As a welcome to 2015, Boston has been gifted … winter.  It is cold out there!  So this morning on my way in I stopped at Starbucks for a grande non-fat extra-hot decaf caramel latte. You know, just a cup of coffee.

I had to say my order three times – once to a barista who was filling in at a cash register, once to a manager who shooed her back to the bar and rang me up, and then again to another cashier who finally charged for my order because the first cash register was out of service.

And that is a mouthful – grande non-fat extra-hot decaf caramel latte.  (Nevermind the stress of saying it in the right order and the shame of accidentally calling it a “medium”)  So by the time I got to the guy who was ringing me up, I just said “grande caramel latte” because those were the items that impacted what they were going to charge me.

I got to thinking – who at Starbucks actually need to know what parts of that order?  How much value is there in their business systems knowing that particular combination is important to me?

I’ve had the same thought at the supermarket, when I’ve bought, say, two pints of ice cream (don’t judge!) and they are different flavors, but the cashier rings them up as two of the same flavor.  I mean, with all the computing power and analytics behind Big Data, don’t Ben and Jerry or at least Stop & Shop care that the same person who buys Chunky Monkey also buys Chubby Hubby, rather than two pints of Chunky Monkey?

And it seems like there are two issues – one is knowing the customer (she buys those flavors together) and the other is inventory (we have two fewer shots of decaf than we had before this transaction).

Maybe there is enough volume at supermarkets and Starbucks that it just doesn’t matter. Maybe Starbucks knows that decaf is under-reported by 15% and rung up as high-test, so they compensate for that.  Or maybe they track stock based on other indicators, not based on transactions.  Maybe two Chunky Monkeys get rung up as often as two Chubby Hubbys.  

That leaves the customer side of it – knowing what customers want what things. And here’s where I really wonder what’s going on.  Something like “extra hot” probably isn’t rung up on the cash register.  (Even if I’m a more fairly compensated Starbucks barista, when it gets busy at the store, I’m going to focus on getting the drinks out, not ringing up all the adjectives).

But wouldn’t Starbucks want to know if everyone who orders decaf also orders extra-hot?  Even selfishly? (Maybe there’s something chemical that’s causing the coffee to come out colder. Or maybe the people who order decaf want it to warm their hands.)  What about a sudden increase in extra-hot requests without a change in weather?  (Maybe there’s something wrong with the machines at that store.)

And milk preference – what if people who order caramel lattes almost always order them with whole milk?  Maybe that means that skim caramel lattes don’t taste as good, so if someone orders that (~$.50 upsell on the caramel) they should be offered whole milk to try to gain them as a syrup-adding customer.  Or maybe it means that people who order whole milk lattes are prone to being open to syrup combinations, so they should be offered a “hazelnut caramel” whatever-a-ccino with an extra syrup upsell.

The truth is, Starbucks is doing pretty well without my crack team analysis.  I just spend so much time reading and learning about Big Data from the technology side that the mechanics of implementing it in retail are a fascinating thought experiment for me.