The short answer
A retainer is the best cash flow instrument a small agency has, and the fastest way to give away margin quietly. The difference is 3 decisions, all made before the first invoice:
Bill in advance, on a fixed date, for the month ahead.
Say what the fee buys, in hours, deliverables or availability. Pick one and write it down.
State what happens to what is unused, because silence here is read generously by clients and strictly by you.
Set a notice period on both sides, so the end of the arrangement is not a surprise.
Get those right and a retainer turns lumpy project income into something you can plan around. Get them wrong and you have a fixed fee attached to an expanding obligation, which is the shape most retainer resentment takes.
Decide what the fee actually buys
Nearly every retainer argument traces back to the 2 sides having different answers to this. There are 3 workable models and they behave differently:
Model | What the client buys | Where it breaks |
|---|---|---|
Capacity | A number of hours or days each month | Hours become a ceiling the client wants to fill exactly |
Deliverables | A defined monthly output, such as 4 posts and 1 report | Each item quietly grows in size |
Availability | Priority access and a guaranteed response time | Hardest to sell, easiest to defend |
Capacity is the most common and the most argued over, because an hour is a unit the client can count and dispute. Deliverables sell more easily, since the client sees what they get. Availability is the honest model for work that is genuinely on-call, and it is the only one where a quiet month is obviously not a refund situation.
Whichever you pick, write the boundary in the same document. A retainer without a stated edge becomes a subscription to unlimited work, which is the version of scope creep that never produces a single visible moment to object to.
Bill in advance
Invoice before the month starts, not after it ends. The argument for it is not preference, it is what the arrangement is:
You are holding capacity. The client is buying the fact that you kept the time free, which has a cost whether or not they use it.
Your costs come first. Salaries and contractors are paid during the month, so billing afterwards means funding the client for 30 days before the invoice even starts its terms.
It removes the argument about usage. Billing in arrears invites a monthly conversation about whether the hours were really spent.
It is normal. Software, rent and insurance are all paid before the period they cover. Clients accept the pattern readily when it is presented as ordinary.
A workable arrangement is invoicing on the 20th for the month beginning on the 1st, on short terms. That gives accounts payable time to process it in an existing payment run without you waiting. Pair it with the payment terms you use elsewhere rather than inventing a separate set.
Where a client refuses advance billing outright, that is worth noticing rather than conceding immediately. Sometimes it is policy. Sometimes it is the first visible sign of a cash problem, and the same client turns up later in the pattern of paying every invoice a week or 2 late.
The money arrives before it is earned
Advance billing creates a gap between cash and revenue that is worth understanding, because it affects both your accounts and your tax position, and because spending an advance as though it were profit is a common way to get into trouble.
Money received for work not yet done is unearned. In your books it sits as a liability until you perform, and it converts to revenue month by month as you do. That matters most when a retainer ends early, because an unearned balance is something you may owe back.
For tax, the US position depends on your accounting method. The IRS sets out in Publication 538 that under the accrual method advance payments are generally reported as income in the year received, though an eligible taxpayer can elect to postpone part of it to the next tax year and no further than that. In the EU there is a parallel timing rule for VAT: the European Commission states that under Article 65 of the VAT Directive, where a supply is paid for in advance, VAT becomes chargeable when the payment is received, on the amount received. UK practice follows a similar tax point on receipt.
The practical consequence is the same everywhere. Tax can fall due on retainer cash before the work behind it is delivered, so the money in the account overstates what you have. Set the tax aside when it lands, and talk to your own accountant about the method that suits you.
The rollover trap
What happens to unused hours is the clause most often left out and most often argued about. 3 options, in descending order of safety:
No rollover. Unused capacity expires at the end of the month. Defensible, because you held the time open, and simplest to administer.
Capped rollover. A stated share, commonly around a quarter of the monthly allocation, carried for a stated window and then expiring.
Unlimited rollover. Avoid. It compounds into an obligation you cannot meet.
The last one deserves the warning. A client who uses half their hours for 3 quiet months arrives in month 4 expecting several times the normal output for the same fee, at exactly the moment you have booked the capacity elsewhere. Unused hours accumulate faster than any month can absorb, and the resulting conversation is about a promise rather than about a preference.
Whatever you choose, report against it monthly. A short note showing what was used, what was delivered and what expired keeps the boundary visible without anyone having to defend it. It also gives you the evidence when the retainer needs repricing.
What the agreement needs to say
1 page covering 7 things, and most retainer disputes never start:
The fee, the billing date and the payment terms.
That invoices are raised in advance for the period ahead.
What the fee buys, in the model you chose.
What is outside it, and how additional work is priced and approved.
What happens to unused capacity.
The notice period on both sides, and whether an advance is refundable on early termination.
A review date, so the price can change without anyone starting a difficult conversation from scratch.
The review date is the item agencies omit and later need. A retainer priced against a scope from 2 years ago is almost always underpriced, and a scheduled review is a far easier route to fixing it than a request out of the blue.
When a retainer client starts paying late
The recurring shape makes this both easier and more dangerous. Easier, because you see the pattern quickly across several invoices. More dangerous, because the work continues automatically, so exposure builds every month while nobody makes a decision.
Set a rule in advance, and apply it without drama: if the current month's invoice has not been paid by a stated day, next month's work does not start. Because the fee is billed in advance, that rule is straightforward to state and easy for a client to understand. It is also far gentler than a mid-project stop, and the suspension clause that governs it should be the same one you use everywhere.
Watch the total rather than each invoice. 3 unpaid months of a $6,000 retainer is $18,000 of exposure to a single client, which is the situation described in when your biggest client is your latest payer.
FAQ
Should a retainer be billed in advance or in arrears?
In advance, for the month ahead. The client is paying for capacity you are holding, your own costs fall due during the month, and advance billing removes the recurring argument about whether the hours were used.
What should happen to unused retainer hours?
The safest position is that they expire monthly, because you held the time open. A capped rollover of roughly a quarter of the allocation, expiring after a stated window, is a reasonable compromise. Avoid unlimited rollover, which compounds into an obligation you cannot deliver.
Is retainer money I have been paid in advance actually mine?
Not yet, in accounting terms. It is unearned until you do the work, and it may be repayable if the arrangement ends early. Tax timing is a separate question and can fall due before the work is delivered, so take advice on your own method.
How much notice should a retainer have?
Enough to refill the capacity, which for most small agencies means 30 to 90 days on both sides. State whether an advance already paid is refundable if the client terminates within a paid period.
What do I do when a retainer client stops paying?
Apply the rule you set in advance: the next month does not begin until the current invoice is paid. Because the fee is billed ahead, that boundary is easy to state and reverses the moment payment clears.
How do I raise the price of an existing retainer?
From the review date you put in the agreement, using the monthly usage reports as the evidence. A scheduled review is a much easier conversation than an unexpected request, which is the main argument for writing one in at the start.
Not legal, accounting or tax advice. Revenue recognition, VAT and the treatment of advance payments depend on your country, your entity and your accounting method — confirm your position with a qualified adviser.