic commonalities with bananas, but they do care an awful lot about consumer data, and they most especially want insights from that data that they can use to influence consumer behavior. In short, they want to rapidly find the insights that are valuable, and discard the ones that are merely interesting. And the difference between the two is summed up in one question: Can I act on this insight? For retailers, the question is maybe even more specific: Can I influence customer behavior based on what this insight tells me?

Paula is writing this week about this same topic. Her take is about what happens when retailers confuse stereotypes with insights. My take here is on big data itself – and why I worry that retailers are collecting a whole lot of information that may ultimately turn out to be interesting, but largely not valuable.

There are two reasons behind this. First, consumers are throwing off more data than ever. This isn’t a story about wearable tech, though that is one component. Actually, this is about how consumers choose to define themselves. Pundits have been talking about the fragmentation of media for years. Trendwatch argues that the more consumers can connect to each other online, the more narrowly they define themselves. If you were, for example, a fan of Boba Fett from Star Wars back in the 1980′s, and you were the only kid in the neighborhood who felt that way, you probably kept that somewhat to yourself. Nowadays, if you love Boba Fett, you don’t have to be lucky to find other people who feel the same way you do – you can just find them online.

This is a different kind of fragmentation than the kind people usually talk about when they talk about media fragmentation – this is more like demographics fragmentation. Another fun example – did you know that Facebook has 58 gender identities? Not relationship statuses, but gender identities. For retailers who are trying to figure out if they should feature the men’s clothing or the women’s clothing as part of their attempt at personalization for consumers shopping there, 58 is a number that makes you pause.

Retailers are trying to keep up. They may not be so focused on trying to assimilate 58 gender identities, but they are increasingly trying to tie together consumers’ loyalty or shopping profiles with their digital and social media ones. And they are trying to collect as much behavioral information about shoppers as they can. Their objective seems to be, the more we know, the more we’ll have something valuable.

The problem is the more data they have, the more difficult it is to separate the things that are valuable from those things that are merely interesting. Let me put it in a product context, because it’s easier to explain. My house is a Coca-Cola house. We do not drink Pepsi. We have never bought Pepsi. We are also not coupon people. If there is a coupon stuck on the label when I buy it, I might redeem it, but even that coupon I might forget. So let’s say one day, as usual, I get the annoying coupon with my receipt that tries to incent me to switch from Coke to Pepsi. And the next day my brother-in-law comes to visit and he drinks Pepsi. Being the accommodating hostess that I am, I buy a 6-pack of Pepsi (nose wrinkled in disgust, but I buy it).

My grocery store doesn’t know that my brother-in-law is visiting. All they see is, she got a coupon, she bought Pepsi. They may not even pay attention to whether or not I redeemed the coupon. Is this information valuable? No. Because if the retailer acted on it, they would be flummoxed that they suddenly again were not able to influence my purchases towards Pepsi, no matter how much the brand paid them to try. The Pepsi data point in my purchase history is interesting, but it’s not valuable. But the reason the retailer doesn’t know this is because they don’t understand anything about why I bought the Pepsi in the first place – which is a whole other story.

Let’s try it in a customer data context. I have actually had a retailer ask me the color of my eyes, and it was a clothing retailer, as part of their loyalty program signup (I don’t think they do this any longer, and I opted not to answer at the time). Now, I can understand that maybe a cosmetics retailer or brand would want to know my eye color. And my eye doctor and optician probably want to know my eye color. And maybe, just maybe, once in my life I have looked at a blouse or a sweater and said “This really makes my eyes stand out ” (I’m being generous here). But I can attest that I have never bought a single item of clothing because of the color of my eyes.

So why the heck did that company want to know my eye color? I assume it was because they thought they could influence my behavior in some way by marketing to me based on the color of my eyes. But if I’ve never considered a purchase in that context, how can they possibly hope to influence my behavior? I’m basically uninfluenced-able based on eye color.

If I did tell that retailer my eye color, that would be a piece of data that retailer now holds about me, has to protect and preserve my privacy in relation to, even though in the end it has absolutely no sway in the purchases I make from that retailer. In short, my hazel eyes are interesting to know, but not in any way valuable. And yet, someone out there wanted to know that information badly enough to ask me for it and create a field to store it – across however many thousands of customers who may or may not have provided that information.

So, if demographic information is misleading, as Paula demonstrates, but digging down into the granular level of demographics and behavior creates more noise than value, why do retailers do it?

I think they have a very strong gut feel that there is value in all of this new behavioral information that is out there – all that big data. And their feeling is, if they collect as much of it as they can, they’ll find something in there that is valuable.

But there is no value – in fact wasted value – in collecting data about things that are merely interesting. And no one can afford to collect all of this data forever. Moore’s law aside (and it has applied to storage too for a long time), this is going to start getting expensive. The more data you have, the more expensive it will be to keep it – and to get value from it as well, because you’ll need powerful analytics to cut through all the noise to find the insights that actually move your business.

Paula says that companies need to get fast at their analytics – the sooner they can get to fine-grained, accurate insights, the more differentiation and competitive advantage they’ll have. And I agree. There are two ways to do that. One, you can amass the most powerful analytics money can buy and force a crunching of the data. Or two, you can do that once, and one time only, with the intention of figuring out which data elements are actually something you can influence. Which data elements lead to actionable insights? They won’t all lead there. The quicker you can pare down the information you collect, the cheaper and faster you can be about collecting it, and the sooner you can get to actionable insights.

But to get there, retailers need to be able to look at all of the indiscriminate customer data they collect today and decide: Is this valuable? Or is it a banana? I think the retailer who can answer that question the soonest will be the one who ultimately wins, even over a competitor with a killer big data strategy.