John Henry, Analytics Expert

I’ve got an old friend from the neighborhood who’s in the retail business. He’s a franchisee for a national chain and a real old-fashioned store manager. After 25 years, he still works the morning shift most weekdays, and he can tell you the name and maybe even the occupation of most of his rush hour coffee customers. He also has a pretty good idea of what they want to buy and what they’re willing to pay for it.

So the last thing my friend needs, as he often tells me, is a bunch of clueless, wet-behind-the-ears planners from corporate headquarters telling him what he should sell and, especially, at what price he should sell it. (Over the years, incidentally, he’s been almost always accurate in predicting which national product initiatives will succeed and which will be laughable failures.)

The last time we talked, he expressed concern about his daughter, who was just about to graduate with a “worthless major” that offered no job opportunities, no career potential and very likely no means of paying back her student loans. When I asked what that major was, I expected to hear something like “Latin” or “philosophy” or (heaven forbid) “journalism.” But no, her dead-in-the-water career choice was “marketing analytics." 

Fortunately, I was able to reassure him that her prospects weren’t nearly as bleak as he imagined. Even though he operated solely on experience, instinct, common sense and, yes, a fair amount of store-level POS data, the rest of the retail industry was moving toward much deeper analytics strategies, I explained. I even suggested he invite his daughter over to take a look at the sales reports some time. Who knows, she might find something interesting lurking in the data. (She landed a pretty good job soon after, by the way.)

On the other hand, maybe the current surge in demand for marketing analytics wonks will be short-lived, now that machine learning is poised to revolutionize the business world. I recently spoke with an executive at an artificial intelligence-based research shop who said that some clients are now asking them to cut right to the chase: They no longer have the time or even the desire to know why they need to execute a specific campaign; they now just want to be told exactly what to do.

Without getting too deeply into debate, I think it’s safe to say that the ideal course of future action for consumer product manufacturers lies somewhere in the middle of these two viewpoints — in a place where old-fashioned “humans” are still allowed to make the final decisions based on knowledge and experience, but where technology-driven algorithms do the heavy sifting through ever-expanding sets of databases to uncover actionable insights that would not have been “humanly possible” to find.

That’s a phrase I heard several times in April at the Retail & Consumer Goods Analytics Summit, the annual conference conducted jointly by Consumer Goods Technology and RIS (Retail Info Systems). I also heard many examples of exactly how “inhumanly” algorithms can perform. Ken Jennings may have held his own against IBM Watson in a game of “Jeopardy!,” but he’d be left in the dust if they went head-to-head in a demand forecasting competition.

Here’s my favorite example from the conference: A spike in the number of Google searches for “morning sickness” is a leading indicator of increased sales for hand and body lotion seven months later. Now my franchisee friend might be able to tell you how many of his regular customers are pregnant and about when they’re due, but would he ever come up with a fact like that on his own? I’m guessing no. (The data point, by the way, comes courtesy of analytics software company Prevedere.)

As an industry, we now have some pretty impressive analytics capabilities at our fingertips. And we’re all in agreement that building these capabilities is critical to future success. But we can’t expect these tools to do all the work for us. We can’t sit back and let the algorithms handle all the customer interaction and make all the plans. There still needs to be some human involvement.

RCAS speaker Sandeep Dadlani, chief digital officer at Mars Inc., suggested one way to divide the workload. AI should be used to solve our business problems so that humans can “free ourselves to find the next problem.”

I’m not sure if my friend is going to buy into that idea. But it might work for the rest of us.