A client of mine spent years on our full-service package — we did the bookkeeping, the monthly reports, the year-end accounts, and the tax return, for a monthly fee. Then they paused the business for a stretch. When they came back, they chose our cheaper self-service tier instead: the owner does their own books, we check in mid-year, file the sales tax returns, and budget a few hours of cleanup into the year-end. That tier exists for exactly this kind of client, and plenty of them run on it just fine.
This one didn't. Their system integrations were never set up right, and by year-end the file was a mess. We blew past the budgeted cleanup hours rebuilding integrations and redoing a year of bookkeeping.
When the cleanup fee landed, the client balked. We discounted. By the time the file was clean, we'd put in more hours than the full-service package would have taken — and been paid significantly less. Then the sticker shock cost us the client anyway.
You've lived this file
If you've been in practice more than a few years, you have your own version of that story. The client does their own books — or a bookkeeper who wasn't up to it did, or now their AI did — and the catch-up bookkeeping lands on your desk at year-end. You do the rescue, and you bill less for it than the monthly engagement would have paid, even though it's the same work plus the diagnosis. That's the inversion almost every firm runs: the harder job is the one we discount.
Nobody says this out loud because it sounds like complaining. It isn't. It's a pricing failure with a mechanism, and AI is about to industrialize it.
Cleanup isn't bookkeeping done late
It's a different and harder job. Original bookkeeping happens with context — you see the transactions as they occur, and you ask questions while the client still remembers the answers. Cleanup is forensic reconstruction of a ledger you can't trust: diagnosis before correction, no contemporaneous context, errors compounded quietly across months. And when it's done, your name goes on numbers whose provenance you didn't control. That's a risk premium, not a rounding error.
The builder who remediates an unpermitted renovation charges more than the builder who'd have done it to code. Remediation always costs more than doing it right would have. No one calls that unfair — except, apparently, us, when we price our own version of it at a discount.
AI industrializes the shoebox
The shoebox client is older than any of us. What's new is the machine. "My AI does the books" is spreading through your client base right now, and let's be honest about the range: some of those files will be fine. The claim isn't universal disaster — it's volume and variance. Unsupervised files are multiplying far faster than their quality is improving, and you can't tell which kind you're holding until you've looked.
Last week I argued that the $49 close assumes a system that hasn't yet been built. This is the same argument arriving at your desk: when the system was never built — or an AI ran unsupervised on top of a broken one — somebody reconstructs the record before anything can be trusted again. That somebody is you. The only question is whether you price it like the harder job it is.
Price the file diagnostic before the fix
Here's the mechanism we should have had in place, and the one I'd now put in front of any firm.
Never quote a cleanup blind. A paid diagnostic — a file review with a fee attached — is the front gate. It's what keeps the higher rate honest: you're not assuming the file is a mess, you're showing the client what you found before either of you commits to the fix.
If you're a year-end or tax-only firm, build the diagnostic into the engagement fee — with a clearly stated term that remediation beyond it is billable. That's the structural fix to my own story. Our self-service tier budgeted the cleanup hours; what it needed was the stated term, priced and agreed at sign-up, for what happens when the file goes beyond them.
And publish cleanup rates that sit above your do-it-right rates — the visible price signal that doing it right with a professional from day one is the cheaper path. The rate card should say what the engagement letter says: the rescue costs more than the system.
The discount didn't save the client
Here's the part that stings. Pricing the cleanup below the monthly package subsidizes the client's decision to decline you — it makes the wrong choice free. And the discount doesn't even buy loyalty.
We didn't lose that client at the invoice; we lost them at re-engagement, when nothing set the expectation of what a mess would cost. By the time the bill argues with the expectation, the relationship is already gone. The discount wasn't generosity. It was a deferred termination, with extra hours attached.
Run-time keeps getting cheaper; the design and the accountability underneath it don't — and neither does the rescue when there was no system at all. So look at your engagement letters this week: what do they say a mess costs? If the answer is nothing, book a free consultation and we'll work out where your diagnostic, your cleanup rates, and your tiers should sit.