Selling AI to skeptical buyers (a field guide)

The enterprise buyer sitting across from you — or on the other side of your Zoom — has seen fifty AI demos in the past six months. They have heard “it's like having a junior analyst on your team,” “it learns your workflow,” and “you'll save 40% of your time” more times than they can count.

They have also been burned. The tool that was going to transform their content pipeline is now a $30K line item that three people use occasionally. The AI email writer generates copy that sounds like it was written by a robot trying to sound human. They approved the budget, championed it internally, and now they own the failure.

This is the room you are walking into. And it is actually a good room to be in — if you know what to do with it.

The instinct that kills deals

Most sellers respond to skepticism by adding more proof. More case studies. More ROI calculations. More demos that show additional features. The logic is: they don't believe me, so I need to show them more evidence.

This almost always backfires. When a buyer is skeptical, more evidence lands as more noise. They are not evaluating your evidence — they are filtering for the catch. Every impressive stat triggers the internal question: “Yes, but what's the part you're not telling me?”

The move that actually works is to stop selling and start diagnosing. Get curious about the specific failure that created the skepticism. The conversation shifts from “let me show you what we can do” to “tell me about the thing that didn't work last time.”

The framework I actually use

I call it proof of work, and it has three components.

Acknowledge the failure first.Not your product's failure — the category's failure. “AI tools have massively overpromised in the past two years. The gap between what got demo'd and what got delivered has been significant. I assume you've experienced some version of that.” This does two things: it signals that you understand their reality, and it separates you from the vendors they've already tried. You are not pretending the category has been perfect.

Get specific about the failure mode.“When you think about the AI tools that have underdelivered for you — what was the gap? Was it that the output quality wasn't there? That the adoption never happened? That the ROI was real but impossible to attribute?” Let them tell you exactly what went wrong. That answer tells you everything about what they need to hear from you.

Match your story to their failure. If adoption was the problem, lead with implementation support and change management — not product features. If output quality was the problem, get into a working session where they see real output against their actual use case, not a generic demo. If attribution was the problem, talk about measurement before you talk about the product.

Handling "we're evaluating everything"

The most common stall in enterprise AI sales right now is the parallel evaluation. “We're looking at six vendors, we'll make a decision in Q3.” It sounds reasonable. It often means the buying process is stuck and no one has authority to move it.

The right move is not to compete on the evaluation rubric — it is to get off the rubric entirely. Ask for a pilot on a specific, bounded use case with a defined success metric. “Instead of a parallel evaluation across six tools, let us earn the right to compete. Give us one use case, thirty days, a clear definition of what success looks like. If we hit it, you have real evidence. If we don't, you have lost thirty days instead of six months.”

Most buyers say yes to this — because it is a better deal for them, and because you are the only one offering it.

The enterprise timeline reality

Enterprise AI deals are slow. Slower than you think they should be, slower than the buyer thinks they should be, and slower than your forecast says they will be. Security reviews, procurement, legal, IT integration, change management — each of these is a real gate, and treating them as obstacles rather than requirements is a mistake.

The sellers I have seen close the most in this space are the ones who get ahead of the process instead of behind it. Who ask about procurement requirements in the discovery call. Who introduce their implementation team before the contract is signed. Who map the internal champion's career risk alongside their business case.

Skeptical buyers are not blocked buyers. They are buyers who have been burned and need a different kind of evidence — evidence that you understand their situation, that you are not going to oversell them into another failure, and that you will still be there six months after the contract to make sure it works.

That is a higher bar than most sellers want to clear. It is also why the sellers who clear it tend to own the accounts they win for a long time.