Will your sales team actually use the AI quoting tool?

The question a dealer's owner asks near the end of an AI quoting conversation is the one that actually decides the project: will the sales team use it? It is usually asked as a worry about people — are they too old for this, too set in their ways, too suspicious of AI. In our experience it is almost never a people problem. It is a design problem, and it is answered before launch by whether the tool removes work from a salesperson's day or adds it.

We build these systems for machinery dealers, and we have watched the same tool succeed on one team and die on another. The difference was not the age of the salespeople or the quality of the training session. It was whether the tool took the copy-paste out of the day — the manual walk from a manufacturer's price list to a Word document — or whether it asked the salesperson to feed a system that gave them nothing back.

AI inside a dealer is not one thing

Start with a fact that surprises owners: their company does not have one relationship with AI. It has several, and they are pulling in opposite directions. At one dealer we work with, a single salesperson had built — on his own money, at home — a whole private AI setup: a server under his desk, automation scripts that prospected and emailed on their own, a chatbot on a personal site that advised customers on which machine to buy, even a synthetic voice reading his mail to him. He was, by any measure, miles ahead of his company's official tooling. In the same building, other people on the same team had never opened ChatGPT.

That gap is the real adoption situation, and it is worth naming honestly because it cuts two ways. The hunger and capability inside a dealer's team is higher than the owner thinks. So is the risk: that private setup ran company email on personal hardware, moved data outside the company without anyone deciding it should, and pointed a competitor-disparaging bot at real customers. One person sprinting ahead on their own is not adoption. It is a governance problem wearing the costume of initiative. And the people who have not touched AI at all are not the laggards in this story — they are the reason the tool has to be boring, legible, and worth opening.

The mistake is to treat this spread as a training gap to be closed with a workshop. It is not. It is a signal that adoption cannot be bolted on after the build. It has to be designed into the thing.

Start from the job, not the tool

The move that works is to stop talking about the tool and start from the specific work a specific person does. Not "we are rolling out AI quoting," but "a salesperson currently spends two hours turning a request into an offer, and here is the part of those two hours we can remove." Adoption follows value, and value is measured in steps taken out of someone's actual day.

This is not only our observation. The largest study of the question found that most executives believe generative-AI success depends more on people's adoption than on the technology itself, and that a third of AI use cases had already been paused after their pilots (IBM, 2024). The mechanism underneath that has been documented in sales forces specifically for two decades: when salespeople were given CRM technology, the strongest predictor of whether they accepted it was perceived usefulness — did it help them do their job — ahead of ease of use, personal disposition, and management support (Avlonitis & Panagopoulos, 2005). Gartner names the same short list of drivers: value, culture, ease of use, and skills (Gartner, 2024). The through-line is that "the team won't use it" is almost always a verdict on the tool's fit, not on the team's character. A flawless training session on a tool that does not fit the work just teaches a team, very efficiently, to do something they will never do.

The difference between the tool that gets used and the one that gets abandoned is visible before launch, and it is not about the model.

	The tool that gets abandoned	The tool that gets used
What it does to the day	Adds data entry the salesperson feeds	Removes a step — the copy-paste from price list to offer
Who owns it inside	Nobody; it was mandated	A champion who wanted it
Who shaped it	Bought finished, then trained	A representative cohort, before launch
What it optimizes for	Looking good in the demo	The salesperson's actual job
What happens at month three	Stale data, quiet abandonment	Still open, because it still saves time

The champion and the six

Two practices carry the adoption, and both happen before launch rather than after.

The first is a champion — one named person inside the dealer who owns the project, wants it to work, and has the standing to push it. Without that person, a system ships into a vacuum and drifts. In the family-run distributor we are building for now, that role is explicit: the son runs the project and is the point of contact, while the internal IT lead who owns the ERP is a technical ally on every decision. The tool has an owner on the inside before it has users.

The second is that we put real users on the platform before it is finished — a small cohort, usually four to six people, chosen to represent the team rather than to flatter it. Two senior salespeople, two who joined recently, two who work from the office rather than the road. Each sees the tool differently, and the differences are the point. When someone proposes a new option, the cohort is how we find out fast whether it helps the seniors and buries the newcomers — whether it earns its place or just makes the app heavier. That protocol is our operating method rather than a proven law, so we hold it loosely; the reason each role is in the room is that each one catches a different way the tool can fail. There are two ways an app gains features: someone decides each one earns the added weight, or the app slowly bloats until the newest salesperson cannot find the button that mattered. The cohort is how you stay in the first world.

The tool that gets used

Put those together and the shape of an adoptable quoting tool is clear. It removes a visible step — the copy-paste from price list to offer — on day one. It is owned by a champion who wanted it. It was shaped by the people who have to live in it before it was finished. And it is boring: it does the job, it does not perform. In our experience the initial version that resolves most of the problem is smaller and quieter than anyone expected, because the problem was never a lack of AI. It was that quoting was slow, inconsistent, and invisible across the team.

A working demo is the cheapest part of an AI system; a tool the sales team opens without being told to is the expensive, real thing. Adoption is designed in the data and the workflow behind the offer, not in the cleverness of the model — and it is settled before launch, not managed after it. Whether a dealer should build one such tool or four, and in what order, is the last question in this cluster. The rest is operations, not theatre.