The agency retainer model: what a retainer buys, what it should not, and when to sell something else

Type "agency retainer" into Google and it finishes the sentence with meaning, fee, model and agreement. That was autocomplete on 9 October 2026.

Further down the same list sits "marketing agency retainer roi issues".

That last one is the honest suggestion. People search for what a retainer is, and then some of them search for why theirs is not paying off.

Here is the complication. I tell first-time founders not to lead with a retainer, and yet it is the product almost every agency wants, because it is the only one that lets you plan next month's payroll.

Both are true. The retainer is a good product sold badly, most of the time, by agencies who never wrote down what it buys.

So this is my version of the model, from the operator side.

Step 1: Say what a retainer is in one sentence

An agency retainer is a fixed fee, usually monthly and paid in advance, for a named list of deliverables, reserved capacity and a regular review.

That is the meaning. The "retainer fee" is the monthly number.

Notice what the sentence does not say. It does not say "access to the team", "as much as you need" or "hours".

A lawyer's retainer often works differently, as an advance the firm draws down against time. Agency retainers borrowed the word and changed the deal. The client is paying to keep a slot open and a list of work moving, every month, without renegotiating each piece.

If you cannot finish that one sentence for your own retainer, the client cannot either. And a fee nobody can explain is the fee that gets cut first.

Step 2: Name the deliverables, by count

The first thing a retainer buys is output with a number on it.

For example: eight short videos a month, captions and titles for each, posted to two platforms, one monthly report. The numbers are yours to set. The point is that they exist.

A named list does two jobs. It tells the client what arrives. And it tells you when the month is done.

Without it, the month is never done. Every request looks like it might be included, and the polite answer is always yes.

VALORAE Media does this in public. Its FAQ, read on 9 October 2026, describes the Full Page tier at $2,000 to $3,500 a month and lists what it covers: every platform, the calendar, strategy, repurposing and comments. A buyer can read the scope before the call.

Step 3: Reserve capacity, and say what happens to the unused part

The second thing a retainer buys is a reserved slot.

The client is not only paying for the eight videos. They are paying for the fact that when they send footage on a Monday, somebody is already waiting for it. That availability is real cost to you, because you turned other work down to keep it.

So write down what happens when the client does not use it.

I prefer a simple rule: unused capacity does not roll over past the month. If the client sends nothing, the slot was still held, and the fee still covered holding it.

Some agencies allow one month of rollover. That is fine. What breaks the model is unlimited rollover, because three quiet months turn into a fourth month containing four months of work.

Step 4: Put a review on the calendar

The third thing a retainer buys is a rhythm.

A monthly review, on a fixed date, where you look at what shipped against the list and decide what changes next month.

This is where the "roi issues" search comes from, I think. A retainer without a review drifts. The client stops seeing the work as a list of deliverables and starts seeing it as a monthly charge.

Nobody checks whether the list still makes sense. Then one day the charge looks expensive and nobody can say why it was worth it.

The review is cheap insurance against that: thirty minutes, one page, the same questions each time. What shipped, what it did, what we change.

Step 5: Refuse open-ended hours and unlimited revisions

Unlimited is a lie.

Every agency that sells unlimited revisions has a limit. It is just not written down, so it gets discovered in an argument instead of in a contract.

Open-ended hours fail the same way. "Up to forty hours a month" sounds generous until the client asks what was done in hour thirty-eight, and you are now defending a timesheet instead of a deliverable.

So a fair retainer names the revision rounds per piece and prices anything beyond them. I would set one or two.

It sells output, not time. Clients understand "eight videos". Nobody enjoys auditing hours.

This also protects the margin. I wrote about where agency margin comes from: finding the client, guaranteeing a standard, and carrying the cost when work goes wrong. Unlimited revisions quietly hand the third one back to you with no price attached.

Step 6: Write the clauses that keep it fair

A retainer agreement is mostly a list of questions answered before anyone is annoyed.

These are the ones I would not sign without:

• Deliverables per period. The named list, by count.
• Revision rounds. How many per piece, and the rate for more.
• Unused capacity. Rolls over or does not, and for how long.
• Out-of-scope work. A rate, or a rule that it becomes a separate project.
• Payment timing. In advance, at the start of the period.
• Notice to end it. How much, from either side, and what the last month includes.
• Price review date. When the fee can change, so a rise is a scheduled event.
• Ownership and handover. Who owns the finished files, the project files and the accounts, and what gets handed back when it ends.

The price review clause matters more than people think. It turns a future price rise into a date both sides agreed to, instead of a surprise. I covered the rest of that in how to tell clients you are raising prices.

On the client side, VALORAE Media's guide to seven clauses to read before signing a social media agency contract covers ownership and leaving in more depth. Read it from the buyer's chair once. It makes your own agreement better.

One thing a fair retainer does not need is a long lock-in. The same Media FAQ says there are no contracts to sign, and that a client can pause or wind down.

That is a deliberate choice. A retainer that only survives because the client cannot leave is a retainer that should end.

Step 7: Know when to sell something else

A retainer is the wrong product more often than agencies admit.

Sell a project fee when the work has a finish line. A website, a launch, a one-off campaign. Wrapping finite work in a monthly fee confuses everyone about when it is finished.

Sell a per-unit price when volume moves month to month. If the client wants three videos one month and twelve the next, a per-piece rate is honest and a retainer is a guess. Media's Edit tier works that way: priced per video and quoted on the call.

Sell a project first to a new client. A retainer asks a stranger to trust you for an open period. A short paid project asks them to trust you until it ships.

Convert to a retainer once the work has a monthly shape and both sides have seen it delivered.

The test I use is simple. If I can write next month's deliverable list today, without guessing, it is a retainer. If I cannot, it is something else.

I do not think any agency gets this right every time, and I would not trust one that claims to. When a monthly fee starts to feel wrong, the fix is the same: stop, write the list, and change the product if the list will not hold still.

The short version

• Define your retainer in one sentence: fixed fee, named deliverables, reserved capacity, a review.
• List the deliverables by count, every month.
• Decide what happens to unused capacity before the first quiet month.
• Hold a monthly review on a fixed date.
• Never sell unlimited revisions or open-ended hours.
• Write the clauses down, including a price review date and a handover.
• Sell a project fee or a per-unit price when the work does not repeat.

I write these operator notes at tgsidd.com.