29% of agencies now get hourly-rate pushback citing AI productivity gains, and a quarter have already dropped hourly billing for good. Here's the AI repricing system that moves you to fixed-fee without the scope-creep bloodbath.
by Ayush Gupta's AI
The problem
The objection used to be rare and easy to deflect. Now it shows up on renewal calls as a flat assumption: AI makes this faster, so it should cost less. A recent Forrester Consulting study found 29% of agencies report client pushback on hourly rates specifically citing AI-driven productivity gains, and a quarter of North American agencies have already moved exclusively to fixed-fee pricing rather than keep defending an hourly number client-side procurement no longer accepts at face value. The problem is most agencies haven't actually recalculated what a deliverable costs to produce with AI in the workflow. The old hourly estimate is still sitting on the rate card unchanged, which means founders are either defending a number they can't fully justify anymore, or guessing at a new fixed fee with no real data behind it.
The fix
Rebuild the rate card on actual AI-era delivery hours pulled from real time-tracking data, then move the deliverables that are stable enough to fixed-fee, with a client conversation that names the AI productivity question directly instead of dodging it.
The Playbook
Pull real delivery hours per deliverable, not per retainer
Retainer-level hours hide too much to reprice anything safely. Pull the last 6-12 months of time-tracking data broken out by specific deliverable type — a blog post, a landing page build, a monthly report, a campaign launch — so there's an actual hours-per-unit number for each thing clients pay for, separate from the account-management time that shouldn't be in the fixed fee at all.
Feed Claude the hours data to find what's actually safe to fix-fee
Not every deliverable has settled into a predictable AI-era hour count yet. Some have compressed hard and consistently. Others still swing wildly depending on the client or the revision count. Claude should separate those two groups before any new number gets quoted to a client.
Here is time-tracking data for the last 6-12 months, broken out by deliverable type, including hours per instance and number of instances:
[PASTE TIME-TRACKING DATA]
For each deliverable type, calculate:
1. Average hours per instance
2. Variance across instances (low/medium/high)
3. Whether this deliverable is stable enough to safely quote as a fixed fee, or still too variable and better left hourly or capped-hourly for now
Flag anything where variance is high — those are the ones most likely to blow up a fixed-fee quote. Don't round variance down to make more deliverables look fixed-fee ready than they are.Build the new rate card with the AI efficiency gain priced in on purpose
Hiding the fact that AI cut the hours doesn't help once a client can see the same thing from their side. Build the fixed fee off the new, real hours number plus a deliberate margin buffer for scope variance, so the fee reflects actual delivery cost instead of a defensive markup nobody can explain if asked.
Using this hours-per-deliverable data and our target margin of [TARGET MARGIN %]:
[PASTE STABLE DELIVERABLE DATA FROM STEP 2]
Generate a fixed-fee rate card: deliverable, average hours, blended cost, recommended fixed fee, and a short note on what's included versus what triggers a scope change. Build in a buffer for revision rounds based on the variance data — don't price it as if every instance will be the fast one.Draft the client conversation that names the AI question before they ask it
The agencies losing this argument are the ones caught improvising an answer when a client brings up AI efficiency first. Get ahead of it: confirm AI did cut delivery time on specific deliverables, then show exactly where that saved time is going — more revision rounds, faster turnaround, senior review that used to get skipped under time pressure — instead of just a lower number.
Draft talking points for a client call about moving from hourly billing to fixed-fee pricing on [DELIVERABLE TYPES].
Context: AI has reduced our delivery hours on these deliverables by roughly [X%]. We're passing some of that efficiency through to a lower fixed fee, and reinvesting the rest into [faster turnaround / more revision rounds / senior-level review].
Requirements:
- Open by naming the AI efficiency gain directly, don't wait for them to bring it up
- Be specific about where the saved time is going, not just "we got more efficient"
- State the new fixed fee and what's included plainly
- No defensive tone, no over-explainingSet a quarterly repricing cadence instead of a one-time switch
AI-driven hour reductions aren't a one-time event — they keep compounding as tools and workflows improve. Recheck the hours-per-deliverable baseline every quarter and adjust the rate card deliberately on a schedule the agency controls, instead of waiting for a client to notice the gap first and force the conversation on their terms.
What changes
A fixed-fee rate card built on real delivery numbers instead of guesswork, a ready answer to the AI productivity objection before it derails a renewal call, and a repricing habit that keeps pace with how much faster AI actually makes delivery — instead of waiting for a client to raise it first.
The objection used to be easy to wave off. Now it shows up on renewal calls as a flat assumption, stated like a fact: AI makes this faster, so it should cost less.
A recent Forrester Consulting study puts a number on it — 29% of agencies now report client pushback on hourly rates that specifically cites AI-driven productivity gains as the justification. A quarter of North American agencies haven't just absorbed the pushback, they've moved exclusively to fixed-fee pricing rather than keep defending an hourly number procurement no longer takes at face value.
The rate card didn't move when the work did
AI genuinely cut delivery hours on a lot of agency work over the last two years — first drafts, research, reporting, revision cycles. The rate card mostly didn't follow. It's still built off estimates from before the workflow changed, which leaves two bad options when a client raises the AI question: defend a number that's quietly become hard to justify, or guess at a new one with no data behind it. Neither holds up well on a call.
Fix the deliverables that have actually settled, not all of them
Not every deliverable has compressed into something predictable. Some have — a blog draft, a landing page build, a standard monthly report. Others still swing hard depending on the client, the revision count, or how messy the brief was. The fix isn't to fixed-fee everything at once. It's to separate the stable deliverables from the volatile ones using real hours-per-instance data, and only move the stable ones first. A fixed fee quoted on a high-variance deliverable is the fastest way to turn a pricing win into a margin problem three months later.
Price the efficiency gain on purpose, not by accident
Once the stable deliverables are identified, the new fee should be built off the real, current hours number plus a deliberate margin buffer for scope variance — not a vague markdown of the old hourly rate, and not a defensive number padded to protect against the AI question. A fee a founder can explain line by line survives a tough renewal conversation. A fee nobody can explain doesn't.
Name the AI question before the client does
The agencies losing this argument are the ones improvising an answer when a client brings up AI efficiency first. The agencies holding their ground open the conversation themselves: yes, AI cut the hours on this work, here's where that saved time went instead — more revision rounds, faster turnaround, a senior review step that used to get skipped under deadline pressure. That's a different conversation than "why does this still cost the same," and it's one the agency can actually win.
This isn't a one-time switch
AI-driven hour reductions keep compounding as the tools improve, which means the rate card isn't done once it's rebuilt. The agencies that stay ahead of this recheck the hours-per-deliverable baseline every quarter and adjust on their own schedule — instead of finding out the rate card is stale again the next time a client brings receipts to a renewal call.
Bottom line
The AI productivity objection isn't going away, and it's not unreasonable — AI really did change the cost of producing a lot of agency work. The agencies that lose this fight are the ones still pricing off a number from before that was true. The ones that win it are the ones who did the math first, moved the stable work to fixed fees built on real hours, and walked into the AI conversation with an answer instead of a defense.