Retainer vs Project vs Performance: Agency Pricing

There is no best agency pricing model. There is a right one for a specific engagement, and picking wrong burns either the agency or the client every time. The three live options are the retainer, the fixed-price project, and performance pricing. Each aligns incentives differently, and each fails in a predictable way. If you understand the failure modes, you can pick correctly in about five minutes.

What is the difference between retainer, project, and performance pricing?

A retainer is a recurring fee for ongoing access and capacity. The client pays monthly, the agency reserves time, and the work is continuous. A project is a fixed price for a defined deliverable with a start and an end. Performance pricing ties some or all of the fee to a result: leads, revenue, ranking, whatever both sides agree to measure.

The core question underneath all three is the same. What are you actually selling? Access, output, or outcome. Get that answer straight and the model picks itself. I argued the death of the lazy version of this in the marketing retainer is dead, and this post is the practical follow-up.

When does a retainer make sense?

A retainer fits ongoing work where the value is the relationship and the accumulated context, not any single deliverable. Brand stewardship, always-on demand gen, content programs that need someone who knows the account cold. The agency's memory of the account is the asset. Rebuilding that every project would be waste.

The failure mode is the lazy retainer. The client pays every month, the work drifts, and nobody can point to what the last three payments bought. This is what killed the retainer's reputation. A retainer only earns its keep when it is tied to a standing scope and a visible cadence of output. If you cannot say what this month's fee produced, the retainer has rotted.

Rule: use a retainer when the work is genuinely continuous and the context compounds. Attach it to a defined program, not a vague promise of availability.

When does project pricing make sense?

Project pricing fits bounded work with a clear finish line. A brand system. A website. A campaign with a defined run. The deliverable is concrete, the scope is knowable up front, and both sides can agree on "done."

Fixed price protects the client from runaway hours and rewards the agency for being fast and good. If I can deliver a brand system in three weeks instead of eight because I have done it fifty times, I keep the upside. That is the correct incentive. Efficiency should pay the operator who earned it.

The failure mode is scope creep. Fixed price only works with a fixed scope. The moment requirements drift, someone eats the cost. The fix is boring and non-negotiable: write the scope down, define what is out, and treat additions as new projects. A clear boundary is a favor to both sides.

When does performance pricing make sense?

Performance pricing fits when the outcome is measurable, attributable, and mostly inside the agency's control. Lead volume, booked calls, tracked revenue from a specific channel. When all three of those are true, tying the fee to results aligns everyone perfectly.

They are rarely all true. Attribution is messy, the client's sales team touches the outcome, and the market moves. That is why I treat pure performance pricing as the exception, not the default, and why I broke down the traps in performance-based agency pricing. Use it when the measurement is clean. Avoid it when you would be betting your margin on variables you do not control.

How do I actually choose between the three?

Run three questions in order.

• Is the work continuous or bounded? Continuous points to a retainer. Bounded points to a project.
• Is the outcome measurable and in your control? If yes, layer performance in. If no, do not.
• What is the client anxious about? Runaway cost points to fixed project pricing. Wasted spend points to performance. Needing a reliable partner points to a retainer.

Most healthy engagements are hybrids. A project to build the brand system, then a retainer to run it, with a performance kicker on the one channel where attribution is clean. Blending is not a cop-out. It is matching each part of the work to the model that fits it.

The mistake is picking a model out of habit and forcing the work to fit. Decide what you are selling first: access, output, or outcome. Then pick the model that prices that thing honestly. That is the whole discipline behind how I price work at Girard Media, and it is why clients always know what they are paying for.