There are four practical ways to price agency work: hourly, fixed-fee per project, monthly retainer, and value-based. None is universally better - each one moves risk between you and the client. Hourly puts all the risk on the client and punishes you for getting faster. Fixed-fee moves the risk to you and rewards efficiency, but only if your scope has real exclusions. Retainers give both sides predictability and quietly bleed margin when nobody tracks burn. Value-based pays the most and requires you to know your delivery costs precisely, which is why it is the last model to adopt rather than the first. Most agencies should run fixed-fee for projects and retainers for ongoing work, and use hourly only where scope is genuinely unknowable.
Most agencies do not have a pricing strategy. They have a pricing habit - whatever they charged the first client, adjusted upward when it felt uncomfortable.
That works until it does not, and the moment it stops working is rarely obvious. Revenue looks fine. The team is busy. And yet margin keeps thinning, the good clients feel expensive to serve, and nobody can say which projects actually made money.
This guide covers the four pricing models agencies actually use, what each one does to risk and behaviour, how to choose between them, and the specific mistakes that make a well-chosen model fail anyway.
Pricing is a risk transfer, not a number
Before the models, one idea that makes all of them easier to reason about.
Every pricing model is an answer to a single question: if the work takes longer than expected, who pays for it?
- Hourly: the client pays. All overrun risk sits with them.
- Fixed-fee: you pay. All overrun risk sits with you.
- Retainer: whoever is not tracking pays. Usually you.
- Value-based: you pay for overruns, but you also keep the upside if you are fast.
That is genuinely the whole framework. Once you see pricing as risk allocation rather than a rate card, the arguments become much simpler - because the question "should we charge hourly or fixed?" turns into "who is better placed to carry the uncertainty on this particular job?"
And the answer to that is usually: whoever has the most information. If you have built this exact thing forty times, you know the cost better than the client does, so you should carry the risk and charge fixed. If neither of you has any idea what is involved, making the client carry it via hourly is honest.
The four models
1. Hourly
You charge a rate per hour, track time, and invoice the total.
Scoro's breakdown of agency pricing models and Teamwork's guide both cover the mechanics; the version worth internalising is what it does to incentives.
Hourly punishes you for getting better at your job. A team that halves its delivery time through experience, tooling, or a good template halves its revenue on the same work. Every efficiency gain is a pay cut. Over years, that is a genuinely perverse arrangement, and it is the strongest argument against hourly as a default.
It also makes the client an auditor. When the unit of sale is your time, the client's only lever for controlling cost is scrutinising how you spend it - which produces the timesheet interrogations that make agency-client relationships adversarial.
Where hourly is genuinely right:
- Discovery and advisory, where the scope legitimately cannot be known in advance.
- Ad-hoc support requests that do not fit a project shape.
- The first engagement with a client whose working style you cannot yet predict.
- Anything where you would otherwise have to price in a large uncertainty buffer that the client would reasonably refuse to pay.
The one thing to get right if you use it: your rate has to be built from your real capacity, not from a number that sounds acceptable. See the section on calculating a rate below.
2. Fixed-fee per project
You quote a total for a defined scope, and that total does not move unless the scope does.
This is the right default for most agency project work, for one reason: it aligns your interests with the client's. They want the outcome; you want to deliver it efficiently. Nobody is watching the clock adversarially.
It also lets you sell on value rather than effort. A client comparing two agencies at £18,000 and £22,000 is comparing propositions. A client comparing £150/hr and £180/hr is comparing commodities.
Fixed-fee only works with real exclusions. This is the failure mode, and it is universal. An agency quotes a fixed price against a scope that lists what is included and says nothing about what is not. Every ambiguity then resolves in the client's favour, because they are reading the same document and reasonably assuming the unlisted thing is covered.
A scope with an explicit exclusions section is the difference between a fixed-fee project and an unlimited-scope disaster at a fixed price. Our guides to writing a statement of work and preventing scope creep cover the structure; if you take one thing from them, take the exclusions list.
Where fixed-fee is right: anything you have done before and can estimate within about 20%. Website builds, brand identities, defined campaigns, migrations - work with a recognisable shape.
Where it is wrong: genuinely novel work, or a client whose decision-making you have not yet observed. A fixed price against an unpredictable approver is a bet on someone else's behaviour.
3. Retainer
The client pays a recurring monthly fee, either for an agreed volume of work or for ongoing responsibility for an outcome.
Retainers are the most commercially attractive model on this list - recurring revenue, predictable capacity planning, lower sales cost - and the easiest to run badly.
There are two distinct kinds, and conflating them causes most retainer problems:
Capacity retainers buy an agreed number of hours or a defined deliverable set per month. "40 hours of design support" or "four blog posts and two email campaigns."
Outcome retainers buy ongoing responsibility for a result, regardless of hours. "You own our SEO." The hours are your problem.
The failure mode is specific and almost universal: the unmanaged capacity retainer. Hours are not tracked against the agreed volume in real time, overage accumulates silently, and by the time anyone reconciles it the month is closed. You absorb it. That becomes precedent, and the client's mental model of the retainer shifts from "40 hours" to "access."
The fix is not a stricter contract. It is visibility - burn tracked against the agreed volume, visible to both sides, mid-month rather than after. A conversation on the 18th about being at 80% of the allocation is easy. The same conversation on the 3rd of the following month is a dispute.
Retainers also need an explicit answer to two questions most contracts skip:
- Do unused hours roll over? If yes, you will eventually owe a client 90 hours in one month and it will break your capacity plan. Most agencies cap rollover at one month or disallow it, and say so plainly at the start.
- What happens at overage? Billed at a stated rate, or stopped pending approval? Either is fine. Silence is not.
4. Value-based
You price against the outcome's worth to the client rather than the effort involved. A pricing project that adds £400,000 of annual margin is worth more than three weeks of a consultant's time, and value-based pricing captures some of that difference.
The upside is real and large. So is the prerequisite most agencies skip.
You cannot price on value until you know your costs. A firm that moves to value-based pricing before it understands its own delivery economics ends up either undercharging - because it has no floor - or losing deals to a price the client cannot benchmark and the agency cannot justify. The instinct that value-based pricing is a way to escape the discipline of knowing your numbers has it exactly backwards: it requires more of that discipline, not less.
It also requires access. To price on outcome you need to know the client's economics - their margin, their conversion rate, the value of the problem. A client who will not share that cannot be sold value-based work, and most clients will not share it with an agency they have not worked with.
Which is why value-based pricing is realistically the fourth model an agency adopts, on the second or third engagement with a client, in a domain where the outcome is measurable. Treating it as a starting point is how agencies end up with a beautifully argued price and no signature.
How to actually set your rate
Even fixed-fee and value-based pricing need an internal hourly cost, because that is your floor. Here is the calculation, which most agencies do wrong in the same specific way.
Step 1: Work out real available hours. Not 40 a week. Subtract internal meetings, admin, recruitment, training, holiday, and sick leave. For most agency roles the honest figure is 25 to 32 hours, and it drops with seniority. Derive it from tracked data rather than assuming - our guide to capacity planning covers the method.
Step 2: Total your annual costs. Salaries plus employment costs, software, rent, insurance, tooling, and an allowance for non-billable time you have already excluded above. Everything.
Step 3: Divide costs by billable hours. Total annual cost ÷ (available hours × billable utilization × number of people) = your break-even hourly cost.
That number is usually a shock. It is also the single most useful figure in the business, because every price you quote can now be checked against it.
Step 4: Add your target margin. Whatever the market supports, on top of a floor you actually know.
The mistake almost everyone makes is at step 1 - using 40 hours, or 2,080 a year. That inflates the denominator, produces a break-even cost far below reality, and yields rates that feel healthy and lose money. Asana's guide to utilization rate and Scoro's breakdown of billable utilization both explain why the denominator is the part that matters, and our post on agency utilization rate goes into the two competing formulas.
Choosing a model: three questions
1. How well do you know this work? Done it many times → fixed-fee. Never done it → hourly, or a paid discovery phase priced fixed and the build priced after.
2. Is it one-off or ongoing? One-off → fixed-fee. Ongoing → retainer.
3. Can you measure the outcome, and will the client share the numbers? Both yes, and you have delivered for them before → consider value-based. Otherwise, not yet.
A perfectly reasonable mature pricing model for a mid-sized agency is: paid discovery (fixed), then build (fixed), then retainer (capacity, with tracked burn) - and hourly used only for out-of-scope ad hoc requests. That covers the lifecycle without any exotic pricing theory.
The mistakes that break a good model
Discounting to win, then resenting the client. A discount is not a one-off concession; it resets the client's reference price permanently and every future quote is measured against it. If you must move, remove scope rather than reduce price - it protects the rate and teaches the right lesson about what things cost.
Pricing from the client's budget rather than your cost. Anchoring on what they say they can afford is fine as a qualification filter. It is not a pricing method, because it has no relationship to whether the work is profitable for you.
One rate for everyone. A blended rate is convenient and safe only when the actual seniority mix matches the mix the rate assumed. A project that ends up staffed more senior than planned loses money invisibly, because nothing in the invoice reveals the shift. If you use a blended rate, check the realised mix afterwards on at least a sample of projects.
Never revisiting. Costs rise every year. A rate set three years ago and never touched is a real-terms price cut compounding annually.
Not measuring realisation. You can price perfectly and still lose, if tracked billable hours never reach the invoice. Realisation - invoiced hours ÷ billable hours tracked - is the number that catches discounting, absorbed scope, and unbilled overage. High utilization with low realisation means everyone is busy and the revenue does not reflect it.
How to present a price so it holds
A well-calculated price still loses if it is presented badly. Four things change the outcome more than the number does.
Present the price inside the value, never on its own. A number arriving in an email with no surrounding argument invites comparison shopping, because comparison is the only tool the reader has. The same number at the end of a document that restates their problem, your approach, and what changes is a conclusion rather than a quote.
Give three options rather than one. A single price is a yes/no decision, and "no" is always the safer answer for a buyer. Three - a reduced scope, your recommendation, and a larger version - turns it into "which", which is a fundamentally easier question. Structure them so the middle option is the one you want, because it usually wins. Our guide to writing a proposal covers how to lay this out.
Never apologise for the number. The sentence "I know it's a lot, but…" tells the client the price is negotiable and that you are uncomfortable with it. Say the number plainly and stop talking. The silence after a price feels much longer to the person who said it than to the person hearing it.
Put a decision date on it. Not artificial urgency - a real one, tied to your capacity. "This holds until the 20th, after which the team scheduled for it moves onto other work." That is true, it is fair, and it prevents the quiet death by deferral that kills more proposals than rejection does.
What client objections actually mean
Most price objections are not about price, and answering them literally is how agencies discount unnecessarily.
"That's more than we budgeted." Usually true and usually irrelevant, because the budget was set before they understood the scope. The right response is not a discount, it is a scope conversation: "That's useful to know. Here's what we could deliver inside that budget, and here's what we'd leave out." Removing scope protects your rate. Cutting the price teaches them your rate was inflated.
"Another agency quoted half that." Almost always comparing different things. Ask what is included - revision rounds, strategy, testing, project management, support after launch. In most cases the cheaper quote excludes several of those, and the gap explains itself. If it genuinely does not, you may be facing someone who is either underpricing or better than you, and both are worth knowing.
"Can you do better on price?" A reflex, not a position. Many buyers ask this on every purchase regardless. "The price reflects the scope - if the budget is fixed, I'd rather adjust what's included than reduce the quality of what we deliver" answers it without conceding anything, and it is true.
"We need to think about it." Usually means an unaddressed concern they did not want to raise. Ask directly: "Of course. Is there a part of this you're unsure about? I'd rather answer it now than have you decide around it."
Genuine budget constraint. Sometimes it is exactly what it says. Then the honest answers are a smaller scope, a phased engagement, or declining the work. Discounting to fit is the one option that hurts you twice - once on this project and again on every future quote to that client.
Pricing by agency type
The framework is universal; the sensible default differs.
Design studios. Fixed-fee, with revision rounds stated as a number and the price of an additional round stated up front. Design is judged subjectively, so unbounded iteration is the specific risk and the scope has to bound it explicitly.
Development shops. Fixed-fee for defined builds, hourly or a capacity retainer for maintenance and support. The complication is that "one small change" is frequently a schema change plus a migration plus a regression, so estimation buffers need to reflect systems rather than screens.
Marketing agencies. Retainer-dominant, because the work is genuinely ongoing. The discipline that matters most is tracked burn against the allocation - see retainer models.
Consultancies. Day rates or fixed-fee engagements. The metric that decides profitability is realisation - how much tracked billable time actually reaches an invoice - because advisory work is written off at invoicing more often than any other kind.
A note on raising prices
Every model above assumes your rate is roughly right today. For most agencies it is not, because rates get set once and then quietly erode.
Costs rise every year - salaries, software, insurance, rent. A rate held flat for three years is a compounding real-terms price cut, and the agencies most likely to be underpriced are the ones whose clients are happiest, because nobody complains and so nothing prompts a review.
Two habits fix this permanently. Review rates annually on a fixed date, whether or not it feels necessary - a scheduled review removes the need to work up the nerve. And raise new-client rates first, letting existing clients follow at renewal. New prospects have no reference price, so they simply hear the new number; existing clients get notice and a reason at a natural boundary.
The fear is always that clients will leave. In practice a modest, well-communicated increase loses very few, and the ones it does lose are usually the lowest-margin accounts - which is a filter rather than a loss.
What to do this week
If your pricing is a habit rather than a decision, three steps in order:
- Calculate your real break-even hourly cost using genuine available hours. Do it once, properly. It reframes every conversation that follows.
- Pick the last five completed projects and calculate actual margin on each - real hours at real cost against what you invoiced. You will find at least one you thought was fine that was not.
- Choose a default model per work type and write it down. Not per client, per type of work. Consistency is what lets you compare projects at all.
Pricing rarely improves through a single dramatic change. It improves because you can finally see which work makes money, and that visibility comes from three connected things: a real cost floor, a scope with exclusions, and time data honest enough to trust.