My verdict on hourly billing for AI-assisted work is simple. It won’t survive the first client who watches a two-day job come back the same afternoon. That client was paying for a finished result all along, which is exactly what outcome based pricing charges for, and I’d change your price to say so before the minutes turn into an argument.
For Calgary professional firms that still quote by the hour, this is the conversation waiting at the next client meeting. Statistics Canada now counts 19.2% of Canadian businesses using AI to produce goods or deliver services, so more clients can time the work themselves.
The hourly rate made sense when time was the honest measure of the work. A job took ten hours because ten hours of a person is what it cost, and the client could picture every one of them. AI breaks that picture. The work still takes judgment, checking, and a name on the line. It just no longer takes the hours your quote was built on, and both sides of the invoice can see it.
Why does the hour fall apart as a unit?
Because the client can now time the work. When a draft arrives twenty minutes after the call, nobody believes it carries a day of labour, and a price built on hours invites the client to argue about speed instead of value.
You end up defending how long something took, which is the weakest ground a seller can stand on, because the honest answer keeps getting shorter. Your clients are prompting the first draft themselves at their own desks now, so they have a fair idea what a first pass costs in minutes. They are not guessing anymore.
The clients who push on your hours are not always being difficult. Some of them ran the same math you ran when you bought the tool. They have a fair sense of what the work costs you now, and an hourly quote reads to them like a price still set for the old way of working.
Nobody ever wanted your hours. They wanted the problem gone.
What did one vendor see early?
That buyers were going to stop paying for effort and start paying for outcomes. One large software vendor that sells to service teams made the jump before clients forced it. The charge stopped being about seats and time and became a price on the finished resolution itself, one case closed and settled.
An outcome price can start leaner than the old arrangement while the work gets rebuilt underneath it. The vendor moved anyway, on a plain reading of where pricing was heading. A finished-result price also helps when a lean team quotes work above its weight, because the client is buying the outcome and never counts heads.
A seller who prices the result early sets the terms. A seller who waits gets the terms set for them.
What does one finished result cost you?
More than the tool bill and less than you fear, but you have to count it before you can price it. Take one service you sell again and again. Count the minutes you and your staff spend setting the work up, running it, and checking the output.
Count the tool costs for that one job. Count the review pass, the corrections, and the time spent sending the finished piece out under your name. That stack, divided by the finished pieces it produces, is your cost per finished result, and it is the floor under any outcome price you quote. I made the same all-in count from the buying side in the finished job, where the point was knowing what an AI job costs before you pay for it. The seller’s side runs the identical count, pointed at your own price list.
You cannot price a result you have never costed.
How do you quote a result instead of a rate?
By naming the finished thing and what counts as finished. A quote for a resolved case, a completed report, a filing ready to send, each with the review and the corrections inside the price, reads clear because the client can picture exactly what lands on their desk.
That is outcome based pricing in plain clothes. Pick one repeatable service and put a single outcome price on it while the rest of your price list stays as it is. The client stops timing you, because there is no clock left in the quote. I’d start with one service, not the whole list, so you learn on a single quote.
The cost stack for one finished result has four lines.
| Cost line | What to count |
|---|---|
| Set-up | Staff minutes to prepare the job |
| Run | Tool cost for that one job |
| Check | Review pass and corrections |
| Send | Time to send it out under your name |
My take: the hour was always a stand-in for value, and AI exposed it.
Which service should get an outcome price first?
Take the most repeatable service on your list. Work out the true cost of one finished resolution, time, tools, and review counted together, and write the number down. Then put one outcome price on the next quote for that service and watch what the client asks about.
They will ask what finished means, who checks the work, and what happens when something goes wrong. Those are the questions a seller wants, because they are about trust instead of minutes. Repricing a service list around finished results is AI Strategy Roadmap work.
Common questions about outcome based pricing
What is outcome based pricing?
Charging for a defined finished result instead of the hours it took. A resolved case, a completed report or a filing ready to send, each with review and corrections inside the price. The client pictures exactly what lands on their desk, so there’s no clock left to argue about.
How do you set an outcome based price?
Cost one finished result first. Count the staff minutes to set up, run and check the work, the tool costs for that job, and the review and correction time, then divide by the finished pieces it produces. That’s your floor. Price above it on the value to the client.
Why is hourly billing breaking down with AI?
Clients can now time the work. When a draft arrives twenty minutes after the call, an invoice built on hours invites an argument about speed instead of value. Many clients use the same tools and know roughly what a first pass costs in minutes.
Which service on your list could carry one outcome price on the next quote? Start there, and I’ll keep working through pricing shifts like this in the AI Newsletter with ZAK.
2 Comments