What an AI Consultant Hears When You Talk About Price

Ask an AI consultant what a project costs and you'll get a number. What I'm actually listening for is how you react to it — because that reaction predicts how the whole project will go far better than the number itself ever could.

The Question I'm Actually Answering

When a prospective client asks what something will cost, they think they're asking about money. What I'm actually deciding, in that moment, is something else entirely: will this be a partnership where we solve problems together, or will every invoice become a negotiation? The number is almost beside the point. I've quoted the same project at the same price to two different companies and watched one become a three-year relationship and the other fall apart within a month — the difference was never in the figure.

I don't say this to sound mysterious about pricing. Our rates are not a secret, and I'll happily explain exactly what goes into a number the moment someone asks. But the way a founder or a manager responds to that number — before we've done a single hour of work together — tells me more about the next six months than their brief does.

Three Reactions That Tell Me Everything

Over enough of these conversations, three reactions keep repeating, and each one predicts something specific.

The immediate counter-offer. Some people hear a number and instantly propose a lower one, before asking a single question about what's included. That's not thrift — thrift asks questions first. It's usually a signal that the relationship will be adversarial from the start, that every future scope conversation will be treated as an opportunity to claw something back, and that "done" will always be a moving target because the goal was never the outcome, it was the discount.

The question about what's excluded. The clients I end up working with longest almost always ask the opposite question: not "can you do it cheaper," but "what happens if this takes longer than expected" or "what's not included in this number." That's someone trying to understand the shape of the risk, not trying to shave the number down. It's the single best predictor I have of a client who will still be a client in a year.

The silence. Occasionally someone says nothing at all — no pushback, no clarifying question, just agreement. That one surprised me the first few times. It isn't always good news. Sometimes it means the number was never really evaluated against a budget, which means the first invoice becomes the moment reality sets in, and that conversation is much harder to have after the work has started than before it.

Why the Cheapest AI Consultant Almost Never Turns Out Cheapest

I'm not going to pretend price doesn't matter — it obviously does, and if you want the actual mechanics of how AI project pricing works, fixed price versus hourly billing is a separate, more technical piece worth reading before you sign anything. What I will say is that the lowest quote in a batch of three is rarely the cheapest project by the time it's finished, and the pattern is almost always the same: a low number was hit by leaving something out, and that something reappears six weeks later as a change request with its own invoice attached.

That's not a criticism of anyone quoting low — sometimes a lower number reflects genuinely lower risk, a narrower scope, or a consultant hungry for the work who's pricing fairly. The problem isn't the number, it's not knowing what's inside it. A client who asks what's excluded from a quote, in my experience, almost never gets surprised later. A client who only compares the headline figure across three proposals is comparing three different scopes and calling it one comparison.

The Fixed-Price Question I Get Asked Constantly

Almost every first conversation eventually arrives at some version of "can we just do this fixed-price so we know what we're spending." I understand the appeal completely — a fixed number feels safer than an open-ended one. But I say no to it more often than people expect, and the reason has nothing to do with wanting to bill more hours.

A fixed price only works when the scope is genuinely fixed, and in the earliest stage of an AI project, it rarely is. Nobody actually knows yet how messy the underlying data is, whether the exception cases will be simple or a nightmare, or how many rounds of "actually, can it also do this" will surface once people start using a working prototype. Quoting a hard fixed price on that kind of uncertainty means one of two things happens: I pad the number heavily enough to absorb the unknowns, which means the client overpays for certainty they didn't need, or I quote tightly and eat every surprise myself, which is a fast way to resent a project I should have enjoyed. Neither is honest. I'd rather scope a small, fixed-price first phase — narrow enough that the unknowns are genuinely small — and only fix the price on the larger phase once that first phase has removed the guesswork.

What I Actually Say Before I Quote Anything

Before a number goes anywhere, I ask what happens if this project goes over budget or over time — not because I expect it to, but because how someone answers that hypothetical tells me whether they're thinking about this as a partnership or a transaction. I ask what "success" looks like in a sentence a technical person and a non-technical person would both agree on, because a client who can't answer that yet isn't ready for a number at all, regardless of how badly they want one.

And I say, plainly, that the number I give reflects the scope we've discussed, not a promise that nothing will change. Projects change. Data turns out messier than described. A stakeholder who wasn't in the first meeting has opinions in the third one. None of that is a failure of the quote — it's the reason a quote needs a documented scope behind it, not just a figure on an email.

Five Signs the Price Conversation Is Really a Trust Conversation

A handful of patterns repeat often enough across first conversations that I now read them as a single signal, not five separate ones.

Someone asks about payment terms before they've asked what the deliverable actually is — that's a business that's been burned by a vendor before, and it's worth asking gently about that history rather than treating the question as a red flag on its own.

Someone wants a written scope document before a verbal yes — that's not distrust, that's someone who's run a project before and knows exactly where vague scopes go wrong.

Someone asks who specifically will be doing the work, not just what the company charges — that's a client who's evaluating a relationship, not procuring a commodity, and those tend to be the best clients to have.

Someone's first question about the invoice is about the payment schedule, not the total — that's someone managing cash flow carefully, which is a sign of a well-run business, not a warning sign about the client.

And someone who negotiates the price down but not the scope down — that's the one pattern that reliably predicts friction later, because the gap between what was promised and what was paid for has to come from somewhere, and it's usually goodwill.

Saying No Because of How Someone Talks About Money

I've turned down projects because of how the pricing conversation went, not because of the number itself. Not often, and never over a single tough negotiation — negotiating fairly is a normal, healthy part of doing business, and I'd be suspicious of a client who never pushed back at all. What makes me say no is a specific pattern: someone treating every clarifying question I ask about scope as an attempt to inflate the bill, rather than an attempt to make sure we both understand what we're agreeing to.

That pattern rarely stays contained to the first conversation. A client who assumes bad faith during pricing tends to assume bad faith during delivery too — every delay gets read as incompetence rather than a legitimate complication, every judgment call gets second-guessed after the fact. I'd rather lose the project at the quote stage than discover that dynamic three months into a delivery, when walking away costs both of us far more.

A Pattern I See Often Enough to Trust

I don't name clients in essays like this one, but the pattern below has repeated often enough, across unrelated industries, that it's worth describing in general terms.

A company gets three quotes for the same AI project. Two are close together; one is noticeably lower. The lower quote wins, understandably — it's a meaningful amount of money, and on paper the deliverable looks identical. Six weeks in, the first change request arrives: turns out the original scope didn't include handling the messiest 10% of the data, or didn't include a second language, or assumed a level of internal IT support that didn't actually exist. None of this is dishonest exactly — it's what happens when a number gets built backward from "what would make this look competitive" rather than forward from "what does this actually require."

By the time the project is finished, the total often lands close to what the higher quotes would have cost from the start — except now it's arrived in three invoices instead of one, each one a small negotiation, and the trust that would have made those negotiations easy was never built in the first conversation. The company that chose based on the lowest number didn't get a worse consultant. They got the same project at the same eventual price, with the friction moved from before the contract to after it.

The clients who avoid this pattern almost always did one specific thing differently at the quote stage: they asked what wasn't included before they asked what the total was. That single question, asked early, does more to protect a project's budget than any negotiation ever could.

What I'd Ask Instead of "What Does This Cost"

If you're evaluating an AI consultant and want a question that reveals more than the number will, ask what's explicitly excluded from the quote, not just what's included — the honest answer tells you more about how the rest of the relationship will go than the total ever could. Ask what happens, specifically, if the project runs over — a consultant with a real answer to that has done this before; one without a real answer hasn't. And if you want the broader checklist for choosing who you work with in the first place, the three questions worth asking before you hire an AI consultant is the piece to read alongside this one — pricing is one signal among several, not the whole picture.

If you're at the stage of wanting the actual numbers rather than the psychology behind them, our pricing page lays out how we structure engagements, and how we work with SMEs choosing an AI partner covers the rest of what we actually do once the quote is signed.

The Part That Isn't About Money at All

None of this is really about price. It's about what a number is a proxy for — whether two people are entering an agreement with the same understanding of what's being promised, and whether both sides plan to act in good faith when reality inevitably differs from the plan. A client who negotiates hard but fairly, who asks precise questions and accepts honest answers, is telling me something I need to know before we start, not after. The number on the quote is the least interesting part of that conversation. It's just the part that happens to get written down.

Frequently asked questions