Have you ever heard that thing that jazz aficionados say about their favorite musicians? “It’s not the notes he plays, it’s the notes he doesn’t play.” It’s one of those cliche-because-it’s-true cliches, because music is about more than the notes you hear. It’s the rhythm, too, which can often be heard in the sudden stops and starts between the notes. It’s like that with your data, and yes, I am drawing a comparison between the propulsive swing of improvised bebop and the ones and zeroes that create the collective record of your business. The longer you stay open, the more data you’re accumulating. But it’s not always about how you play the data you have, it’s how you play the data you don’t have. Confused? That’s by design. Before I unpack that statement, though, let’s first look at a real world example of what I’m talking about–which will make it easier to understand as well as bloat my word count to meet quota.
Not too long ago, the world learned the extent to which the United States surveillance program collected data on its own citizens’ cell phone use. Touchy subject, I know, but stick with me; it serves my point and is not an attempt to rage against an unwarranted surveillance state. Not where they can read it, anyway. Ahem. At the time, citizens were reassured that their calls weren’t being monitored, their text messages weren’t being read or stored, and their nude selfies were already on the hard drive of a 23-year-old named Eric and they really should be more careful about that. What the NSA was collecting was metadata–that is, the data about the data. It’s the kind of stuff you’d see on your monthly bill: phone numbers you called, numbers you received calls from, length of call, number of texts, and any other non personal kinds of information. Essentially, it was transactional data, the kind you might find on something like, say, a point of sale system. But there’s tons of information that can be massaged out of that data, if you know how to play it.
A group of researchers at MIT did just that, showing how they could predict personality types using nothing but cellphone metadata. By examining such seemingly innocuous data points, and the space between those points–
- number of phone calls, and a breakdown of the number initiated vs. received
- comparing the total # of contacts to the frequency of interaction with each one
- average time-to-respond to incoming text messages
- the regularity of any given calling routine (i.e. how often you call your significant other)
- “radius of gyration,” which uses location data to see how much ground a given person covers every day (and whether they return to the same spot each day or night).
–they were able to demonstrate correlations between user behaviours and five different predefined personality types. From there, it’s a small leap to an algorithm that could be fed all this data and have it alert you to everyone who turns out to be, let’s say, a “neurotic” type with a just the right amount of “openness” to be dangerous. That’s a long way from the info on your phone bill, but it can be done.
But enough about dystopia, let’s talk about your business. Your POS is storing a bunch of data, maybe even sharing it out with other apps. Your sales revenue, labor costs, and inventory expenses go to your accounting software; your customers, too. Their contact info is shared with a marketing app. You’ve got very specific uses for that data. But your POS has a reporting function for a reason, and you can filter those reports based on any number of criteria to identify trends that you can capitalize on: you can see when certain products sell and when they don’t; get an idea of which items tend to be bought together; compare your revenue with your labor costs for any given time period; track performance of promotions; identify all customers named Bruce who show up on Wednesdays and order Disco Fries, extra gravy. You can go further by adding tags to your products; you can create tags to describe anything that would be helpful for you to know. For instance, tag items as “vegan” or “gluten-free,” and when you’ve identified the people who buy those things exclusively, don’t invite them to Shepherd’s Pie night. Or parties: don’t invite them to parties, either. As with any jazz musician or data collecting police state, the shape of the data isn’t solely determined by what you have. Play the notes right, and you start hearing things you didn’t know were there.