The short version
- A retainer prices a cadence. We ship on every account every week, and that rhythm is what the number pays for.
- Weekly delivery gives a year roughly fifty chances to correct course. A monthly rhythm gives twelve.
- The floor exists because overhead per account is fixed. Onboarding, the weekly call, reporting and the pre-publish check cost roughly the same at any fee.
- When the number has to come down, take out workstreams and keep the cadence. A removed workstream is reversible. Cutting the review pass instead shows up later as a retraction.
A retainer prices a cadence. Every account we run gets something shipped every week, and that weekly rhythm is what the number pays for. Most difficult retainer conversations are two people using different units: one side counting hours or deliverables, the other holding weeks of capacity.
It gets easier once both sides are pricing the same thing. It does not get comfortable.
What does a retainer actually buy?
Held capacity on a fixed rhythm. Something ships on every account every week, and the week is the unit. Which workstream fills it can move. The week does not, and guaranteeing that is what costs money.
The rhythm does work that hours do not. A weekly cycle produces roughly fifty points in a year where a client can look at something real and say that is wrong. A monthly cycle produces twelve, and a wrong assumption survives four weeks of production before anybody sees it.
The shapes an engagement gets priced in, and where each one gives way:
| Pricing shape | What the client buys | What varies | Where it fails |
|---|---|---|---|
| Retainer | A held cadence and a standing team | Which workstreams run inside it | Scope grows quietly until the cadence slips |
| Project fee | A defined outcome with an end date | Nothing after signature | Discovery finds work nobody priced |
| Hourly | Time | Everything | Both sides watch the clock instead of the work |
| Per deliverable | A countable unit | Quality inside the unit | Thinking time is unpaid, so it stops happening |
| Performance share | A cut of the result | Attribution | Neither side controls the other side’s sales team |
Why hold a floor at all?
Because overhead per account does not scale down. Onboarding, the weekly call, reporting and the review before publish cost roughly the same whatever the fee is. Under some number the fee funds the overhead and the delivery work is whatever survives.
That overhead is not optional on our side. On one account 44 content pages sit behind an automated validator that has to pass before anything publishes, and every number in a draft is traced back to a source file in a separate pass. Neither gets cheaper because the retainer is smaller.
What happens when a client asks for a lower number?
We take out workstreams and hold the cadence. Fewer things ship each week, the week still ships, and the client can see exactly what came out and put it back when the budget returns.
What does not flex is the check before publish. Removing a workstream is visible and reversible. Nobody sees a review pass disappear until something goes out wrong, and the validator on that 44 page account was written after the fact, which is the normal order and the expensive one.
How do you make the renewal conversation boring?
By making the record structural rather than reported. Each client has a portal with their own access, so what shipped is visible without anybody assembling a summary the week before a renewal.
Our updates carry status and stop there. Where a decision belongs to the client we put the options and the trade in front of them and leave it. Renewal then reviews a record both sides already hold.
A weekly cadence gives a year roughly fifty chances to correct course. A monthly one gives twelve.
When is a retainer the wrong shape?
When the work has a real end. A migration or a rebuild. Price those as projects. A retainer stretched over finite work quietly becomes a maintenance fee nobody decided to buy.
The other case is an account with nobody who can approve weekly. A cadence needs a counterpart with authority. On one engagement two approvals sat with people who had never been in the kickoff, one on security review and one on pricing sign-off, and the launch date moved twice for reasons unrelated to delivery.
The two decisions sitting underneath any retainer number are why we set a minimum in the first place and why the review pass is the last thing to cut.
Frequently asked questions
What does a marketing agency retainer actually pay for?
Held capacity on a fixed rhythm. The retainer reserves a team and a recurring delivery slot, and the workstreams inside that slot change without the price changing.
Why do agencies have a minimum retainer?
Because overhead per account does not scale down. Onboarding, reporting, the weekly call and the pre-publish review cost roughly the same whatever the fee is. Under a certain number the fee covers overhead and the delivery work is whatever survives.
Should a marketing retainer be priced by the hour?
No. Hourly pricing makes both sides watch the clock rather than the work, and it charges the client more when the agency is slower at something. Price the cadence, then hold it.
What should an agency do when a client asks for a discount?
Remove workstreams and keep the weekly cadence. A removed workstream is visible and reversible. Cutting the rate while promising the same scope means the review passes get cut instead, which surfaces later as work somebody has to retract.
Is a retainer or a project fee better for marketing work?
A retainer fits work that compounds and needs correcting weekly, such as content, campaigns and lifecycle. A project fee fits work with a real end date. Run a finite project on a retainer and it becomes a maintenance fee nobody decided to buy.
How do you tell whether a retainer is delivering?
Look at what shipped each week rather than what was summarised each month. A per-client portal showing the actual deliverables makes that checkable at any time, and a weekly rhythm gives fifty-odd points a year to object to real work.
Sources
- Chua Network delivery data across 8 client accounts (internal fact bank)
- Chua Network engagement records, anonymized (internal experience bank)
