Some things don’t have a single truth: a bird in the hand is worth two in the bush, but you have to speculate to accumulate. Some things do: bad data in, bad data out. That’s why we should all be taking every lesson we’ve learned from using CRM and applying them to AI fast-like-now.
Join me in a halcyon past, at a long lunch on a leather banquette, opening a second bottle and trying to figure out if you’re the whisky or bourbon sort. Or on the golf course, tactfully ignoring your dropped ball.
As a seller, I keep handwritten notes: prospects and customers, corporate structures, partners and children, birthdays, holidays, hobbies and sports teams. For quick reference, I file these notes in a Rolodex. And when that Rolodex gets nice and fat, it writes its own cheques. Which is where the problem starts.
No business likes its business to walk out the door into someone else’s business. So – flash forward – my business buys a CRM and asks me not to bother with that out-of-date Rolodex anymore.
The problem is, from a seller perspective, the CRM was wrong from the start, maybe because ‘the Rolodex you don’t own’ never really flew as a strapline. I passively resist: I’m hitting my numbers, you don’t want me to walk, and sometimes I don’t put in all the information. Maybe I replace a mobile number with another number for the lolz. Now your CRM contains more local restaurant numbers than TripAdvisor.
I asked a group of 60 sales leaders about this at a recent RevGen Insight summit. “Technology and management drove the evolution of CRM,” they said, “not sellers.” Other popular soundbites: “CRM is about manager benefits, not seller benefits.” “It turns talented sellers into data entry clerks.” And: “have you ever heard a seller say CRM was responsible for a deal?”
“Technology and management drove the evolution of CRM...not sellers.”
If you’re pro-CRM (and most of us are), you argue – quite correctly – that the reason CRM can’t claim a deal is because bad data in equals bad data out. Maybe you think that’s the seller’s fault; maybe you’re right. But is this the hill you want to die on? Because being right isn’t as useful as solving a problem.
And here’s the problem:
Built for the people who do the data entry.
Built for the people who pay for efficiencies.
It doesn’t help me do my job so much as it helps you report on my job.
All of which is quite boring, and we all know, and nowadays there are lots of add-ons and apps that fix these problems, and why have I bothered with this? Because I worry that exactly the same thing is happening with a particular type of AI: the one that most of us will have to use at work.
What Amazon-, Google-, and IBM-sized companies can make AI do by machine learning enormous data sets is amazing. My concern is for the AI that SMEs, our business world’s glorious engine room, will be implementing in the next couple of years. AI animated on smaller and meaner datasets. The sort of AI that quickly becomes racist and sexist.
After all, if a group of Twitter trolls can turn Microsoft’s wholesome-by-design AI chatter bot, Tay, into an unhinged Nicholas-Cage-acts-the-Internet masterclass in just 16 hours, what might happen if you feed an AI your half-baked CRM? That’s a lot of high-level strategy planning based on having the contact details for Pizza 2U.
2019, here’s my rallying cry: let’s get this one right. Let’s build AI processes, systems and technologies that genuinely help the people who do, do. Let’s use it to make people’s jobs easier, less repetitive, more rewarding and fun. Yes, fun. Let’s use it to empower rather than to manage, to make equal parts more effective as more efficient. Or at least let’s not make the same mistakes again. Let’s not build another wall.