Moving a client off hourly billing
A mid-relationship migration from hours to fixed deliverables: rebuild their own timesheet history into a rate card, run a shadow-quote month, then switch.
Every efficiency you find on an hourly contract is a pay cut you administer to yourself. You spend a weekend building a prompt library that turns a two-day edit into a single afternoon, and the reward is the smallest invoice you have ever sent that client. Nobody else in the supply chain works this way. Your accountant does not bill less because their software got faster.
The problem is not the rate. It is that hourly billing publishes your cost structure on the invoice and invites the client to negotiate against it. Every line is an argument waiting to happen, and every tool that speeds you up hands the client the winning side of it.
Moving off hourly at the start of a relationship is easy — you just quote differently. Moving off hourly in month fourteen, with a client who has an approved rate on file and a procurement process built around timesheets, is the actual problem. This is that migration.
The rate you would need does not exist as a comparable
Run the arithmetic before the conversation, because it explains why "just raise my hourly rate" is not the fix.
Say a deliverable used to take 24 hours at $90/hour: $2,160. Now it takes 6. To hold the same fee you need $360/hour.
Nothing in the market lets you defend that number. The hourly rates a client can find for AI work are consulting rates, quoted for advisory and implementation engagements, and there is no published day-rate benchmark for AI-first creative production at all. So when you propose $360/hour for making short-form video, the client compares it against the only anchors available to them, all of which sit lower, and you lose an argument you should never have entered.
The fee was never wrong. The unit was. A deliverable that was worth $2,160 to the client's business is still worth $2,160 to the client's business, and how many hours you spent producing it was always private information you volunteered.
Migrate on their history, not on your promise
The single mistake that kills these conversations is arriving with a rate card the client has never seen a basis for. They will read it as a price rise disguised as a restructure, because that is what most of them are.
Instead, build the rate card out of their own approved invoices. You already have the data.
- Pull the last six months of timesheets. Twelve if the work is seasonal.
- Re-cluster the hours by deliverable, not by activity. Your timesheet says "editing 6h, revisions 3h, exports 1h." Rewrite it as "Product hero cut, 10h." This is the whole trick. Hourly logs are organised around what you did; a rate card has to be organised around what they bought.
- Divide by their approved rate. Now every deliverable has a historical price the client has already paid, at a number their finance team already signed off.
- Take the median, not the mean. One catastrophic month should not set the rate, and neither should your best one.
- Band anything with a spread wider than about 2x. If "social cut" ranges from 3 to 14 hours, it is not one deliverable — it is two or three, and the difficulty drivers behind that spread are what the bands should be built on.
You now have a rate card where every line traces back to an invoice they approved. That is a completely different conversation from a rate card you invented.
The shadow-quote month
Do not switch on the strength of a spreadsheet. Run one cycle in both units.
For one month, keep billing hourly exactly as before. In parallel, issue a shadow quote at the start of the month — the same work, priced per deliverable off the new card — and send it alongside the normal invoice at the end with both totals side by side.
Three things happen. You find out whether your bands hold on a real month rather than a historical average. The client sees the two numbers land close together, which removes the suspicion that this is a price rise wearing a costume. And the scope conversations start happening at the beginning of the month, where they belong, instead of appearing as a variance line at the end.
If the shadow quote comes in materially under the hourly total, do not celebrate. It means your bands are too low or the month was unusually heavy, and you have one free cycle to find out which.
Switch on a natural boundary: contract renewal, a new quarter, or the start of a new campaign. Migrating mid-project gives you two units of account on one piece of work and an argument about which hours fall on which side.
What replaces the timesheet
Hourly billing comes with a whole apparatus, and if you remove it without replacing it the client feels the loss of control immediately. Every artifact has a fixed-fee equivalent:
| Hourly artifact | Fixed-deliverable replacement |
|---|---|
| Timesheet | Deliverable schedule with dates |
| Hours estimate | Shot count and difficulty tier |
| "This will take more hours" | Change order at a published rate |
| Utilisation report | Output report plus performance data |
| Rate increase | Tier change or scope change |
| Overrun conversation | Revision cap already in the agreement |
The revision cap is the one that carries the most weight, because hourly billing was silently doing that job. Under hours, round four was simply more hours, so nobody had to define anything. Under a fixed fee, round four is free unless you wrote it down. Bring a revision policy into the same document as the rate card, on the same day, or you have converted an uncapped-cost problem into an uncapped-work problem.
The reporting swap matters too. A client who loses the timesheet loses their only monthly evidence that they got something. Replace it before they notice it is gone — client reporting for content freelancers covers what actually belongs in that document.
The two objections, and what to say
"How do I know I'm not overpaying now that the work is faster?"
This is the real question and it deserves a direct answer, not a deflection. The honest version: you are paying for a deliverable at a price you have already been paying, and the speed is what protects your timeline and my capacity to take your urgent requests. If the client raises AI specifically — and client-side surveys have shown AI coming up in a majority of agency renewal conversations, with Clutch reporting 61% in 2024 — the productive move is to reframe rather than discount. Talking openly about how the work gets made is usually better received than dodging it; what to tell clients about using AI is the version of that conversation that does not end in a rate cut.
What you should not do is volunteer a discount to pre-empt a request that, in most reported cases, never actually arrives. The fear of the AI discount demand is considerably more widespread than the demand.
"Procurement needs an hourly rate on file."
Common in mid-market and enterprise, and usually solvable. Give them an hourly rate for out-of-scope advisory work only — strategy calls, workshops, platform consulting — and keep production entirely on the deliverable card. Procurement gets a number for their system, production stays on a unit that does not punish you for being fast.
The pieces that come after the switch
Fixed fees only work if you know your own inputs, and hours were doing that job badly but at least doing it. Once the timesheet is gone, the numbers that keep a quote honest are shot counts, attempt counts and credit spend per deliverable. Versely bills in credits with no free allowance behind it, so a month of variants is a real, forecastable cost — tracking credits per client and per deliverable is the ledger that replaces the timesheet on your side of the wall, and the plan and credit pricing page is the input side of whatever rate you land on.
Then pick the unit deliberately rather than by default. Per minute, per deliverable, or retainer walks the same month of work through all three, and fixed-price versus cost-plus quoting covers the case where the brief is too unstable for a flat number and you need a base plus a change-order trigger instead.
FAQ
Should I tell the client the work now takes fewer hours?
Not as a headline. If they ask, answer honestly — deceiving a client about how the work is made is a bad trade for a temporary fee. But leading with "this is faster now" invites exactly one response. Lead with the deliverable, the schedule and the revision cap, which is what they are actually buying.
What if the deliverable-based total comes out lower than their hourly spend?
Then the hourly relationship was carrying padding, admin time or scope creep that a fixed fee will now expose. Fix the bands before you switch, not after — once the card is signed, correcting it upward is a price increase with all the friction that carries, and you have spent your goodwill on the migration.
Can I migrate one workstream and leave the rest on hours?
Yes, and it is often the safest route. Move the most repeatable workstream first — the one where you already know the shape of every deliverable — and leave genuinely exploratory work on hours until you can price it properly. Mixed units are only a problem when both apply to the same piece of work.
Does this work for a client who buys unpredictable one-offs?
Fixed deliverables work, retainers do not. A client whose volume swings from two assets to twenty pays too much in one month and too little in the next under any capacity-based model. Keep them on a per-deliverable card with banded rates and revisit after three months of real data.