Retainer vs per-project translation pricing: which model fits your clients

Sooner or later every agency owner we talk to asks a version of the same question: keep quoting each job as it lands, or move the best accounts onto a monthly arrangement? Translation retainer pricing sounds like the grown-up answer. Steadier cash flow, fewer hours lost to quoting, a client who stops price-shopping every January. What we've watched happen is messier. Retainers work well on maybe a third of accounts and quietly bleed margin on the rest, and the thing that decides which is which has very little to do with how big the client is.

What translation retainer pricing actually covers

"Retainer" gets used for at least three different arrangements, and agencies get into trouble when they sign one and manage another.

The most common is a prepaid word bank. The client commits to a monthly volume, say 25,000 words, at a rate slightly below your standard, and unused words roll forward for a month or two before expiring. Risk sits mostly with the client: if they send nothing, they've still paid.

The second is reserved capacity. The client is not buying words, they're buying a claim on your schedule. A guaranteed 48-hour turnaround on anything under 5,000 words, or a named senior translator held available two days a week. You bill whether or not they use it, because the cost to you is real either way. This is the version agencies most often underprice, because holding capacity feels like it costs nothing until the month a good client asks for something and you have to say no.

The third is a fixed-scope subscription: a defined set of recurring deliverables at a flat monthly fee. Weekly release notes plus a monthly newsletter, four languages, done. Here the risk is yours, and it only works when the scope genuinely repeats.

The distinction matters more than the label. A client who thinks they bought reserved capacity while you think you sold a word bank will be unhappy the first time you quote a two-week turnaround in a busy month. Write down which one it is, in the contract, in plain language.

Where per-project pricing still wins

Per-project pricing gets treated as the beginner's model, the thing you graduate from. We don't think that's right. There are accounts where it's simply the better commercial answer.

Take a manufacturer we've seen this pattern with repeatedly: they refresh a technical manual twice a year, roughly 80,000 words each time, into four languages. Nothing else. Between refreshes, months of silence. A retainer here is either priced so low it insults the work, or priced at the annual average and the client stares at eight invoices for nothing. Quote each refresh, price the terminology maintenance between cycles as a small separate line if they want it, and both sides sleep fine.

The same applies when the client's own budget is project-funded. Plenty of corporate buyers can approve a purchase order against a named initiative and cannot approve a recurring line item without a procurement cycle that takes a quarter. Pushing a retainer at that buyer isn't commercial ambition, it's asking them to do something their finance system won't let them do.

And there's the case that catches newer agencies: brand-new clients. You have no volume history, no sense of how much rework their reviewers generate, no idea whether their "final" files are final. Price a retainer on three data points and you'll be wrong. Run them per-project for two quarters, collect the numbers, then have the conversation. The retainer offer lands better anyway when you can say "you sent us 190,000 words last year across 31 jobs, here's what that would have cost under a monthly arrangement."

The predictability you gain and the margin you trade

The argument for retainers is revenue predictability, and it's a real argument. Knowing that $18,000 arrives on the 5th regardless of what lands in the inbox changes how you hire, how you talk to your bank, and how calmly you handle a slow August. Coverage of LSP results in outlets like Slator tends to reward exactly this: buyers of translation businesses pay more for recurring revenue than for the same revenue arriving as one-off jobs.

What nobody puts in the pitch deck is the price of that predictability.

Clients expect a discount for committing, usually somewhere in the 10 to 15 percent range, and they'll anchor higher if you let them. That comes straight off gross margin on work you were probably getting anyway. If the retainer only formalises volume the client was already sending, you've handed back a slice of margin for a signature.

Capacity commitments cost more than they look. A guaranteed turnaround means keeping slack in the schedule, and slack is unbilled time. Agencies that promise 24-hour turnaround on a retainer and then run at 95 percent utilisation are not managing capacity, they're gambling that two clients won't have a busy week simultaneously.

There's also concentration risk that per-project work spreads out naturally. Four retainers at $15,000 a month is a lovely number until one renewal goes badly and a quarter of your predictable revenue leaves in a single email. Per-project clients churn too, but they churn gradually and you see it coming.

None of this argues against retainers. It argues for pricing them as what they are: a trade of margin and flexibility for predictability, which is only a good trade when the predictability is worth something specific to you.

How to price a retainer without guessing

The mistake we see most often is pricing the commitment at the client's peak month. It feels safe. It isn't, because the client will look at their actual usage after two quarters and ask for a reduction, and now you're negotiating from a weaker position than if you'd started lower.

Pull twelve months of history for the account. Not the total, the month-by-month distribution. You want the median monthly volume and the spread. Set the committed volume at or slightly below the median, not the mean and definitely not the peak. The median is what they reliably need. Everything above it should be overage.

Price overage at your standard rate, or very close to it. Discounting overage teaches the client that the commitment doesn't matter, and the whole point of the commitment is that it does.

Cap rollover at one or two months. Uncapped rollover is how agencies end up owing a client 90,000 words of work in a quarter when the client suddenly reactivates, usually at the worst possible moment.

Then write down what's included beyond the words. This is where retainers quietly go underwater. Glossary maintenance, reviewer feedback cycles, file preparation for awkward source formats, the QA step and the report that comes with it. Each of those is real hours. If the client's expectation is that a retainer covers "everything translation-related," you'll be absorbing work you'd have billed separately on a per-project basis. Our guide to translation quality assurance covers what a defensible QA step actually involves; whatever you decide, put the scope of it in the retainer document rather than leaving it to custom.

One more thing worth pricing explicitly: rush requests. Under per-project billing, a rush surcharge is normal and nobody blinks. Under a retainer, clients often assume urgency is included, because they're already paying you every month. Say otherwise in writing before it comes up.

Repetitions deserve a decision too. If your per-project quotes apply a fuzzy-match discount grid and your retainer prices flat per word, you've just changed the economics of every high-repetition file the client sends. That can go either way. On a client whose content repeats heavily, flat pricing is a gift to you and they'll eventually notice. On a client with genuinely fresh copy every month, flat pricing is a gift to them. Either pick the flat rate with eyes open, having modelled it against last year's actual match distribution, or carry the grid through into the retainer and accept that your monthly invoice is now slightly harder to explain. What doesn't work is quoting flat because it's simpler and discovering the answer in your annual accounts.

Which clients are actually ready for a retainer

There's a pattern to accounts where retainers hold up, and it's about the rhythm of the client's content rather than their size.

The strongest signal is content produced on a cycle the client doesn't control. A SaaS company shipping every two weeks generates release notes and support articles whether anyone feels like it or not. A regulated manufacturer producing updated safety documentation on a compliance schedule has the same property. The volume arrives because the calendar says so, and that's exactly what makes a monthly commitment sane for both sides.

A second version of the same pattern shows up in HR and internal communications. One client we've watched runs a workforce across five countries and pushes out policy updates, onboarding material, and safety briefings on a quarterly rhythm set by their compliance calendar. The volume per quarter varies, but the fact of it doesn't. That account works on a retainer because the client's own planning horizon matches the commitment they're being asked to make. Contrast that with a marketing team whose translation needs depend on which campaigns get funded, and you can see why one converts and the other doesn't.

The next signal is multiple stakeholders. If three departments send you work through one contact, the account survives that contact leaving. If everything runs through one enthusiastic marketing manager, the retainer is that person's personal project, and it ends when they move on.

The third is a client who has already complained about your quoting turnaround. That complaint is a buying signal. They're telling you the friction of getting a price is costing them time, and a retainer removes it.

The anti-signals are just as legible. Seasonal concentration, where 70 percent of the year's volume lands in two months. Budget that gets approved per initiative. A client whose volume trend is declining, where a retainer is really a request that you help them smooth out a wind-down. And any client who opens the conversation by asking what discount a retainer gets them, before discussing volume. That's not a partnership conversation, it's a procurement tactic, and it's fine, but price accordingly.

What goes wrong with retainers and how to write around it

Scope drift is the main one. The retainer is signed for DOCX and XLSX files in two language pairs. Eight months later you're handling PPTX decks, a third language, and a set of PDFs somebody scanned. Nobody negotiated any of this. It arrived one file at a time and your project managers absorbed it because saying no to a good client over a single deck feels petty. Define the covered formats and language pairs, and treat additions as amendments even when they're small. The amendment doesn't have to cost the client anything the first time. It just has to exist, so the boundary stays visible.

Quiet non-usage is the counterintuitive failure. A client pays for six months and sends almost nothing. Your margin looks wonderful. Then renewal arrives, they run the numbers, and they either cancel or demand a reduction so steep the account stops making sense. If usage drops below a reasonable share of commitment for two consecutive months, raise it yourself. Offer to reduce the tier or extend rollover. You'll lose a little revenue and keep the account, and the client will remember that you flagged it instead of pocketing it.

Then there's internal confusion, which is less dramatic and more expensive. Project managers who don't know which client is on which model will quote a retainer client for work that's already covered, or fail to log overage on the one account where overage is the margin. This is an operations problem more than a pricing problem, and it's the point at which most agencies discover their spreadsheet doesn't cut it any more. A proper business management system tracks committed volume, consumption, and overage per account, which is roughly the moment retainers stop being a source of billing arguments.

How to test this without rewriting your price list

Don't convert your book of business. Pick two accounts that show the cycle pattern described above, and offer each a three-month pilot with terms written down: committed monthly words, the rate, overage at standard, rollover capped at one month, covered formats and language pairs named, rush handled as a surcharge, QA scope stated. Three months is short enough that a mispriced tier doesn't hurt, and long enough to see real usage.

At the end, compare three numbers against the same period under per-project billing: gross margin per account, hours your PMs spent on quoting and admin, and how many times you turned down other work because of a capacity promise. If margin held and admin time dropped, you have a model worth extending. If margin fell and nothing else improved, you sold a discount and called it a retainer.

Most agencies we've seen end up running both permanently, which is the honest answer. Retainers for the accounts with a real cadence, per-project for everything episodic, and a clear internal rule about which is which so nobody has to guess.