Reprice the lines AI made faster, not the whole retainer. Drafting, variants and report assembly can fall in price or grow in scope, while strategy, optimisation and accountability keep theirs. A client asking for an AI discount should get a choice: a lower fee on the changed lines, or the same fee with more work.
Marketing agency pricing guides mostly list typical retainer ranges drawn from before AI, so I quote none here. The number that matters is your own hours per scope line, because the discount conversation is really a conversation about hours: which ones fell, which didn't, and where the saved time went.
Everything numeric in the worked example is illustrative. Survey figures come from vendor-run or small samples, and each is named and hedged where it appears; every source was rechecked on 4 November 2026.
The method in short: log hours by scope line for a month, split the lines into those AI changed and those it didn't, choose a pricing structure for each group, and write the review step into the scope so that accountability stays priced.
Which lines of a retainer AI actually changed
A retainer is several kinds of work sold as one fee. AI changes the production-shaped kinds (drafting copy, producing variants, resizing assets, assembling reports) far more than the judgement-shaped kinds: choosing what to test, deciding where budget goes, telling a client no. Time-log six weeks by line before you reprice anything, because most agencies overestimate the saving on strategy and underestimate it on reporting.
The table shows a single $4,000 monthly retainer for a client buying social content, articles, a newsletter and paid social management. All figures are illustrative, and the hours after AI include the human review that every deliverable still gets.
| Scope line | Hours before | Hours after | Price before | Price after | Reason |
|---|---|---|---|---|---|
| Strategy, planning and account management | 11 | 11 | $1,000 | $1,000 | AI helps prepare calls; the thinking is unchanged |
| Social content, 12 posts | 14 | 8 | $1,100 | $850 | Drafts and variants are quicker; human review stays |
| Articles, 2 a month | 12 | 7.5 | $800 | $640 | Drafting faster; expert input and editing unchanged |
| Newsletter | 4 | 2.5 | $250 | $200 | Same pattern as articles |
| Paid social management | 8 | 7 | $600 | $600 | Ad variants are quicker; budget decisions are not |
| Monthly reporting | 5 | 2.5 | $250 | $150 | Data pull and first-draft commentary automated; insight is still yours |
| Total | 54 | 38.5 | $4,000 | $3,440 | Hours down 29%, price down 14% |
Hours fell by 15.5, or 29%, yet the price on the reprice-by-line route falls only 14%, because two of the six lines, strategy and paid social, hold their price. Reporting shows the largest percentage drop. If you already automate it, how marketing agencies use AI to automate client reporting shows the steps; if you want the hours evidence for each line, timesheet software for agencies covers budgets and billable hours.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowRetainer, package or per-deliverable: what each does with AI-cut hours
The three structures react differently when production gets faster. A retainer sells access and ongoing judgement, so falling hours expose it to the "why am I paying the same?" question. A package sells defined output for a fixed fee, so falling hours raise your margin as long as scope holds firm. A per-deliverable menu sells units, so unit prices on commodity items will drift down.
| Model | What the client buys | Effect of AI on your costs | Main risk | Use when |
|---|---|---|---|---|
| Retainer | Ongoing expertise and a steady flow of work | Production hours fall; strategy hours don't | The client sees hours fall and asks for a cut | Ongoing paid media, SEO or content programmes that need learning over time |
| Package | A defined set of outputs for a fixed fee | Output per hour rises; margin rises if scope is firm | Scope creep, with the package treated as a menu | Launches, audits, quarterly content bundles |
| Per-deliverable | One unit of output at a unit price | Unit cost falls; commodity unit prices can fall | A race to the bottom on cheap units | Ad hoc work, overflow, a new client testing you |
The strongest structure for the example client is a split. The three production lines become a fixed package of $1,690 (social $850, articles $640, newsletter $200), and the three lines that lean on judgement and optimisation become an expert retainer of $1,750 (strategy $1,000, paid social $600, reporting with insight $150). The total is the same $3,440, but the client now sees two products with two purposes. The package can change shape or price when tools improve, while the retainer stays priced on expertise.
For work outside either, a unit menu keeps ad hoc pricing consistent. These illustrative prices come from the same hours logic: static social post $70, six-slide carousel $140, short-video script and captions $180, 1,500-word article with expert review $320, newsletter $200, monthly performance report with commentary $150. Write the scope for each, including the review step, with the help of an AI scope of work generator with change control.
In a proposal, the change on the page is small. Before: "Social media management, 12 posts a month: $1,100." After: "Social content, 12 posts a month, drafted with AI tools and edited by a named strategist: $850. Extra posts $70 each." The second version sells the review step and states the overage price, which the first never did.
Agencies that charge paid media as a percentage of ad spend face a different problem. The fee doesn't move when hours fall, so there is no discount to give, but the percentage can stop covering the work at low spend and look generous at high spend. Say a client spends $20,000 a month and the agency charges 12%, which is $2,400. If management takes 20 hours before AI and 16 after, the fee stays at $2,400 and the effective rate rises from $120 to $150 an hour. A minimum fee for low spend and a tiered percentage as spend grows keeps the price tracking the work and not only the budget.
Four answers to "where is my AI discount?", with the sums
Assume a loaded internal cost of $45 an hour and $25 a month of AI tools allocated to this client. Before AI, 54 hours cost $2,430, leaving a $1,570 margin (39%) on the $4,000 fee. After AI, 38.5 hours plus tools cost $1,757.50. The options differ in who receives the $672.50 a month that the saved hours are worth.
| Option | Monthly fee | Hours | Margin | Margin % | Client receives |
|---|---|---|---|---|---|
| Before AI | $4,000 | 54 | $1,570 | 39% | The original scope |
| A. Hold the fee, bank the saving | $4,000 | 38.5 | $2,242.50 | 56% | Nothing extra |
| B. Flat 15% discount | $3,400 | 38.5 | $1,642.50 | 48% | $600 a month off |
| C. Reprice the lines AI changed | $3,440 | 38.5 | $1,682.50 | 49% | $560 a month off, shown line by line |
| D. Hold the fee, reinvest 12 hours | $4,000 | 50.5 | $1,702.50 | 43% | 12 more hours of testing and content |
Look at options B and C: they leave you with nearly the same margin in dollars, $1,642.50 against $1,682.50, and both hand the client more than 80% of what the saved hours are worth. If a client's motive is budget, either gives them most of the saving. C is the better version because it tells the client what changed, which makes the next conversation easier: the unchanged lines are visibly unchanged, and the strategy fee is never in play.
Option D is the one agencies overlook. The client pays nothing less but receives 12 extra hours of work on top of the original scope, about 31% more, and your margin ends up about 3.3 points above where it began. It suits a client whose motive is growth rather than cost. Option A is defensible only when the client never asks, and it is the weakest position if they do.
My own order: offer C and D together at renewal and let the client choose. A client who picks D has told you they want results, which is the best retention signal you will get. Spotting at-risk retainers with AI helps you notice the ones who ask quietly instead.
What the surveys say, and how far to trust them
Four sources are worth knowing. Each has limits, which is why none of them sets a price in this tutorial.
- Productive, a software vendor, surveyed more than 180 agencies and published the results in November 2025. Around a third had already been asked the "AI discount" question, and nearly half expected to be. Its second wave, with 174 agencies and consultancies that use Productive, surveyed from 17 to 31 March 2026, said clients "haven't increased the rate of asking for discounts" and called value-based billing "the preferred direction for agencies that are adjusting". Among agencies reporting several positive AI results, 52% were keeping or raising prices with better margins.
- Promethean Research, in 2025 survey data from its 2026 State of Digital Services Report, has marketing agencies leaning towards retainers (45% prefer them) and says value-based pricing fell to 18% of surveyed agencies in 2025 and underperformed more standard models.
- Digiday interviewed seven independent agency executives in August 2025. Billing ranged from hourly, billed to actuals so clients see savings directly, through output-based to outcome-based. One agency's AI-made video and stills for a campaign cost 26% of what live shoots would have, and most said off-the-shelf AI tools weren't yet good enough to create final campaign assets routinely.
- Agency Management Institute, an agency consultancy, asked 400 client decision-makers for its Agency Edge 2026 research what they expect from agencies because of AI. Only 29% said reduced fees, while 89% expect full accountability for AI used on their behalf. The summary doesn't describe how respondents were found, so read it as one survey rather than a benchmark.
The samples are self-selected or small: a vendor's own customers, seven interviewees, a consultancy's survey with undescribed sampling. Read them as signals rather than benchmarks. Discount requests exist but haven't surged, agencies disagree about the right pricing model, and clients care about accountability. None of them tells you what to charge, and the first two cut against the idea that you must switch to value-based pricing in a hurry.
Handling the discount conversation
Swydo's guide suggests four steps in order: concede what AI has made cheaper, clarify the pressure behind the question, show two or three decisions that changed results, then restructure the scope and let the client choose. The order makes sense: skipping straight to "show" sounds defensive, and defending the whole retainer makes clients more suspicious. A reply that follows the sequence:
"Hi [first name], fair question, and yes: drafting, variants and report assembly are quicker for us now. Here is exactly where, line by line. Strategy and paid media decisions haven't got quicker, and that's where your results come from. I'd like to offer two options for the next quarter: a lower fee on the lines that changed, or the same fee with the saved hours put into testing and content. Which suits your goals better?"
Back the reply with a short hours ledger rather than a promise. A month-three example, illustrative: 15.5 hours saved against the original scope; 6 reinvested in testing four new ad angles, 4 in refreshing two older articles, 3 in a landing-page review, and 2.5 held back as margin. The ledger takes five minutes to build from timesheets and answers "where did the time go?" before it is asked.
If the pressure comes from a finance lead rather than your day-to-day contact, offer a call between the two leaders, and give your contact a one-page summary they can forward: what changed, what didn't, and the two options.
What a priced review step has to include
Review is what clients are paying for when they ask agencies to be accountable, so list what it covers and cost its minutes. As an illustration, review of an AI-assisted social post takes about 15 minutes and review of an article with expert input about an hour. Across the example retainer that is 3 of the 8 social hours and 2 of the 7.5 article hours: about a third of production time once AI drafts. Keeping it visible in the scope is what lets you refuse to cut it.
- Every claim, figure and quote traced to a source or to the client.
- Brand voice and terminology checked against the style guide.
- Claims the client can't substantiate removed or softened, especially comparative, health and financial ones.
- Images checked for licence terms and, where the channel or the client's policy requires it, marked as AI-generated.
- No confidential client data entered into tools outside the agreed list.
- A named approver and a date on every deliverable.
Contract wording that keeps review and scope priced
A lower price or a wider scope only holds if the contract says what is being sold. The block below is a filled-in scope line you can adapt.
Scope line: Social content, 12 posts a month (copy and graphics).
Production: drafted with AI tools and edited by a named strategist. Every post is approved by [client contact] before publishing.
Pricing basis: fixed monthly fee for the quantity above. Extra posts at $[unit price] each.
AI disclosure: we use AI tools for drafting and variants. We do not enter client confidential data into free consumer accounts. A list of tools is available on request.
Savings review: every six months we compare hours logged with the scope and agree whether to adjust the scope or the fee.The "named strategist" and approval lines keep accountability visible, which matters if clients do expect agencies to answer for how they use AI. The unit price for extras handles the habit AI creates, where more variants feel free to ask for. The savings review gives the discount conversation a scheduled home. For fuller drafting of disclosure, ownership and approval terms, see AI clauses for agency contracts.
Absorb the AI tools or pass them through?
Most agencies fold tool costs into the fee, and the sums usually justify it. A five-seat team on ChatGPT Business at the monthly rate of $25 a seat pays $125 a month, or about $10.42 per client across twelve clients; on annual billing at $20 a seat it is $8.33. Even adding a design tool and image credits, an allowance of $25 per client a month, as in the example above, is generous.
Set a written policy and apply it consistently: tools are absorbed under a stated monthly allowance per client, and anything beyond it, such as large volumes of video-generation credits, is passed through at cost with the client's prior approval. What a five-person agency team pays for AI tools each month lists a realistic stack to cost from.
When to hold firm and when to walk away
Hold firm on the lines where AI changed nothing. A client who wants the strategy fee cut because "you use AI" is asking to pay less for judgement, and the honest answer is that the judgement still takes the same time.
Walk away, or at least repackage, when a client:
- expects senior accountability at the price of a tool subscription;
- wants outcome-based pricing but won't share the data needed to measure the outcome;
- wants unreviewed AI output published under your agency's name;
- asks you to take all the risk of AI use, such as ownership or accuracy claims, while taking all of the savings.
In each case the price isn't the real problem; the scope is. Offer a smaller package that fits the budget, or a self-serve version, and keep the relationship open.
A quarterly check that your pricing still fits
Pricing set after one good month ages quickly, because tools keep improving and clients keep comparing. Once a quarter, work through five measures per client.
- Effective hourly rate: monthly fee divided by hours actually logged, compared with your target.
- Margin by scope line, so a fading line is spotted before it drags the total.
- Hours against scope, to catch scope creep early.
- Where saved hours went, from the hours ledger.
- Renewal signals: unprompted questions about price, requests for in-house training, and a drop in approvals turnaround.
If a line's effective rate keeps climbing while the client is happy, you have room to expand that line. If it falls, check whether the scope has crept before you blame the tools.
Agency pricing questions after AI
Should I tell clients when AI cuts my production time?
Yes, in general terms, and show where the saved time went. A client who learns about it from a rival's pitch assumes you have been keeping the margin. An hours ledger that shows what changed and what was reinvested turns a fee argument into a scope discussion.
Is value-based pricing the answer to AI discounts?
Not by itself. Promethean Research's 2025 survey data had it at 18% of surveyed agencies and underperforming more standard models, while Productive's March 2026 survey of its users calls it the preferred direction for agencies that are adjusting. Price outcomes where you can measure them, and keep a retainer for ongoing work.
How do I stop a package turning into endless deliverables?
Write the quantity and the review steps into each scope line, price extra units from a menu, and review hours against scope every six months. Scope creep gets worse when AI makes extras feel free, because clients ask for more variants. Quote the overage unit price up front.
Should I charge extra for AI tools?
Most agencies fold tool costs into the fee. A five-seat team paying $100 to $125 a month for assistants, spread across a dozen clients, is under $11 each, too small to itemise. If tools become a large cost, such as video generation credits, pass them through at cost under a written policy.
What if a client wants to bring the work in-house with AI?
Don't fight it. Offer a split where their team owns daily production and you provide strategy, testing and quality control as a smaller retainer. Help them price the in-house route in full, with tools, salary, review time and ramp-up, so the comparison is honest.
Further reads
- Consulting Pricing Models After AI: Fixed Fee, Retainer or Value? — Fixed fee, retainer or value pricing for advisers
- Agency Utilisation Rate: Track Billable Hours and Spot Overload With AI — Track billable hours and spot overload by person
- How to Choose a Marketing Agency That Uses AI Responsibly — The client's side of the same conversation
- Which Agency Tasks AI Handles Well, and Which It Doesn't — Sort agency tasks into AI-ready and human-only
- Website Maintenance Plan: Packages a Web Agency Can Add With AI — Package recurring work for another kind of agency
- AI Content Workflow for Agencies: From Brief to Approved Draft — Brief to approved draft, with review steps built in
- AI Media List Tools: Muck Rack vs Semrush AI PR Toolkit vs ChatGPT — A repeatable test: build one media list three ways, check every name, and score Muck Rack, Semrush AI PR Toolkit and ChatGPT on what you find.
- Best AI UGC Video Editors for Marketing Agencies: Costs and a One-Brief Trial Plan — One client brief, one scoring sheet and a cost-per-video sum for HeyGen, Creatify, Arcads and Synthesia, plus what you must disclose.
- Video Production Pricing Sheet: Build a Rate Card With AI — A rate card for video work built from your own costs: day rates, edit hours, revision rules and licence fees, with a worked brand-film quote.
- Virtual Assistant Pricing When AI Does the Drafting: Rates and Packages — How a virtual assistant can price hourly, by package or on retainer once AI cuts drafting time, with worked sums and a repricing message.
- AI Media Planning for Small Agencies: Channel Mix and Budget Splits — A prompt for a first-draft media plan, a worked $30,000 split, and five checks that catch bad sums, false sources and channels that cannot spend.
- ChatGPT Prompts for a Marketing Agency: Briefs, QA and Client Updates — Eight agency prompts for briefs, QA and client updates, with sample outputs, a QA checklist and rules on which client data stays out.
- Consulting Utilisation Rate: Track It Weekly and Price Projects With AI — Billable versus available hours, bench time, a weekly AI roll-up from timesheets and a worked project price checked against your rate card.
- Content Automation With AI: A Weekly Pipeline for a Small Agency — A Monday-to-Friday content pipeline for a small agency: briefs, AI drafts, human edit, client approval gates and scheduling, with a quality checklist.
- How Does Claude Help a Marketing Agency? Six Jobs and Their Limits — Six agency jobs Claude does well, the limit of each, a test you can run on your own brief, and plan costs against ChatGPT.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Productive agency surveys (November 2025 and March 2026), Promethean Research pricing-model posts, Digiday (August 2025), Agency Management Institute's Agency Edge 2026 summary and a Swydo guide (October 2026), rechecked on 4 November 2026. Sums are illustrative.