Start with the number that should decide how you read the rest of this.

Research by the World Federation of Advertisers with Agency Mania Solutions, covering 69 multinational companies across six sectors with roughly $147 billion in combined global marketing spend, found that only 20% of those advertisers have already changed their commercial models with agencies because of AI — while 61% intend to. The WFA has tracked the future of agency remuneration across several editions. ContentGrip's account of the study sets out the figures.

So the repricing is coming, almost nobody has done it, and if you are paying an Indian agency a retainer today you are almost certainly on a structure designed before any of this. We are an agency, we charge retainers, and we are about to argue that the basis of those retainers is wrong — so read this as a commercial argument and test it accordingly. The test is at the end, and it applies to us as much as to anyone pitching you.

What is the Proxy Problem?

Agency fees have always been anchored to effort. Even a flat monthly retainer that bills no hours is usually built from an internal estimate of how many people-days the scope requires. Hours were never what you were buying — you were buying judgement, accountability and output — but hours were the easiest thing to count, so they became the proxy.

Generative tooling broke the proxy, and it broke it unevenly.

The strongest evidence is a pre-registered field experiment by Dell'Acqua and colleagues at Harvard Business School with Boston Consulting Group, running 758 consultants across tasks chosen to sit inside and outside the model's competence. The paper is public. Inside that competence: 12.2% more tasks completed, 22.5% to 27.6% faster, human-graded quality up more than 40%. Outside it, correctness fell by 19 percentage points against a control group that was right 84.5% of the time. Separately, Noy and Zhang's randomised experiment in Science found substantial time reductions with quality gains on standard professional writing tasks.

Read those two findings together and the pricing implication is exact. The tasks that compress hardest — drafting, first-pass research, routine production, variant generation — are precisely the ones that filled the most billable hours. The tasks that do not compress, and actively degrade when delegated, are the senior judgement calls that occupied the fewest.

Which puts both sides of a conventional retainer in an awkward position. The agency on an effort-derived fee now profits from its own inefficiency, because getting faster reduces what it can justify charging. The client on the same fee is funding a cost base that has partly stopped existing. Neither party has to be acting badly for that to be true.

Has the market actually moved?

More than most Indian businesses realise, and the direction is not toward outcome pricing.

From the same WFA research, as reported by VideoWeek:

• Labour-based remuneration is now the main model for under 20% of these advertisers, against 54% in 2011
• 35% use fixed-fee or output-based models; 23% use labour-plus-performance structures
• Asia-Pacific leads on fixed-fee and output-based pay at 44%, against 32% in Europe and 29% in the United States and Canada

That last line matters for an Indian buyer. The region you operate in is already ahead of the West on moving away from time-based fees, which means asking for a different structure is not an exotic request.

There is also a historical precedent worth knowing, because it predicts how this plays out. Research by Horsky and Zeithammer, reported by UCLA Anderson Review, traces the decline of the old 15% media commission. Advertisers challenged it from the 1920s and won a consent decree under antitrust law in 1956 — and the commissions survived into the 1980s anyway. What actually killed the model was competition and unbundling: advertisers' use of unbundled services rose from 18% in 1991 to 47% in 1994, and billings-based compensation fell from 61% of agencies in 1994 to 10% by 2003.

The lesson is not that rules change pricing. It is that competition changes pricing, and it takes about a decade. Anyone telling you the industry will reprice itself quickly out of fairness has not read this history.

Should you ask for outcome-based pricing?

Usually not, and this is where we part company with most of our own industry.

Outcome pricing is the answer everyone reaches for and the evidence on it is poor. From the WFA study: for 67% to 80% of respondents, performance-based remuneration accounts for less than 20% of total agency compensation. And when advertisers were asked what actually drives agency performance, financial incentives ranked as a less important factor at 3.9 out of 6 — behind great briefing at 5.6, respect and trust at 5.2, and high-quality feedback at 5.1.

That is advertisers, not agencies, saying that paying for outcomes is not what makes agencies perform. Meanwhile 63% of the same multinationals expect to increase their use of performance-based fees. Aspiration and evidence point in opposite directions, which is worth sitting with.

Five structural reasons it disappoints, and none of them are about bad faith:

The agency does not control the variables. Your product, your price, your inventory, your sales follow-up and your brand equity determine most of the outcome. Paying an agency on a result it can influence but not determine is paying it for your own operations.

Attribution has to be agreed, and it rarely survives a real funnel. Adoption is highest in media buying precisely because the signal is cleanest there. For brand and creative work, isolating the agency's contribution is close to impossible, and the contract becomes a standing argument about a measurement model.

Priced correctly, it costs you more. An agency absorbing risk it cannot control will charge for that risk. If it does not, it has mispriced and will either renegotiate or underinvest. Clients asking for outcome pricing are usually asking to pay less, which is a different conversation and a more honest one.

It degrades strategy. An agency carrying downside on a monthly number will rationally choose safe, fast, highly measurable tactics — bottom-of-funnel capture, branded search, retargeting — over anything that builds demand slowly. You get the strategy the incentive buys, and it is the short one. This is the failure mode behind a lot of what looks like lead generation going wrong.

In India it has a tax consequence most people miss. Under Rule 33 of the CGST Rules, expenditure an agency incurs as a pure agent of the client — media bought on your authorisation, invoiced separately, over and above the agency's own services — is excluded from the agency's value of supply. The rule's conditions require a contractual agency arrangement, no title held in what is procured, no use of it in the agency's own interest, and recovery of only the actual amount spent plus the agency's fee.

Blend your media spend into a single cost-per-acquisition invoice and you risk losing that treatment, which can bring GST onto the whole amount including the pass-through. Separately, how a fee is characterised can affect which withholding provision applies to your payments. Neither of these is a reason to avoid outcome pricing, but both are reasons to have your chartered accountant look at the structure before you sign it rather than after. This article is not tax advice.

Even the most sophisticated advertisers have struggled here. Coca-Cola launched a well-publicised value-based compensation programme in 2009, asking agencies to move off labour-based fees and tie margin to business results. The difficulty of cleanly attributing business outcomes to marketing interventions is the reason that kind of programme is still discussed as an experiment rather than a template.

So what should a fee be anchored to?

The inputs that did not compress. There are four, and all four are specifiable.

Senior judgement, named. Who exactly is making the calls on your account, how senior, and how much of their attention you have. This is the input that degraded by 19 percentage points when delegated to a model, which makes it the scarce one. A fee that does not tell you whose judgement you are buying is not telling you much.

Accountability. A named person answerable for the work, who will defend a recommendation and own a failure. You cannot automate this and it is a real part of what a retainer has always bought.

Throughput capacity, stated and countable. Not "social media management" but a defined quantity per period: this many assets, this many tests reaching a decision, this many pieces published, this cadence of reporting. Fixed-fee and output-based models are where the market has actually gone, and this is why — it is the one thing both sides can verify without arguing.

Compliance and governance exposure. This is a genuinely new cost line. Reviewing generated output before it ships, maintaining production records, and handling the disclosure obligations now attaching to synthetic advertising creative in India all take competent human time. We covered those obligations in our work on AI-powered marketing, and they should appear in a scope of work rather than being absorbed silently. The same logic applies to search work, where what counts as the deliverable has changed more than most scopes reflect.

Note what this is: a capacity-and-judgement fee, not an outcome fee and not an hours fee. It prices what is scarce, it is countable by the client, and it does not ask either party to pretend they can attribute a quarter's revenue to a campaign.

The Two-Question Test

Here is what to ask any agency, including us, and what the answers tell you.

First: what does this fee buy, in numbers I can count at the end of the month?

A good answer is specific and verifiable — assets delivered, tests concluded, pages published, named senior hours on the account, reporting cadence. A bad answer is a list of service categories. If you cannot audit the deliverable at month end without the agency's cooperation, the scope is decorative. The discipline here is the same one that applies to measuring the work itself.

Second: has your price changed at all as your own production costs have fallen — and if not, why not?

This is the uncomfortable one and it is the more revealing. There are honest answers. An agency may have held price and increased throughput, in which case ask to see the throughput numbers against last year. It may have reallocated hours from production to senior time, in which case ask who, and ask to see the reporting that would show it. It may have added governance and compliance work that genuinely did not exist before, in which case ask what that involves.

What should worry you is an agency that cannot answer, or that treats the question as hostile. Falling costs with a static fee and unchanged deliverables is the thing this entire article is about, and an agency comfortable with the question will have a ready answer.

Two further disclosures worth asking for, because they are reasonable and rarely refused by anyone operating cleanly: who is actually on your account and at what seniority, and whether anything is subcontracted. And on media, confirm that spend is invoiced separately as a pass-through rather than blended into a fee, for the Rule 33 reason above.

What we cannot tell you

There is no published, methodologically sound data on Indian mid-market agency retainer levels, margins, utilisation rates or pricing-model prevalence. The WFA, Gartner and industry-body figures in this article describe multinational advertisers and largely North American and European markets. Anyone quoting you an Indian benchmark is quoting something nobody has measured — which is the same gap we found when looking at Indian creative production costs, and we wrote about what agency pricing actually involves here.

What you can do instead: collect three or four comparable proposals, insist each states throughput in countable terms, and compare on that basis rather than on headline fee. That produces a real local benchmark for your category and budget, which no survey would improve on.

The case against everything above

We would rather state this than have you think of it first.

You may want predictability more than fairness. Retainers exist because both sides want a stable number. Any structure that makes your invoice move is worse for your own planning, and that is a legitimate reason to keep a flat fee even knowing its basis has shifted.

Hours may never have been the point. If what you were buying was access, accountability and insurance against being ignored, then production getting faster does not reduce what the relationship is worth. That argument is stronger than agencies admit, because it is the one that justifies their fee honestly.

The market will do this without a framework. Competition repriced the 15% commission over roughly a decade without anyone adopting a new theory. Your simplest available move is to negotiate the number down and skip the rest of this article. If you can, do that.

Context is working against you, though. Gartner's 2026 CMO Spend Survey of 401 marketing leaders across North America, the UK and Europe found marketing budgets flat at 7.8% of company revenue, with 15.3% of budget now going to AI and 56% of CMOs saying their budget cannot fund their strategy. Money is being pulled toward tooling from somewhere, and agency fees are the nearest adjustable line.

And the obvious objection to us publishing this. An agency arguing for repricing while charging a conventional retainer has a credibility problem, and we accept that. Our position is that the second question in the test above should be asked of us, we should have an answer, and if we do not then the argument fails where it should — in a meeting, not in a blog post.

How we would restructure an engagement

For anyone who wants to act on this rather than file it:

1. Separate the fee from the media. Pass-through invoiced distinctly, for commercial clarity and for the Rule 33 position.
2. Write throughput into the scope. Countable quantities per period, with a definition of what counts as delivered.
3. Name the senior time. Who, how much, and what they are accountable for.
4. Price governance explicitly. Review, records and disclosure compliance as a stated line, not a hidden cost.
5. Leave outcome elements narrow if you use them at all. A bonus on a single clean, agency-controllable metric with an agreed measurement model and a cap — not a results-only arrangement.
6. Review on a fixed cadence against the throughput numbers, not against a traffic chart.

Want this applied to your current arrangement? Send us your existing scope of work and what you pay, and we will tell you what we think the fee is anchored to, which deliverables you cannot currently audit, where media and fee are blended in a way that may cost you, and what we would change. If your arrangement is already sound, we will say so — and if the honest answer is to negotiate your current agency down rather than switch, we will say that too. Talk to Midgrow, or message +91 9632398896. Our digital marketing services page sets out how engagements are structured, and social and content retainers are where throughput definitions matter most.

Frequently asked questions

What pricing models do marketing agencies use?

Nine arrangements with different risk allocations: hourly or day-rate billing, fixed-fee projects, flat monthly retainers, capacity or throughput subscriptions, dedicated full-time-equivalent resourcing, commission on media spend, base fee plus performance bonus, pure outcome or results-only compensation, and revenue or equity share. The important distinction is that several of these are hour-derived in substance even when no hours are billed, because the fee was built from an internal effort estimate.

Is hourly billing dead for marketing agencies?

It is becoming marginal among large advertisers. WFA research across 69 multinationals with roughly $147 billion in combined marketing spend found labour-based remuneration is now the main model for under 20% of them, against 54% in 2011, with 35% on fixed-fee or output-based structures. Asia-Pacific leads on fixed-fee and output-based pay at 44%.

Should I pay my agency on results instead of a retainer?

Usually not as the primary structure. The same WFA research found performance-based remuneration accounts for under 20% of total agency compensation for 67% to 80% of respondents, and advertisers ranked financial incentives as a less important driver of agency performance at 3.9 out of 6, behind good briefing at 5.6. The agency cannot control your product, price, inventory or sales follow-up, and an agency properly pricing that risk charges more, not less.

Why is attribution a problem for outcome-based agency fees?

Because the contract then depends on both parties agreeing a measurement model that assigns business results to marketing activity. That is achievable in media buying, where the signal is relatively clean, and close to impossible for brand and creative work. Adoption of performance fees is correspondingly highest in media and lowest in creative and production.

Does blending media spend into an agency fee cause a GST problem in India?

It can. Rule 33 of the CGST Rules excludes expenditure an agency incurs as a pure agent from its value of supply, provided the payment is made on the client's authorisation, invoiced separately, and is additional to the agency's own services, with the agency holding no title and recovering only the actual amount. A single blended cost-per-acquisition invoice may fail those conditions. Take your chartered accountant's view before structuring it.

What should an agency fee actually be based on?

The inputs that did not compress under AI: senior judgement with named people, accountability, a stated and countable throughput capacity, and the governance and compliance work that reviewing generated output now requires. That is a capacity-and-judgement fee — verifiable by the client, and not dependent on attributing revenue to a campaign.

How do I tell whether my agency fee is fair?

Ask two questions. What does this fee buy in numbers I can count at month end, and has your price changed at all as your production costs have fallen? A specific, auditable answer to the first and a considered answer to the second are both reasonable to expect. Discomfort with the second question is the signal worth acting on.

Are there benchmarks for Indian agency retainers?

No reliable published ones. There is no methodologically sound public data on Indian mid-market retainer levels, margins, utilisation or pricing-model prevalence, and global figures describe multinational advertisers in Western markets. Build your own comparison by collecting several proposals that each state throughput in countable terms.