The Agency Retainer Pays for the Wrong Work

The Retainer Isn't Broken. The Definition of Work Is.

The vast majority of marketing agencies still don't offer performance-based pricing. Not because clients haven't asked for it, and not because agencies are protecting some comfortable racket. It's because paying for outcomes is genuinely hard to design, attribute, and defend when a hundred variables sit between the agency's work and the business result.

But that complexity doesn't excuse what most retainers actually buy: a standing team, a calendar of deliverables, and the reassuring illusion of activity. The 2025 ANA/4A's study found that most agencies' top clients remain in agency-of-record or retainer-based relationships, and that client-agency relationship tenure has doubled since 2016. Long-term relationships are not disappearing. What those relationships are measuring is the problem.

The core argument here is not that retainers are obsolete. A retainer paid for capacity and activity is a structural guarantee that the wrong work gets done.

Why Activity Wins Inside a Retainer

A retainer creates a specific economic logic: the agency must justify its monthly fee. Visible output is the safest justification. Reports get written. Campaigns get variants. Decks get updated. Meetings fill the calendar.

None of that is fraud. But it is a slow drift away from what actually matters.

When the contract rewards availability, prioritization becomes optional. Agencies maintain their teams and fill their hours. If the client's most urgent problem shifts from paid media efficiency to lifecycle email, the retainer structure doesn't automatically redirect effort. The team already has a job description. Changing it requires renegotiation, not judgment.

Patrick Gilbert describes this dynamic directly in Never Always, Never Never, drawing on a case study involving a long-term client of AdVenture Media. Paid media campaigns were performing well and revenue attributable to those campaigns grew on the same budget year over year. By every metric visible inside the agency's scope, things looked good.

What wasn't visible: the company's email list had a large number of subscribers and almost no automation. A consultant brought in later found that email was significantly underperforming relative to what would be expected for a brand of that size and tenure. The CMO eventually resigned. The agency was let go.

Campaigns didn't fail. Scope did. The retainer paid for what it was contracted to cover, and nobody's incentives pushed hard enough to flag what was being left on the table.

That is the structural problem. A retainer can pay very well for the wrong work.

The Measurement Trap

Performance-based pricing seems like the obvious fix. Tie the fee to results. Align incentives.

Numbers suggest this hasn't happened. Among surveyed SEO agencies, 9% use performance-based pricing. ANA compensation research puts 82% of marketers on fee-based models, with 41% layering in some form of performance incentive. Categories overlap because pure performance pricing almost never works cleanly on its own.

Here's why: agencies don't control pricing, product quality, distribution, inventory, sales execution, or macroeconomic conditions. If a brand cuts its conversion rate by cheapening the product and the agency is paid on revenue, who absorbs the loss? If a competitor runs an aggressive promotion during a key sales period, does the agency take the margin hit?

Attribution problems run the other direction too. If an agency is compensated purely on last-click conversions, it will concentrate on the bottom of the funnel. Brand advertising, upper-funnel demand, and creative that takes time to compound get deprioritized. Les Binet and Peter Field's long-running analysis of the IPA DataBank makes clear that brand investment drives long-term growth in ways that don't show up in short-term conversion data. A performance-only retainer creates exactly the wrong incentive structure for brand vs performance marketing.

Pure performance pricing is also wrong. Not because agencies are avoiding accountability, but because clean attribution is often impossible and the incentives warp strategy when it exists.

A 9% adoption rate of performance pricing isn't agency stubbornness. It's a signal that the model breaks down in practice.

The Activity Trap

If performance pricing creates attribution distortions and retainers reward output volume, what's left?

Measuring agency work against client progress, not against agency availability or platform metrics alone, is the answer.

In Never Always, Never Never, the backyard game go-to-market case study illustrates what this looks like. Before spending a dollar on media, the team built a forecasting model: expected reach costs, projected site traffic, required purchase rates. Early indicators weren't proof of profitability, but they were signals of directional progress. When the college audience appeared to underperform in week three, the team didn't move the budget. They looked at the full picture, held the plan, and watched that audience become the biggest revenue driver by week four. Not a single campaign setting changed.

Most agencies would have reallocated the budget. They would have logged the change, reported the decision, and moved on. It would have looked like active, responsive management. It would have been the wrong call.

This is what the book calls a strategy vs tactics distinction in practice. Value wasn't in the media buy. It was in the discipline to not change the media buy when the dashboard looked concerning. That kind of judgment doesn't fit neatly into a deliverables list or a utilization report.

As typically structured, the retainer makes this kind of restraint invisible and undervalued. If you're not doing things, it looks like you're not working.

AI Makes the Problem Worse, Then Better

Generative AI is accelerating the activity trap. Agencies can now produce more outputs, faster, at lower cost. Reports, creative variants, content calendars, audience analyses: all of it compresses in time and labor. If the retainer is still priced around headcount and hours, clients will eventually notice that the same work requires fewer people to produce.

An honest version of this transition is that AI should change what agencies sell. Patrick Gilbert's argument in Never Always, Never Never is that AI is bandwidth: it lets small teams cover ground they previously had to ignore. But that framing only creates value if the agency uses that bandwidth to fill real gaps in the client's marketing program, not to produce more of the same deliverables more cheaply while keeping the invoice the same.

Economic pressure here is real. If AI tools reduce the labor cost of routine campaign management, reporting, and content production, the justification for a large standing retainer built around those tasks weakens. Clients will ask the question eventually. Agencies with good answers will be the ones that shifted from selling labor to selling judgment.

Agencies without good answers are the ones still counting logged platform changes as proof of work.

What a Better Model Looks Like

Abolishing the retainer isn't necessary. ANA data is clear that long-term relationships remain the dominant structure, and there are good reasons for that. Institutional knowledge, strategic continuity, integrated execution: these are real. They take time to build and they're expensive to rebuild after a switch.

What needs to change is what the retainer covers and how progress gets measured.

A more defensible structure separates three things:

• A base fee for strategic access and core capabilities that genuinely require continuous availability
• Fixed fees or project pricing for major initiatives, transformations, or channel expansions
• Incentive components tied to outcomes where attribution is clean enough to be fair to both parties

Layered on top: quarterly reprioritization. Scope should flex with the business, not stay fixed because that's what was written at the start of the year. If the client's biggest constraint shifts from acquisition to retention, the agency's effort should shift with it. A contract that prevents that isn't a partnership. It's a subscription.

Measurement design matters too. How to evaluate marketing attribution is one of the harder problems in the discipline, and most retainer agreements don't address it directly. They measure what's easy: platform metrics, output counts, deliverable completion. They don't define what the agency actually controls, what will be measured to assess impact, and how external factors will be accounted for.

Without that framework, performance conversations become adversarial. Clients point to a number that went down. Agencies list the variables they don't control. Nobody learns anything.

The Honest Conversation Most Agencies Avoid

Retainers pay for the wrong work because nobody wants to have the conversation about what the right work is.

Clients sometimes don't know their biggest constraint. The home decor brand in Never Always, Never Never had a paid media agency, not an email agency, so email sat broken for years. The CMO knew. Budget holders didn't prioritize it. The agency had no mandate to fix it.

Agencies sometimes do know, and stay quiet because raising the issue threatens the current scope. If the most valuable thing you could do for a client is tell them to cut their paid media budget and invest it in rebuilding their lifecycle program, that advice ends your retainer. Most agencies don't give it.

Deeper still: the retainer model creates an incentive not to surface the client's real constraints. Better to optimize what you've been hired to optimize, report the numbers that look good, and let the relationship continue.

A marketing strategy framework that actually serves the client starts with the business problem, not the agency's service menu. What is the actual constraint on growth? Is it awareness, consideration, conversion, retention, pricing, distribution? Answering that question determines what work matters. Retainer scope should follow from that diagnosis, not precede it.

Most retainers work in reverse. Scope is agreed first. Strategy gets built around what's already been contracted.

The Conclusion Is Uncomfortable

Marketing agencies aren't collapsing. Long-term client-agency relationships are becoming more common, not less. But longevity is not the same as value, and the fact that a relationship has persisted for years is weak evidence that it's solving the right problem.

Retainers fail when they measure agency availability instead of client progress. That's a design problem, not a pricing problem. Switching from a monthly fee to a performance bonus doesn't fix it if the underlying scope is still wrong.

Fixing it requires harder work: defining what actually moves the business, being honest when that definition changes, designing measurement that both parties trust, and building contracts that reward the judgment to hold a plan as much as the activity of executing one.

As Patrick Gilbert argues throughout Never Always, Never Never, strategy vs tactics in marketing is ultimately a question of what the work is actually for. Lever-pulling optimized for the dashboard is not strategy. A retainer that rewards it isn't either.