A client writes to a partner in March. One line: you mentioned the team is using AI now, and the invoice is the same as last year, so help me understand.
The Question Nobody Prepared For
At large accounting firms, 35% now report clients questioning their pricing model because of AI, and 73% are changing their messaging around pricing. Those figures come from a General Assembly survey of 258 director-level and above leaders, reported by Accounting Today, and the sample matters: every firm in it has at least 1,000 employees. This is not yet a small-firm statistic. It is a question arriving first where clients have procurement departments and the leverage to ask it.
Changing what you say is not an answer to a question about what you charge.
The reflex response is that AI is killing the billable hour. For most CPA work, the billable hour packed up years ago. Ignition's 2025 benchmark, drawn from 219 US accounting firms using its own platform, found that only 3% charge hourly for tax prep, and just 17% charge hourly for advisory services, down from 21% the year before.
So the client is not questioning your rate. In most engagements there is no rate to question. They are questioning a fixed fee, and a fixed fee is a different kind of promise.
What a Fixed Fee Actually Contains
A fixed fee is a bet on effort. It gets set by looking at what the engagement took last year, and the year before, then adding something for the parts that always go wrong. Nobody writes that down, but that is the arithmetic underneath almost every quoted number in the profession.
That arithmetic has stopped holding. Sixty percent of tax professionals now use AI for tax research at least weekly, up from 33% a year earlier, according to a Blue J and CPA.com survey of more than 1,000 US tax professionals. In the same survey, 84% agreed AI saves them time.
The effort moved. The price stayed where it was. The client noticed the gap before most firms got around to repricing it.
Two Answers That Both Cost You
The first is to absorb it quietly and hope nobody asks. This works until someone asks, and at the large end of the profession a third of firms are already past that point. It also means the first serious conversation about your fee happens on the client's timing, in a tone they chose, with you responding rather than explaining.
The second is to cut the fee to match the new effort. This feels principled and it hands the entire benefit of a multi-year technology shift to the buyer. It also reprices your firm around its lowest-effort year, which is a difficult number to walk back from when a complicated return lands.
Neither is a strategy. Both are what happens when the pricing model was never articulated in the first place.
The pressure here is not theoretical. Sixty-seven percent of accounting firms in the General Assembly survey plan to keep headcount flat while relying on AI. That gain is real, and it is landing somewhere. Margin that nobody has explained is margin waiting to be argued about.
Price the Judgment, Not the Remembered Hours
What survives automation in this profession is judgment. Which treatment to take. What got caught in review. What the firm carries if the position is wrong. None of that got faster, and none of it is what the fee was implicitly measuring.
The honest answer to why the fee is the same is a description of what the engagement actually involved: the reconciliation that surfaced a misclassification, the position that needed a second look, the exception someone handled so quietly it never reached the client. That answer is available in almost every firm and provable in almost none.
Which points at the real problem. The work that justifies the fee is the least documented work in the building. It lives in a reviewer's head, in a comment on a workpaper, in a message from a Tuesday in February. When a client asks what they are paying for, the firm reaches for that record and finds an anecdote.
Where This Leaves the File Room
MetaWurks exists for the part of this that is a document problem rather than a pricing philosophy. It ingests a client's returns, statements, invoices and correspondence and lets an accountant query them in plain English, so that finding what a client spent on something in the third quarter stops being a twenty-minute hunt through folders. Role-based access controls and audit logs record who opened which file and when. And because documents ingested into the platform are not used to train models or exposed to other users, the client's tax file does not leave the firm's control to make any of that happen. That is a different proposition from pasting a client return into a consumer chat window, which is still how a good deal of this work quietly gets done.
That does not, on its own, produce the record a partner needs in a fee conversation. No software writes down why a position was taken. What it does is remove the excuse: when the searching, retrieving and cross-referencing stop eating the week, the work that remains is the work worth describing, and there is time left to describe it. A firm that can account for what an engagement involved has something to say when a client asks why the number did not move. A firm that cannot is asking the client to take its word for it, in a year when 84% of the profession has just told a survey that the work got easier.
The fee was never really for the hours. It was for the part of the job where someone decides. That part is still there, still expensive, and still yours. It is just no longer buried under enough visible effort to speak for itself.
Join the Conversation
If a client asked tomorrow what your fee actually covers, could you show them, or would you have to describe it from memory?