Retainer or project? An honest decision guide

Studios, including ours, are not neutral witnesses in the retainer-versus-project debate. Retainers are lovely businesses: predictable revenue, deeper client knowledge, calmer Mondays. So when an agency tells you a retainer is what you need, apply the same scepticism you'd apply to a barber recommending a haircut. Sometimes you need the haircut. Sometimes you need a pair of scissors and a mirror.

We've run both models for a decade. Here's the honest map — including when each one fails, which is the part sales decks leave out.

What each model is actually for

A fixed-scope project buys a defined outcome: a website, a rebrand, a product build with a beginning and an end. You're purchasing deliverables against a price. The model works when the outcome can be specified in advance with reasonable accuracy, and when "done" is a real state — a site launches, an identity system ships, an app hits the store.

A retainer buys capability over time: a senior team's ongoing attention to a living thing — growth programme, product iteration, content engine, design system stewardship. You're purchasing capacity and continuity. The model works when the work is genuinely continuous, when priorities change month to month (which is the point, not a defect), and when accumulated context is a competitive advantage the client wants to keep renting.

The single biggest predictor of a bad engagement is buying one when the situation calls for the other. A project-shaped job forced into a retainer becomes an expensive subscription to guilt. A retainer-shaped need forced into a project becomes a launch followed by a slow decay, and we've written before about what that looks like in rebrand rollouts — six months after the big reveal, the system is quietly eroding at the edges because nobody owned the edges.

When a project wins

The outcome is specifiable. You know what "done" looks like and can write it down. New marketing site: specifiable. "Be better at growth": not.
You have internal capability for the after. Your team can run, extend and maintain the thing once shipped. Projects end; if nothing catches what they hand over, the project's value decays from day one.
Budget comes in lumps. Many organisations fund capital work annually and can't hold an ongoing line. A fixed-scope project fits how the money actually moves.
You're testing a new agency. A bounded project is the right first date. You learn everything that matters — how they run a week, how they deliver bad news — with a defined exit.
The risk is technical, not strategic. Fixed scope transfers delivery risk to the agency in exchange for a premium on the price. If what you want is certainty, that's a fair trade. Paying it.

When a retainer wins

The work is continuous by nature. CRO programmes, SEO and content, product iteration, lifecycle email — these compound or decay; there's no "done". Our growth engagements run this way because experiment programmes are explicitly a velocity game: the value is in the cadence, not any single test.
Priorities legitimately change monthly. A retainer converts reprioritisation from a contract amendment into a Tuesday planning call. If your roadmap changes quarter to quarter, that's not failure — that's the environment. Price for it.
Context is the asset. Six months in, a retained team knows your constraints, your codebase, your customers and your politics. That knowledge is where the speed comes from. Rotating project teams pay for re-learning it on every engagement — invisibly, in the estimate.
You've already worked together. Retainers reward trust. Starting a stranger on a retainer is buying the subscription before the first date; the reverse order of operations.

How each fails

Projects fail through scope rot and handover cliffs. Scope rot: the definition of done was optimistic, and the last month is a grinding negotiation about what "included" meant. Handover cliff: the site ships, the agency leaves, and the internal team inherits a machine nobody was trained to drive. Both are mitigable — honest estimation and structured handover, covered in our piece on how we estimate software — but a client should know the failure shapes before signing, not after.

Retainers fail through drift and bad physics. Drift: without a forcing function, retained teams quietly become expensive maintainers of the status quo, and the monthly invoice starts funding presence rather than progress. Bad physics: the client buys ten days a month and plans for fifteen days of work; the retainer becomes a permanent queue of disappointment. The fix for both is the same: retained work needs goals per cycle, reported in outcomes — ours report in revenue or its leading indicators, as Priya writes about — and a quarterly review where either side can propose resizing or ending it without that being a dramatic event.

The hybrid shapes that actually work

The shapes we see succeed most:

Project, then retainer. The most common healthy pattern. Build the thing as a project; hand it to a retainer that runs, grows and improves it. The budgets are different sizes and that's fine.
Retainer with project pockets. Ongoing team for the continuous work, with fixed-scope spikes for definable builds (a new section, a campaign platform). Costs need discipline — the spike gets its own estimate and schedule — but it matches reality better than pretending all work is one kind.
Retainer with a sunset clause. Twelve-month commitment with a three-month notice either way and a quarterly sizing review. Keeps both sides honest; kills zombie retainers.

The questions to ask before signing either

Ask the agency:

"What does this engagement look like in month nine?" (A project answer and a retainer answer should sound entirely different. Beware identical answers.)
"Which of your current clients have ended retainers amicably, and why?" (Everyone's retainers end; the how is the tell.)
"What's the handover plan, in detail, if we never renew anything?"

Ask yourselves:

"Is what we're buying an outcome or a capability?"
"What internal capacity do we have to receive whatever ships?"
"Would we rather transfer delivery risk (fixed price) or scope risk (retainer)?" — somebody holds one of these in every single engagement, and the contract decides who.

The pricing mechanics of both models at Brassfern live on our pricing page, and if you're genuinely unsure which shape fits, the cheapest answer is a bounded discovery sprint: two weeks, fixed price, output includes our recommendation for what should come next — occasionally including "you don't need us for this", which is a finding worth paying for, and one we stand behind enough to put in writing.

Key takeaways

Projects buy outcomes; retainers buy capability over time. The biggest predictor of a bad engagement is buying one when the situation calls for the other.
Choose a project when the outcome is specifiable, you have internal capacity for "after", budget comes in lumps, or you're testing a new agency.
Choose a retainer when the work compounds, priorities change monthly, context is the asset, and you've already worked together.
Projects fail by scope rot and handover cliffs; retainers fail by drift and bad physics. Both failure shapes are contractible problems — mitigate them in the agreement, not the retrospectives.
Hybrid shapes (project-then-retainer, retainer with project pockets, retainer with sunset) map to reality better than purity. Ask "what does month nine look like?" before signing anything.

FAQ

Isn't a retainer just more expensive per hour? Per available hour, usually yes — retainers price in continuity, priority access and the team's retained context. The honest comparison isn't hourly rates; it's cost per outcome over twelve months, including the re-learning tax you pay a rotating series of project teams and the opportunity cost of months where nobody was working on the thing at all. For continuous work, retainers are usually cheaper per outcome. For one-off outcomes, they're a subscription to guilt.

What minimum commitment makes a retainer sensible? Three months to start producing compound value; six to feel it. Shorter than that and you're buying an on-call project team under a fancier name, with none of a project's scope discipline. If a client genuinely can't commit past two months, we recommend a well-scoped project instead. It serves everyone better.

Can we pause a retainer mid-year? Good retainer agreements build this in: typically one or two pause cycles per year with notice, or a step-down to a maintenance tier. What pausing can't preserve is the team — the people working on your account get allocated elsewhere, and "same team back in six weeks" is a hope, not a term. If continuity of specific people matters, get it written down.

How do we keep a retainer from becoming invisible maintenance? Goals per cycle, reported in outcomes, reviewed quarterly with an explicit continue/resize/end question on the agenda. If your agency resists outcome reporting on a retainer, that's diagnostic. The entire point of buying capability over time is that its value should be measurable over time.

We had a bad retainer experience. Why would the next one be different? Diagnose which failure shape you lived: drift (no goals, no demos, quiet decay) or bad physics (chronically more demand than capacity). Both are contractible. Then test the next agency with a bounded project first; agencies that run excellent retainers almost always run excellent projects, and the reverse is reassuring but not guaranteed. Past pain is useful data — bring it to the first meeting and watch what the agency does with it.