The Customer Knowledge That Never Made It into a System

Spend any time in an ASA task group conversation right now, and the questions arrive in a predictable form. Can it read our call notes and tell us what happened? Can it pull the accounts that have gone quiet? Can it sit on top of the CRM and surface the things a busy salesperson would miss?

The questions are good ones, and they are almost always about AI. Over the past year, the area that has generated the most interest in those conversations is AI applied to customer data. Members want to know what the new tools can do for their customer relationships.

There is a quieter pattern underneath those questions, and it surfaces the moment you talk with people who have actually built this capability. They tend not to talk about AI at all.

On a recent ASA Embracing the Future podcast, two people who have lived through CRM implementations, Charlie Parham, CEO of the manufacturer's rep firm Pepco Sales, and Scott Stockham, CRO at RepFabric, kept circling back to the same unglamorous thing. Not the software. The data, and specifically the question of whether it exists at all.

Stockham told a story that should sound familiar to anyone who has spent a day in the field. He did a ride-along with a plumbing wholesaler's best salesperson. Forty-five minutes across town to a meeting, a quick conversation, a couple of action items, then forty-five minutes back in the other direction to a similar meeting. A full day of real selling by a good rep. And, in his words, "there weren't a whole lot of notes taken."

The customer knowledge from that day, who wants what, which job is coming, what almost went sideways last time, never left the salesperson's head. None of it entered a system.

Now point an AI tool at that. Ask it to summarize the account, flag the risk, or draft the follow-up. There is nothing for it to read. The intelligence everyone wants to pull out was never captured in the first place.

This is the part that gets skipped. The customer data a distributor would want to act on does not start in a CRM, and it does not start in an AI tool.

Parham put it plainly: that data "has to come from somewhere. It's coming from your ERP or your customer list or another CRM."

For most distributors, the ERP is the system of record. Customers, orders, pricing, and history, it is all in there, and a lot of it is a mess. Duplicate customers. The same contact entered three different ways. Fields that some branches fill in and others leave blank.

Stockham's first move with a new client is the least exciting work there is: cleaning up duplicates. Parham's advice was to start that cleanup before you even shop for a system because, if you start shopping too early "and you don't know what you want and you don't know what you have," nobody can help you.

What makes this worth raising now is the timing. The reflex Stockham warned about, that you can "ship money off to a service provider" and get "a magical solution," is exactly the reflex AI invites. The tool is new and capable, so it is easy to believe it will do the foundational work for you.

It will not.

As Stockham put it, "you can't automate something that doesn't exist."

He was talking about sales process, the steps a tool is supposed to speed up, and the same holds true for the customer data those steps are meant to capture. AI increases the value of clean, captured customer data. It does nothing for data that was never recorded.

That is why the customer intelligence framework ASA's task group has been building puts AI at the end rather than the beginning. The sequence runs from readiness to selection, adoption, using customer intelligence for growth, sharing data with partners, and only then to AI.

The order is not a ranking of importance. It reflects what operators describe, which is that the value compounds slowly.

Parham, fifteen years and several CRM generations into Pepco, described year one as getting people to use the system at all, and year two as getting the organization to use the data. AI sits a long way beyond where the work starts.

None of this is an argument against AI, and it is not a reason to wait on it. The interest is real, and the tools are improving quickly.

But the operators who are further down the road were not framing their experience around AI at all. They talked about clean data, a sales process people actually follow, and the slow work of getting an organization to use what it captures.

That is the foundation any AI tool has to stand on, and it is where interest is currently thinnest. The energy is concentrated at the far end, while the work that makes that end possible sits at the beginning, in the ordinary business of deciding which customer data is worth keeping, capturing it the same way every time, and cleaning up what is already there.

The encouraging part is that the foundational work pays off no matter which tools show up next. A clean customer record, a sales process people actually follow, account knowledge that lives in a system instead of one person's memory, all of it holds value through every new release.

Parham's line about when to begin was aimed at CRM, but it travels. The best time to start was five years ago, and the second-best time is now.

The salesperson two towns over already made the drive. The only question is whether anything he learned made it somewhere a tool, today’s or next year’s, could ever use it.