What actually determines whether work should be a retainer or a project?
Whether the value compounds or completes. Work compounds when doing more of it keeps producing more return: content that keeps ranking, campaigns that keep learning, a codebase that keeps needing maintenance as it grows. Work completes when there’s a defined end state after which the return from doing more of the same activity drops sharply: a migration is either finished or it isn’t, a rebrand ships or it doesn’t.
Retainers are the right instrument for the first kind because the engagement genuinely has no natural stopping point. Cancelling it isn’t finishing the job, it’s abandoning an ongoing one. Fixed-scope is right for the second kind because a retainer wrapped around completing work either drags on past the point of diminishing returns, or quietly narrows into unrelated busywork to justify the monthly invoice once the real job is done.
Where does fixed-scope clearly win?
A defined migration or rebuild (moving platforms, replatforming a store, a full redesign) has a real deliverable and a real finish line, and deciding whether the migration is warranted at all is its own bounded piece of work; wrapping it in a retainer just adds billing ambiguity to work that should end when it ends. A one-time audit is the same shape: the SEO audit product exists specifically because diagnosis is a bounded task with a clear output, distinct from the ongoing execution that might follow it.
A narrow, single-outcome brief is the other clear case: a landing page for one campaign, a specific integration, a defined content set for a launch. If the brief can be fully described up front and doesn’t imply an obvious next phase, forcing it into a retainer just means billing monthly for a project that was actually finished in month one.
Why would an agency built on recurring revenue ever push someone toward a project instead?
Because getting the shape wrong costs trust on both ends eventually, even if it’s the easier close today. A client sold a retainer for work that actually completes notices, a few months in, that the reports have started padding activity to justify the invoice. That’s a worse outcome for the relationship than a smaller, correctly-scoped project would have been. The friction is real and worth naming rather than pretending it doesn’t exist: fixed-scope work is a smaller, one-time number, and recommending it against our own preference for recurring engagements is the actual test of whether the advice is trustworthy.
What about work that starts as a project and becomes ongoing?
That’s genuinely common, and it’s the right sequencing rather than a loophole. A fixed-scope audit surfaces a prioritised fix list; some of those fixes are one-time (a technical error, a broken redirect chain) and some imply ongoing work (a content programme, continuous monitoring). The honest move is to scope the project as a project, deliver it, and then let the client decide separately whether the follow-on work justifies a retainer, not to presell the retainer before the diagnosis exists to justify it.
What does a correctly fixed-scope proposal actually look like, structurally?
It names a specific deliverable, a specific price, and a specific definition of done: all three, not just the first two. “SEO audit and fix list” with a fixed price is a project. “SEO audit and fix list, then ongoing implementation at the same rate” quietly isn’t, because the second half has no defined end and shouldn’t be priced as if it does. The test for whether a proposal is genuinely fixed-scope: could you hand it to a third party after delivery and have them independently verify the work matches what was promised? If yes, the scope was real. If the honest answer is “it depends what happens after,” the retainer language has crept back in under a project’s pricing.
The other structural marker is that a fixed-scope proposal names what happens if the work uncovers more than expected: a defined change-order process, rather than open-ended scope creep absorbed silently into the same fixed fee. Without that, a fixed-scope project either loses money quietly when it runs long, or gets rushed to protect the fee, neither of which serves the client honestly.
