How to Price Bookkeeping When AI Makes the Work Faster

When AI makes the work faster, hourly billing punishes you for it. If a client's books took six hours and now take three, an hourly fee halves, even though the client gets the same or better books. That is the core problem behind a lot of 2026 practitioner discussion about pricing.

This post sets out the numbers that are actually published, what they do and don't tell you, and a simple way to decide what to do with your own fees.

The numbers that are out there

How fast AI can be. Blake Oliver had Claude code two years of books (nine bank accounts, about 2,200 transactions) in roughly four hours, with six misses (Blake Oliver, Jun 2026). That was one test by one author, but it shows the scale of change for catch-up and cleanup work.

How much time firms say they save. Karbon's 2026 survey reports an average of 60 minutes a day saved per person, roughly 21 hours a month (Karbon, 2026 report, vendor survey). It is a vendor survey with a self-described lean toward progressive firms, so treat it as an upper-end picture, not your number.

Where fees are going. Blake Oliver cites CPA Trendlines data showing firm pricing up 4.2% year over year after a 2.1% decline, with tax preparation up about 8% and advisory up 6% or more (Blake Oliver, Jun 2026). There is no bookkeeping-only figure in that data.

What clients see as the low end. Pilot advertises AI-only bookkeeping at $99 a month; Bench runs $199 to $599 a month (Rework, Aug 2026, vendor list). Your clients may have seen those prices.

What practitioners say they charge. On a practitioner forum in April 2026, a bookkeeper with ten years in practice and 22 clients proposed monthly retainers of $500, $1,000 and $1,500, with $1,500 onboarding, and estimated the effective rate would rise from $60 an hour to $150 to $200. A certified public accountant (CPA) in the same thread described retainers of $1,200, $1,800 and $2,500 with onboarding of $2,000 to $3,500 (beancount.io forum, Apr 2026). These are anecdotes, not a survey. They are useful as a sense of range, not as a benchmark.

The labor picture. The Bureau of Labor Statistics puts median pay for bookkeeping, accounting and auditing clerks at $24.36 an hour (May 2025) and projects 6% fewer jobs from 2025 to 2035 (BLS, Aug 2026). Meanwhile the number of US payroll and bookkeeping businesses grew 2.2% to 331,316 in 2026 (IBISWorld, 2026). More firms, fewer clerk jobs: competition for clients is not easing.

The three options

Blake Oliver frames it simply: if you bill hourly for write-up work, you either lower the price as the hours fall, or you charge for the outcome (Blake Oliver, Jun 2026). In practice a small firm has three moves.

1. Stay hourly and accept lower bills

Honest and simple. It works if you can fill the freed hours with new clients at the same rate. It fails if the freed hours stay empty.

2. Move to a fixed monthly fee for the same work

The client pays the same each month; you keep the time you save. The forum thread's phrase for this is "price on their metric, not yours" (beancount.io forum, Apr 2026). The risk: if a client compares your fee to a $99 AI service, a fixed fee for invisible work is hard to defend.

3. Fixed fee, with something new the client can see

Package the time saved into output: a monthly note on what changed, a short quarterly call, a cash-flow check before payroll. The client gets something they didn't have before; you get a fee that doesn't depend on hours. This is the option most of the published advice points toward, and the one that holds up best against low-price AI services.

A worksheet you can run this week

For your ten largest clients:

1. Hours before. Average monthly hours over the last six months before you turned on the ledger AI.
2. Hours now. Average for the last two or three months.
3. Current monthly revenue from the client.
4. Effective rate before and after. Revenue divided by hours.
5. What they'd see. Write one line on what this client would value that you don't send today.
6. Tier. Put each client in a draft tier: books only; books plus monthly note; books plus note plus a quarterly call.
7. Gap. Compare current fee to the tier price you'd want.

If effective rates have risen and clients are happy, you may not need to change anything yet. If you are hourly and hours have dropped, the worksheet shows you how much revenue is leaving.

Making the change without losing clients

• Give notice. Practitioners in the same thread describe raising or restructuring fees at renewal with 60 to 90 days' notice (beancount.io forum, Apr 2026).
• Show the new thing first. Send two months of commentary before the new fee starts, so the client sees what they're paying for. Our commentary workflow post shows how to draft it with AI and review it quickly.
• Be straight about AI. If a client asks whether you use AI, say yes, and say how your team reviews it. Your AI rule sheet gives you the words.
• Let some clients go. The forum thread calls them "price shoppers, not value buyers." A client who only wants the cheapest possible books has a $99 option now.

The decision in one line

If AI is cutting your hours, decide this quarter whether you are passing the savings on or turning them into something clients can see. Doing nothing is a choice, and on an hourly fee it is a pay cut.

For the full picture of what the ledger AI does today, start with our 2026 guide.

---

If you'd like a second opinion on your numbers, start a free conversation. We'll go through your month-end and tell you plainly where the time is going and what we'd do with it.