The most instructive fact in the accounting pricing debate this year is not a survey. It is a transaction.
The Argument Does Not Belong to Either Side
KPMG demanded a reduction in its own audit fee from Grant Thornton UK, on the basis that AI had reduced the cost of the work. Public filings confirm a significant reduction followed.
Sit with the shape of that for a moment. A Big Four firm that sells AI-driven efficiency to its clients, applying precisely that logic as a buyer, to another accounting firm, and winning.
Most commentary treats this as a threat to firms. That framing is comfortable and slightly wrong.
The proposition is simply that a price built on effort should move when the effort moves. There is nothing unfair in it, and no firm that has ever argued its own fee should reflect the value delivered rather than the hours burned is in a position to call it unreasonable when a client says the same thing back.
What makes it uncomfortable is that most professional pricing was never actually value-based. It was cost-plus with a story attached, and AI is now testing which of those it really was.
What the Numbers Say About How Far Along This Is
A General Assembly survey of 258 director-level and above leaders at consulting, accounting and legal firms with more than 1,000 employees in the US and UK puts some shape on it.
79% say AI is changing pricing conversations. 42% say clients are actively questioning their pricing model, and only 37% say they are addressing it proactively, which leaves a substantial group waiting to be asked.
Among accounting firms specifically, 35% report clients questioning pricing because of AI and 73% are changing how they talk about pricing, 38% of them proactively. Ash Khanna of General Assembly puts the client position bluntly: as AI makes work more efficient, clients are questioning traditional billable hour pricing and demanding transparency.
The firm-size detail is the one small firms should notice. Among firms with 1,000 to 4,999 employees only 9% reported no pricing impact, against 31% at larger firms. Scale is currently providing some insulation. Smaller practices do not have it.
The Two Bad Responses
There are two instinctive reactions and both make the position worse.
The first is silence: keep the fee, say nothing, hope the client does not raise it. That converts a negotiation you could have shaped into an ambush you cannot, and the client who works it out themselves has also worked out that you were not going to mention it.
The second is a defensive discount: drop the price the moment it is questioned, with no change to what is being sold. That ratifies the client's framing that they were buying hours, and there is no floor underneath it.
The workable third option is to change what is being priced. If the AI saving is real, it should show up somewhere the client can see, whether as a lower fee on compliance work or as scope that used to cost extra and now does not. That is a conversation a firm can hold on its own terms, and it is considerably easier to hold first than second.
The Part That Is Actually Hard
None of this is a communications problem. It is a measurement problem.
To hold the conversation at all a firm needs to know where the time actually went on an engagement, what a piece of work genuinely costs now, and which parts AI touched. Most practices cannot answer that with any precision, which is why the instinctive responses above are so common. It is difficult to negotiate about a saving you have not measured.
MetaWurks addresses one slice of the underlying work rather than the pricing question itself. It ingests a client's invoices, contracts, statements and correspondence and lets an accountant query the whole set in plain English, so the retrieval that used to consume an afternoon becomes a question. Documents ingested into the platform are not used to train models and are not exposed to other users, role based access controls decide who can open which client's records, and audit logs record who opened what and when.
Where that changes the hours on a job, it is worth knowing by how much, because that number is the one the conversation turns on.
KPMG did not wait to be asked. Neither will your clients.
Join the Conversation
If a client asked tomorrow what AI has saved on their engagement this year, would your answer be a figure or an impression?