Every agency founder I speak to eventually says the same sentence, usually around month eighteen: "We need to move to retainers."
What they almost always mean is "I am tired of starting from zero every January." Which is fair. Project revenue has a particular cruelty to it. You close a good quarter, you feel like a real business, and then three deliveries land in the same fortnight and the pipeline is empty behind them. So retainers start to look like the answer. Predictable revenue. Calm months. A number you can actually plan hiring against.
Before starting Scopeyard I spent years running a product development studio, and then delivered AI automation work across healthcare, recruitment and operations. I've done both models, and I've watched a lot of founders convert too early and quietly lose money on it for a year.
Here is the uncomfortable thing. A retainer does not create predictable revenue. It transfers risk from the client to you. On a project, the client is buying a defined outcome and carrying the risk that they need more later. On a retainer, you are promising availability, and you are carrying the risk that they want more than you priced for. If your delivery is not systematised, that swap is a bad trade.
So the question is not "should we move to retainers." It's "have we earned the right to."
Retainer Readiness = Repeatable Work × Predictable Demand × Delivery Systems ÷ Founder Dependence
1. First, check whether the work actually repeats
The only retainers that work are the ones where the client has an ongoing problem, not a one-time one.
Test it honestly. Look at your last ten projects and ask: six months after delivery, did this client still have a job to be done that we were the natural person to do? Not "would they be happy to hear from us." Would something break, degrade, or go unmanaged if nobody did the work?
Website builds usually fail this test. Performance marketing usually passes it. Brand identity fails. Content and SEO pass. Custom software sits in the middle — the build is a project, but the maintenance, the roadmap and the support are genuinely ongoing.
AI automation is the interesting case, because agencies keep selling it as a project when it is structurally a retainer. Models get deprecated. Prompts drift. The client's data schema changes. An integration silently breaks and nobody notices for three weeks. If you're shipping automations and walking away, you are leaving the most defensible revenue in the business on the table. Most of the AI project delivery checklist — evals, monitoring, handover — describes work that never actually ends. Maintenance is not a favour. It's a product.
If fewer than half your projects have a genuine ongoing job attached, you don't have a retainer business yet. You have a project business with a retention problem, and those are fixed differently.
2. The four signals that say you're ready
I look for four things before I'd tell a founder to convert.
| Signal | What good looks like | Why it matters |
|---|---|---|
| Repeat work | 40%+ of last year's revenue came from clients you'd worked with before | Proves the ongoing job exists |
| Delivery is documented | Someone other than you can run the monthly cycle from an SOP | Retainers are a systems business, not a talent business |
| You know your delivery cost | You can state hours-to-deliver for your core service within ±20% | You cannot price a retainer you can't cost |
| Cash runway | 3+ months of fixed costs covered | Conversion dips revenue before it lifts it |
The third one is where most agencies fall over. Founders who can't answer "how many hours does our standard monthly SEO deliverable actually take" price the retainer off what the client will accept, not off what it costs. That works for about four months.
If you don't have that number, build the SOP library before you build the retainer. It is boring and it is the whole game.
3. The three signals that say you're not
You're converting to fix cash flow. This is the most common reason and the worst one. Retainers do not fix cash flow on their own — the billing terms do. Ignition's 2025 survey of 273 agency managers and executives found 71% say at least one in four invoices is paid late, and 56% typically wait two weeks to two months past the due date to get paid. A monthly retainer billed in arrears with 30-day terms is just a smaller, more frequent version of the same problem. Retainers billed in advance by direct debit fix it. The mechanism is the payment term, not the pricing model.
Your utilisation is already above 90%. A retainer is a promise of availability. If your team has no slack, you're selling something you don't have, and the first month a client asks for more you'll either eat it or disappoint them.
Every project still runs through you. If you're the quality gate on every deliverable, a retainer just guarantees you a recurring monthly obligation you personally have to meet. That's not recurring revenue. That's a subscription to your own calendar.
4. Sell a defined scope, not "availability"
The single biggest mistake in retainer design is selling hours or access instead of a fixed set of deliverables.
"20 hours a month" invites the client to audit your timesheet and argue about efficiency. "Four blog posts, two landing pages, one monthly performance review" invites them to judge the output. The second conversation is the one you want.
Write the retainer as:
- A named list of deliverables per cycle, with quantities
- An explicit response time for ad hoc requests, and a cap on them
- A change process for anything outside the list, with a rate attached
- A minimum term, usually three or six months, and a notice period
The cap matters more than founders think. The Ignition study found 57% of agencies lose $1,000 to $5,000 every month to unbilled work, and another 30% lose more than $5,000 a month. And 78% say they rarely or only sometimes charge for out-of-scope work. On a project, scope creep costs you once. On a retainer it compounds monthly, silently, forever, because there's never a delivery moment that forces the conversation.
If you're going to run retainers, you need an out-of-scope process you'll actually use. Write the rate into the agreement so raising it is administrative rather than confrontational.
5. Price the first retainer off cost, then check it against the market
Work bottom-up, not top-down.
Take the deliverable list. Cost it in hours at your blended internal rate. Add 20% for coordination, reporting and the ad hoc requests you know will come. Multiply by your target margin. That's your floor.
Then sanity-check against the market. Published benchmarks put small-business retainers roughly in the $1,000–$5,000 a month band, mid-market at $5,000–$15,000, and enterprise above $15,000. B2B and technical work sits higher. These are wide ranges and they vary hugely by service and geography, so treat them as a reality check on your floor rather than a price list.
A simple worked example. Say your monthly deliverable set costs 24 hours of delivery time. At a blended internal cost of $45/hour that's $1,080. Add 20% for coordination — $1,296. At a 55% gross margin target, the retainer prices at roughly $2,880. If you were about to quote $1,800 because it "felt right for a small client," you now know precisely what that decision costs you: $1,080 a month, $12,960 a year, per client.
Don't be lazy with this. The same discipline applies whether you're pricing a software project or a monthly retainer.
6. Convert existing clients before you sell to new ones
New logos are the hard way to launch a retainer. Existing clients who just finished a successful project are the easy way, and the conversation is simple: the thing we built needs looking after, here's what that costs.
Time it at handover, not six weeks later when the relationship has gone cold. Build the retainer offer into the project close — the same meeting where you walk through the handover checklist. Make it the natural next step rather than a new sale.
Price the first cohort slightly under your target and be explicit that it's an introductory rate for a fixed term. You want three or four clients through a full quarter so you can measure real delivery cost before you commit to a public price.
7. Aim for a mix, not a conversion
Nobody sensible goes 100% retainer.
Project work is where you meet new clients, test new services and charge properly for one-off complexity. Retainers are where you build the base that covers fixed costs. The healthy state is both, with recurring revenue covering enough of your monthly overhead that a slow quarter is annoying rather than existential.
As a rough progression: early on, getting 25–35% of revenue recurring is enough to stop the January panic. Once retainers cover your fixed costs — payroll, rent, tools — you've crossed the line that actually changes how you run the business. Beyond that, more recurring revenue buys you optionality rather than survival.
It's worth knowing that this shows up in valuation too. Brokers who sell agencies consistently place project-only shops at the bottom of the EBITDA multiple range and retainer-heavy books several turns higher — the commonly quoted spread is roughly 2–4x for project-only generalists versus 6–9x for agencies with 60%+ recurring. Broker ranges are self-interested and vary by deal, so don't plan around a specific number. But the direction is not in dispute: buyers pay for predictability.
And predictability is worth something even if you never sell. Ignition found 63% of agencies suffer unpredictable cash flow, and 82% of them delayed or cancelled hiring and investment as a result. That is the real cost of a project-only book. Not the revenue you didn't earn. The hires you didn't make.
8. The operational tax you're signing up for
Retainers are quieter revenue but louder operations.
A project has a natural ending that forces reconciliation. A retainer has twelve invoices a year, each of which needs the client to feel they got value. That means a reporting rhythm, a visible record of what was delivered, and approvals that don't drift. AgencyAnalytics' 2025 benchmarks, drawn from 220+ agency leaders, found 70% say client reporting plays a critical role in retention — and 43% report average client lifespans between two and five years. Long relationships are normal in this model. They're also fragile in a specific way: nobody churns because of one bad month, they churn because six months went by and they can't remember what you did.
This is the part we built Scopeyard for. If a client can log in and see every deliverable in the cycle, what's shipped, what's waiting on them, and what they approved last month, the value conversation stops being a slide you make and starts being a fact they can check. Agencies running monthly retainers for marketing clients tend to feel this first, because the deliverable count is high and the individual items are small.
Final thoughts
Move to retainers when the work genuinely repeats, when someone other than you can deliver the cycle, and when you know what an hour of your delivery actually costs. Move for any other reason and you've swapped a lumpy revenue problem for a slow margin problem, which is worse because it takes a year to notice.
Retainers don't make an agency predictable. Systems do. Retainers just make the lack of them expensive every single month.