The email lands on a Tuesday afternoon. A long-term client, eight years on the books, has a question before the quarterly review. She has been reading about AI in her own sector, has seen the case study you published in March, and she would like to understand why the invoice has not started to come down now that “you’re using AI for most of this.” The review is in two weeks. She is asking the question quietly, on her own, before her FD gets sight of it.
The conversation is starting in earnest now, across audit, tax, legal advisory, marketing, and design. The KPMG and Grant Thornton case, where a 14% fee reduction was won on AI-efficiency grounds, gave buyers a vocabulary they did not have last year. PwC’s own Chief AI Officer has acknowledged on the record that clients are saying, in his words, “we want our fair share of those efficiencies.” This is no longer hypothetical. The question is what you say back.
Why the AI discount conversation is happening
Clients have a real basis for asking. They are using AI in their own work, watching it absorb tasks they used to pay people to do, and wondering whether their providers should pass on the same gains. Industry surveys through 2025 and 2026 suggest two-thirds of mid-sized professional firms have now had an explicit AI discount conversation. The pattern is real, and ignoring it does not make it stop.
The trigger is often the firm’s own marketing. When you publish a case study saying AI cut your research time by 70%, that figure does not stay inside your firm. It becomes a number a procurement team can quote back at you, with the most aggressive requests landing in audit, tax, and management consulting. The conversation arrives because you handed the buyer the script.
The flawed buyer logic, and the partial truth inside it
The buyer argument runs like this. AI does the work. Your costs are lower. Therefore your price should fall. It is internally consistent and it is wrong on the second step. Time saved on the production side rarely converts cleanly to cost saved overall. In regulated work, accountants and lawyers commonly report spending more time on verification, documentation, and human oversight once AI is in the workflow.
The partial truth is that productivity has shifted. AI does some throughput work faster, and pretending otherwise is its own credibility problem. The regulator still holds the human signatory accountable for every output, and that accountability has not fallen. You were never being paid for throughput. You were being paid for the outcome, the judgment that shapes it, and the risk you carry if the outcome is wrong. Speed gets you to the same place sooner. It does not lower the value of getting there.
Four conversation moves that hold the line
The first move is to reframe the work as outcomes rather than hours. When the client asks if AI has made the work cheaper for you, the honest answer is, “AI made the mechanics faster, and the result and the risk we carry are the same.” That sentence does most of the work in a real conversation. It accepts the productivity gain and uncouples it from the price.
The second move is to separate AI throughput from human judgment, in writing. A short deliverable map that names what AI accelerated (draft, summarise, search, cluster) and what your team actually did (chose, verified, applied judgment, signed off) makes the value tangible. Firms that document this routinely report fewer discount conversations, because the buyer can now see where the fee is going.
The third move is to surface what AI does not change. The professional indemnity sits with you. The regulatory accountability sits with you. The risk that the analysis is wrong sits with you. These have not fallen because the production tools got faster, and the price reflects them. Naming this out loud is uncomfortable for a sentence and clarifying for the next twelve months.
The fourth move is to restructure the engagement if the work has genuinely shifted. If a deliverable has become more standardised, or a recurring report has dropped in complexity, offer a scope or service-level adjustment, slower turnaround, fewer iterations, a smaller bundle, rather than a percentage off the same scope. The Spin Sucks framing is useful here, “We can absolutely reprice if we change scope or service level. Otherwise, we hold fees and deliver the higher-value package.” It gives the buyer a real choice without conceding the principle.
When a price adjustment is the right answer
Sometimes the discount request is correct. The test is whether the work has changed at the level of outcome, judgment required, or accountability, not whether your internal effort has fallen. If a quarterly compliance report that used to require senior review now genuinely needs less of it, the fee should reflect that. A price adjustment, in that case, is honest and protects the long-term relationship.
The conditions to refuse are equally clear. Same outcome, same risk, same client value, hold the line. If a piece of recurring work has become essentially commoditised because AI took over the judgment as well as the throughput, that is genuine change. If only your internal effort has dropped, that is a productivity gain you have earned, and reinvesting it in quality, faster delivery, or new advisory services is a defensible position. The conversation shifts from “what did this used to cost” to “what can you now do that you could not do before.” That reframing also tends to surface advisory opportunities the firm has not been pricing for, which can offset the discount pressure on the recurring work.
What this means for your business
The owners who handle this well will not be the ones with the cleverest scripts. They will be the ones who have done the underlying work, knowing where AI sits in their delivery, where judgment sits, what their indemnity covers, and what the outcome is worth to the buyer in pounds. Holding pricing power in 2026 is largely about being clear-eyed about your own value.
If your firm is heading into a quarterly review where this conversation is likely, the worst thing you can do is improvise it. The second worst thing is to concede 5% as a goodwill gesture. The best thing is to have the four moves prepared, know which one fits the relationship, and be willing to redesign the engagement if the work has genuinely shifted under your feet.
If you would like a second pair of eyes on how to set this up inside your firm before the next round of client conversations, book a conversation.
