She had talked herself into forty thousand pounds. The proposal sat in a draft tab, the cover note half written, the send button waiting. Her gut said fifty-five. She had buried that number twice in the last week, once when she compared the offer to a similar engagement she had run for a smaller client, and once when she remembered a peer who, she assumed, would charge less for the same shape of work. Both signals had moved the price down without her noticing.
This is the founder decision that is most heavily distorted by emotion. Pricing touches on fear of losing the deal, on the time you have already sunk into shaping the offer, and on a quiet running comparison with people you have never actually surveyed. AI sat in the room with her would not have told her the right number. Used well, it would have made every assumption that justified forty thousand visible before she pressed send.
Why is pricing the most distorted founder decision?
Pricing is the place where three biases meet and reinforce each other. Anchor bias holds the first number you write down as the gravity well around which everything else orbits. Loss aversion means the pain of losing the deal feels roughly twice as intense as the pleasure of winning the extra fee. Sunk cost loyalty to the offer shape makes a low counteroffer harder to walk away from than the marginal economics justify.
Tversky and Kahneman showed in 1974 that even arbitrary anchors hold against deliberate effort to escape them. The same paper found that financial incentives for accuracy did not weaken the effect. Layer on the macro context. The Federation of Small Businesses Small Business Index landed at minus seventy-one in Q4 2025, the lowest reading since the COVID-19 outbreak. Founders pricing in a low-confidence environment quietly anchor lower, accept worse terms, and underweight their own value.
What can AI actually do in the pricing conversation?
AI’s job is to enumerate the assumptions your draft price depends on, with a clarity you cannot achieve alone reviewing your own work. BCG’s 2026 research on B2B pricing argues for a pattern where AI proposes and orchestrates, and rule-based components, or the founder, enforce the non-negotiable guardrails. Wade Foster at Zapier calls his version the ninety percent rule. The model handles the logic, the human handles the last ten percent.
For a pricing conversation, that translates into a specific role. The model surfaces the cost floor implied by your draft, the value ceiling implied by what the client will gain, the market anchor your peers actually charge, the historical reference point you carry from your back catalogue, and the positioning signal the price sends. You then make a conscious decision about which of those anchors should govern. The price stays a founder choice, and the reasoning stops being invisible.
What does the prompt actually look like?
The workhorse prompt is short. Paste the offer description as you would send it to the client. Paste the draft price. Ask the model to enumerate every assumption that justifies the number, line by line. Then in a second pass, ask it to stress test each assumption with one credible alternative. The output reads as a checklist of things you were quietly taking for granted.
A founder running this on the forty-thousand-pound offer would see her assumptions laid out. Two-week scope, six-day delivery cadence, no scope creep, the client’s budget cap matches your guess, your hourly anchor of one hundred and fifty pounds is the right reference point, your peer is charging less than you, this client values process over outcome. The stress test then asks: what if the scope is three weeks, what if the hourly anchor is wrong because the value here is outcome not time, what if the peer comparison is invented. None of those questions is unfamiliar. Seeing them written down changes whether you act on them.
The second pass matters more than the first. The first pass produces a list. The second pass forces you to defend each item, because the model is offering one credible counter-position to each assumption and asking which is closer to the truth. That is the moment the back-catalogue anchor gets challenged. The hourly rate that has held since 2023 might be wrong because your work has matured. The peer who you assume undercuts you might be charging more on outcomes you cannot see. The scope estimate might be padded because the last engagement of this shape ran over and you do not want a repeat. The model does not know any of this. What it does is make space for you to admit it.
How do you keep the judgement step with the founder?
Visibility is one thing, direction is another. Once the assumptions are on the page, the model has done its useful work. The judgement that follows belongs to you, because three things cannot be algorithmised. The specific client relationship, what they have signalled about budget, what they value beyond the deliverable. The market position you want to occupy, and what each price implies for it. Your own appetite, this month, for a difficult close.
Alan Weiss’s formula for value-based fees is one structured way to make that judgement explicit. Tangible outcomes multiplied by annualisation, plus intangible outcomes weighted by emotional impact, plus peripheral benefits, divided by the client’s expected return on investment. The formula matters less than the discipline it forces. The founder names every input. The model can show you the inputs you forgot. It cannot tell you what your work is worth to this client in this market this quarter.
What discipline closes the loop before the call?
Write three numbers down before you pick up the phone. The price you will state, the lowest you will accept, and the point at which you will walk away. All three. Without the third, loss aversion will quietly redraw the line during the negotiation, because the pain of losing the deal feels twice as sharp as the pleasure of holding firm. Blair Enns calls this the fear gap, closed through evidence and prior commitment.
The Harvard Program on Negotiation’s reading of Galinsky’s research adds a useful nuance. If you have market data, understand the zone of agreement, and can defend the number with objective standards, making an ambitious first offer measurably improves the final outcome, even though it feels harder in the moment. The three numbers are how you make that ambition possible. You have already decided. The conversation cannot drag you below a line you have written down.
The same approach to hard decisions, AI as a sparring partner that surfaces the logic and a founder who keeps the call, applies across the operating year, not just at pricing time. Pre-mortems on a new hire, scenario tests on a strategic bet, the difficult client conversation you are rehearsing for Tuesday morning. The pattern holds. You do the thinking out loud with a model that is patient enough to enumerate every assumption. You then make the call yourself, with the assumptions visible and the line already drawn.
If you want a thinking partner for the next pricing decision sitting on your desk, book a conversation.
