Hourly vs Project vs Retainer: Which Pricing Fits Your Freelance Work?

Short answer: Start from the rate floor, then pick the model that fits how the work actually arrives. Hourly fits work whose size nobody can predict. Project pricing fits work with a clear deliverable and a known scope. Retainers fit ongoing work for a client who needs you every month. The model does not set your price — the floor from Guide D1 does that. The model decides when the money arrives, and for irregular income, when it arrives matters almost as much as how much it is.

Who this is for: You know, or are working out, your rate floor — the hourly number below which the work does not pay for itself. Now you need to decide how to charge that floor: by the hour, by the project, or by the month.

Who this is not for: If you have not worked out the floor yet, stop here and read Guide D1 first. Choosing a model before knowing the floor is how freelancers end up with a beautiful retainer that pays less per hour than the job they left, or a project price that felt generous until the third round of revisions ate it. Floor first, model second. Every number below assumes the floor is already known.

The floor, briefly, because everything hangs on it

Guide D1 works the floor out in full: the gross income needed to cover tax, expenses, and the life the work is supposed to fund, divided by realistic billable hours. Its worked example lands on $112.60 an hour at 1,200 billable hours a year (against a naive $38.46 from dividing a salary by 2,080 hours). That $112.60 is the number used in the examples below, so the three models can be compared on equal footing. If your floor is different — and it should be, because your costs and hours are yours — swap your number in wherever $112.60 appears. The comparisons still hold; only the dollars change.

One more ground rule, stated once so it does not need repeating: no model in this guide is presented as a way to charge multiples of the floor on the strength of a pricing theory. A model changes timing and risk. It does not turn a $112.60 floor into a different income by itself. Anyone promising that a pricing model alone multiplies earnings is selling a course, not describing arithmetic.

Hourly: the honest meter

How it works: you charge for each hour worked, at the floor rate or above, and invoice the hours used.

Decision rule — choose hourly when: the size of the work cannot be known in advance. Open-ended support, debugging, consulting where the client brings a new problem each week, work where the scope depends on what you find once you start. If you cannot honestly estimate the hours, do not price the project — meter it.

The maths: 20 hours in a month at the $112.60 floor = $2,252. Ten hours = $1,126. The arithmetic is trivial, and that is part of the appeal: there is nothing to argue about except the hours themselves, so keep time records that would survive a polite question from the client.

Irregular-income consequence: hourly income tracks hours worked, week by week, with no smoothing built in. A 30-hour month followed by a 5-hour month is a $3,378 month followed by a $563 month at the floor rate, and the buffer (Guide A2) absorbs the gap. Hourly also punishes speed: the faster and better you get, the fewer hours a job takes, and the less you earn for the same result. That is not a reason to work slowly. It is a reason to notice when hourly stops fitting work you have mastered — which is usually the moment project pricing starts looking sensible.

Watch out for: scope creep without a rate conversation. Hourly clients sometimes treat the meter as unlimited. A short written note at the start — the rate, an estimated range, and "I will flag it when we pass X hours" — prevents most arguments, and the flag is worth more than the note.

Project: one price for one deliverable

How it works: you agree a fixed price for a defined deliverable, and the client pays that price whether it takes you 15 hours or 25.

Decision rule — choose project pricing when: the deliverable is clear, the scope can be written down, and you can estimate the hours within reason. A website with named pages, a report with named sections, a design package with a set number of concepts and revision rounds. If you can describe what "done" looks like in a sentence, project pricing is available. If you cannot, it is not — yet.

The maths: start from the floor, not from a feeling. A project you estimate at 25 hours prices at 25 × $112.60 = $2,815 before any allowance for revisions, project management, or the risk that the estimate is wrong. That "before" matters. A project price built purely on estimated hours, with no room for the two extra revision rounds that usually arrive, is an hourly rate in disguise, and a worse one, because the extra hours are unpaid. Build the allowance in openly: estimated hours at the floor, plus a stated revision allowance, plus any fixed costs the project itself creates. If the client wants the price lower, the honest trade is a smaller scope, not a smaller hourly equivalent hidden inside the fixed price.

Irregular-income consequence: project work concentrates income. A $2,815 project paid on delivery, after six weeks of work, is six weeks of spending from the buffer followed by one big inflow. That is exactly the feast-and-famine shape Guides A1 and A2 are built to smooth, and project-heavy freelancers lean on the buffer hardest. The fix is partly structural — see deposits in Guide D4, where a 25–50% upfront share, commonly 50% for new clients, splits a project payment into two arrivals instead of one — and partly habit: when the big payment lands, the surplus rule sends it to buffer first, not lifestyle first.

Watch out for: the second project that is really the first project continuing. "Just one more page," "just a quick extra section" — each one is small, and together they are an unpaid project. The written scope is the defence, and the sentence "happy to do that as a separate piece at the same rate" is worth practising until it sounds friendly, because it should sound friendly. It is a normal business sentence, not a confrontation.

Retainer: the same money, every month

How it works: a client pays a fixed monthly amount for ongoing work, usually defined as a set number of hours or a set scope of services each month.

Decision rule — choose a retainer when: the client needs you every month, the work recurs, and both sides would rather not renegotiate monthly. Ongoing support, regular content, monthly reporting, a standing advisory role. If the work is genuinely monthly, a retainer fits. If the client only thinks they need you monthly, the retainer will quietly become a source of friction by month three.

The maths: the retainer is the floor, times the hours, with the same honesty as project pricing. A 10-hour monthly retainer at the $112.60 floor = $1,126 a month. A 20-hour retainer = $2,252 a month. Price it from the floor, not from what the client suggests as a "monthly budget" — a budget below the floor times the hours is a pay cut with a nicer name.

Irregular-income consequence: retainers smooth income, and that is their biggest selling point for a freelancer. A $1,126 retainer that arrives every month is $1,126 the buffer does not have to cover, every month, including January and July (see Guide A5's slow quarters). Two retainers at that level cover $2,252 of monthly essentials-adjacent income before a single project is sold. That smoothing is real, and it is why retainer-heavy freelancers often report the calmest cash flow on this site's whole system. It is not free, though: the smoothing is bought with availability. Retainer clients reasonably expect you to be there in their month, including the month a big project would have paid more.

Retainer pricing: guaranteed hours, unused hours, raising it

This is where retainers actually succeed or fail, so it gets its own section rather than a paragraph.

Guaranteed hours. State the hours plainly: "up to 10 hours a month." The word "up to" does work here — it caps your commitment while giving the client a clear allowance. Also state what counts: which kinds of work, what turnaround, and what happens when a request would push past the allowance (it becomes hourly at the floor rate, or a separate project — pick one and write it down).

Unused hours. This is the question every retainer client asks, and every retainer freelancer dreads, so decide it before the client does. The common options are:

Whichever option you pick, write it in the agreement and state it in the first month, not when the question first arises. A rule announced at the start is a term. A rule invented when hours have already gone unused is an argument.

Raising it. Retainers go stale. The floor rises (costs, tax, experience), the scope creeps, and a retainer priced two years ago quietly falls below the current floor. Raise it on a schedule, not on a grievance: review every six or twelve months, compare the retainer against the current floor times the hours, and raise it in a dated conversation tied to the review, not mid-crisis. A small, regular raise — say, in line with the floor's movement — is easier for a client to accept than a large correction after years of silence, and it is easier for you to ask for, because it is a review, not a confrontation. If the client will not accept a raise that merely restores the floor, that retainer is now below-cost work, and the honest decision is whether the relationship is worth subsidising — a decision to make deliberately, not to drift into.

Choosing between them: a decision table

Most freelancers end up with a mix, and that is fine — even good. A retainer or two for the base, project work for the peaks, hourly for the genuinely unpredictable. The mix itself smooths income: retainers cover the floor, projects provide the surplus that fills the buffer, hourly fills the gaps. What does not work is drifting into a model by accident — charging hourly because nobody discussed it, or project-pricing work whose scope was never written down. The model should be a choice, made with the floor in front of you.

Checklist

• Rate floor from Guide D1 calculated and written down (your number, not Sam's $112.60).
• Each current client matched to a model, with a reason from the decision table.
• Project prices built from floor × estimated hours plus a revision allowance, not from a feeling.
• Retainer agreements state guaranteed hours, the unused-hours rule, and a review date for raising it.
• Deposit share set for project work (Guide D4), so project income does not arrive in one lump.
• Income smoothing checked: retainer base vs monthly essentials, buffer still in place for the rest.

Frequently asked questions

Can I switch a client from hourly to a retainer? Yes, and the cleanest moment is a review, not a complaint. Show the last few months' hours: if they cluster around, say, 10 hours a month, a 10-hour retainer at the floor gives the client predictable cost and gives you predictable income. If the hours swing wildly, hourly is still the honest model, and a retainer would misfit both sides.

Is project pricing always better than hourly once I am experienced? No. Experience makes project pricing available, not mandatory. Work whose size you still cannot predict — a new kind of problem, a client whose scope moves — is still hourly work, however experienced you are. The skill is matching the model to the work, not graduating from one model to another.

What if a retainer client uses far fewer hours than the allowance? That is the unused-hours rule doing its job — if you picked one and wrote it down. If the pattern persists for months, the retainer is oversized for that client, and the review conversation should resize it rather than letting resentment build. A smaller retainer that both sides feel good about beats a larger one that one side quietly regrets.

Does the model change my tax? The model changes when income arrives, which can change which quarter it lands in, but it does not change the tax set-aside rule: tax moves first, on every payment, whatever the model (Guides B1 and A4). If timing questions get complicated, that is a question for a tax professional, not a pricing decision.

What to do next

1. Check your rate floor first, then match the model to how the work arrives.
2. Run your numbers through the rate calculator.
3. Re-check the dated figures and sources in this guide before you act on them.

Sources and method note

• Method: pricing-model comparison for freelancers, built on the rate-floor method; Sam's $112.60 floor and the retainer/project arithmetic derived from it are fictional worked examples from Guide D1, not survey data or market rates. No market-rate table is given or implied.
• Rate floor: Guide D1 on this site. Deposit maths and late-payment ladder: Guide D4 on this site. Income smoothing and buffer: Guide A2 on this site. Seasonal view: Guide A5 on this site.
• Tax ordering: Guide B1 on this site; IRS Publication 505 (estimated tax).

Dated assumptions (25 Oct 2026): tax year 2026. The $112.60 floor is Guide D1's worked example (gross $135,123 at 1,200 billable hours); replace it with your own floor before pricing anything. No figure here is a market rate, a premium, or a promise about what clients will pay.

Related guides: Rates and safety net hub · D1: How much should I charge · D4: A client hasn't paid · K4: Freelance rate calculator

Educational only. Worked examples are fictional and illustrate a method, not a promise about your income. Tax figures depend on your situation — consult a tax professional. Reviewed 25 Oct 2026.

Related guides and tools

• the rates and safety net hub
• how much should I charge — work out your real rate floor
• what to do when a client hasn't paid your invoice

Dated assumptions: draft content as checked 25 October 2026; tax and retirement figures are for tax year 2026 where stated. Verify current-year limits, rates, and deadlines before acting, as the IRS adjusts them annually.