6 min read
Should you lower your fee because you used AI?
Count the whole job before you answer the client's discount request.
An agency should not lower its fee simply because it used AI. Start with what the client agreed to buy, then count the whole job: preparation, production, review and revisions. A faster draft can justify a new conversation, but it does not calculate the answer.
For the owner of a small agency, the useful question is what actually changed in delivery. The example below gives you a way to check before replying to a discount request.
What did the client agree to buy?
The agreed basis of the fee is the starting point. Hours worked, a set of deliverables and an agreed business result are different things to price.
If you bill actual hours, record actual hours. Do not invoice for time you did not spend because an earlier version of the process took longer. Check any agreed minimums, retainers and expenses separately.
For a fixed deliverable, compare the work with the agreed scope, quality and revision allowance. Faster production alone does not tell you whether the fee should change. Nor does it excuse ignoring a promise to share efficiency gains.
The IPA's pricing playbook announcement, published on 12 March 2026, describes models based on inputs, outputs, outcomes and combinations of these. It makes the choice dependent on the work and the relationship, including risk, scope stability and whether results can be measured.
Changing the model is a separate decision from answering today's email. Keep the current agreement in view, and record any proposed change clearly.
How much time did the whole job save?
Compare completed work at the same quality, including the time spent checking it. A reduction in drafting time is only one part of that calculation.
Workday's January 2026 research surveyed 3,200 full-time active AI users at companies with at least US$100 million in annual revenue. Nearly 40% of reported time savings went into rework. That is not a small-agency estimate, so do not use it as your own cost assumption.
Take a hypothetical campaign adaptation: the same deliverable, deadline, quality standard and revision allowance before and after introducing AI. These hours are invented to show the calculation. They are not a benchmark.
A faster draft, measured across the whole job
Illustrative hours for one campaign adaptation. The deliverable and required quality stay the same.
| Stage | Previous process | AI-assisted process |
|---|---|---|
| Preparation | 1 hour | 1.5 hours |
| Drafting | 4 hours | 1 hour |
| Review | 2 hours | 2.5 hours |
| Revisions | 1 hour | 1 hour |
| Total | 8 hours | 6 hours |
Drafting falls from four hours to one, a 75% reduction. Total work falls from eight hours to six, a 25% reduction. Neither percentage includes software costs or differences in the cost of the people doing the work.
Under actual-hour billing at an unchanged rate, the labour charge would reflect six hours rather than eight. Under a fixed fee, the example informs the next pricing discussion. It does not produce an automatic discount.
If senior review replaces junior drafting, fewer hours may produce a smaller cost saving than expected. If quality falls, the comparison is not yet valid.
Use the five agency workflow assessments to identify where review belongs. For a first trial, the small-team adoption sequence explains how to establish a before-and-after measure.
When does a lower fee make sense?
A lower fee is worth considering when it reflects an agreed billing basis, a demonstrated efficiency or a deliberate change to the offer. Decide which applies before choosing a number.
For a repeatable fixed-fee service, a sustained reduction in total cost may support a more competitive offer. Check enough completed jobs to distinguish a repeatable improvement from an unusually easy brief.
A smaller budget can also mean a smaller scope. Name the deliverables, revision rounds or turnaround commitments that would change. Avoid giving away extra work simply because one stage became quicker.
The following table is a decision aid. It does not interpret your contract or prescribe a market rate.
What to check before changing the fee
Choose the row that matches the agreement and the evidence.
| Situation | Check first | Practical next step |
|---|---|---|
| Actual-hour billing | Actual billable time and agreed terms | Reflect the hours worked in the labour charge |
| Fixed deliverable, uncertain saving | Scope, quality and total cost | Measure completed jobs before proposing a new price |
| Fixed deliverable, repeatable saving | Cost evidence and any gain-sharing agreement | Consider a revised offer or agreed share of the saving |
| Client needs a lower budget | Which commitments can change | Propose a smaller, clearly defined scope |
| Client wants more for the same fee | Capacity and additional review work | Agree a bounded addition only if delivery supports it |
What should you say to a client asking for an AI discount?
Explain which part of the work became faster, what still needs doing and how the agreed fee works. Use the client's actual job as the evidence.
For the hypothetical fixed-fee example, a reply could read:
A reply grounded in the work
AI reduced drafting time on this job. Across preparation, drafting, review and revisions, the total fell from eight hours to six.
The agreed fee covers the completed adaptation and its review. We can discuss sharing a repeatable saving on future work once we have checked the full costs.
If the immediate problem is budget, I can also price a smaller scope.
That wording is an example, not a script to send regardless of the facts. Replace the hours, scope and proposed next step with what you can support. If you bill actual hours, explain the reduced hours on the invoice instead.
The conversation need not become an argument about whether every agency is discounting. Productive's 2026 pulse report surveyed 174 agencies and consultancies using its software between 17 and 31 March. It found no increase in discount pressure over the preceding six months and no consensus on pricing.
That sample skews towards digitally mature agencies and is not representative of the whole market. It supports caution about sweeping claims, not a reason to dismiss your client's question.
What to record before the next quote
Record the full delivery process so the next quote uses evidence you can explain. Keep the initial scope beside the time and cost records, rather than comparing two jobs that only share a title.
For each completed job, record:
Time spent preparing inputs and making the draft.
Time spent reviewing, correcting and handling revisions.
Who did that work and its cost.
Relevant tool costs.
Whether the agreed quality, deadline and scope were met.
Separate a one-off setup cost from recurring delivery work. Check the time records with the people who performed the work before treating the result as a saving.
What to do next
Take one recently completed job and reconstruct the whole process with its owner. Compare the total with the estimate, then write down what that evidence changes about the next quote.
If you need a repeatable process to measure, ViraOps's Monthly Workflow Playbooks set out agency tasks in stages, with prompts and expected outputs. Use that structure to make the work visible before the next client asks what AI changed.
Use ViraOps's Monthly Workflow Playbooks to give the team a repeatable process, with each stage, prompt and expected output laid out.
Written by Catarina Mestre, Founder of Viravesso. Over a decade in creative operations across agencies and studios in the UK and Portugal, now writing practical AI guidance for small marketing and creative teams.
Common questions
Productive's March 2026 survey found no increase in AI discount pressure over the preceding six months. Its 174 respondents used Productive, so the result is not a market-wide estimate. For your own fee decision, use the agreed billing basis and completed-job evidence rather than assuming every competitor has cut prices.
There was no consensus in Productive's 2026 pulse survey. The IPA also describes several pricing approaches rather than a single answer. For your agency, start with the stability of the scope and what you can measure. Changing the model needs an explicit discussion about responsibilities and risk, not just a new label.
Faster drafting alone does not establish a profit improvement. Count preparation, review, corrections, revisions and relevant tool costs, then consider the cost of the people doing the work. The hypothetical example reduces total time by 25%, but that is not a measured cost saving or a forecast of your margin.
Read next
31 August 2026 · 6 min read
Is AI cutting agency jobs? What the data actually shows
31 August 2026 · 9 min read
The five AI workflows that hold up in agencies, and where each breaks
