Retainer caps for marketing agencies: units, rollover and the 80% warning

Project agencies lose margin through scope creep on a single build. Retainer agencies lose it through a slow drift in what a month includes. In month one, the retainer covers four landing pages, twelve ad variants and a monthly report. By month nine, it covers six landing pages, twenty variants, a weekly report and a Friday call that has quietly become a strategy session.

Nobody decided to expand the retainer. Account managers, who are measured on retention, said yes to each extra item because the client was happy and the item was small. The fee stayed the same. The hours went up. By renewal, the agency is either losing money on the account or facing a difficult conversation about a price increase that the client sees as coming from nowhere.

An hours-only retainer invites the client to treat the agency as a pool of time, and gives the account manager nothing concrete to point to. A unit-based retainer lists what the month includes in terms the client already thinks in.

Hours still matter internally, so keep an hour value per unit for planning. But the client conversation happens in units. "You have used 11 of your 12 ad sets this month" is clearer and less confrontational than "you are at 54 of 60 hours".

Unused units are as much of a problem as overflow. If the client uses eight ad sets in a quiet month, do the other four roll over? If the rule is not written down, the answer will be negotiated every month, usually in the client's favor.

• No rollover: simplest, but clients feel they lose value in quiet months.
• Rollover for one month: unused units carry forward once, then expire. This is the most common fair compromise.
• Unit swaps: unused units of one type can be exchanged for another at an agreed ratio, for example two ad sets for one landing page.
• Quarterly pooling: units are counted per quarter rather than per month, which smooths campaign peaks.

Whatever you choose, show the balance on every monthly report. A visible balance makes the cap feel like a shared budget rather than a hidden limit.

The most useful single habit for retainer agencies is a warning when the client reaches about 80% of a unit cap in a month. It should go out before the last units are spent, and it should always include options.

Clients rarely react badly to this. It reads as the agency managing their budget carefully. The alternative, delivering the extra work and either absorbing it or invoicing for it afterwards, produces exactly the friction account managers are trying to avoid.

Repeated overflow is a signal, not a nuisance. If a client exceeds the cap three months running, the retainer is the wrong size, and that is a good conversation to have: it is a growth conversation, not a scope fight.

1. Track overflow per unit type for each account.
2. After two consecutive months over cap on the same unit, flag it in the monthly review.
3. Propose a retainer adjustment that matches actual usage, with the data.
4. If the client prefers to stay at the current level, agree what gets deprioritized.

Agencies that do this well often grow accounts at renewal without a hard sell, because the evidence for a larger retainer has been shared month by month.

Retainer caps fail when the people closest to the client feel they cannot enforce them. Account managers are usually rewarded for client happiness and retention, not for margin, so every incentive points toward yes. If leadership wants caps to hold, the incentive has to change: track overflow absorbed per account alongside satisfaction scores, and make it explicit that scheduling work into next month is good account management.

It also helps to give account managers a small discretionary allowance, for example two hours a month per account, that they can spend on goodwill without asking anyone. Discretion within a limit feels very different from a rule with no flexibility, and it keeps the cap credible for everything above it.

The difficulty for a growing agency is visibility. One account manager can count units for three clients. Nobody can do it reliably across thirty accounts with requests arriving through Slack, email and briefs in Asana or ClickUp. TWINJAM parses the retainer schedule into unit caps, counts requests against them across connected channels and alerts the account lead privately at the threshold you choose.

The client never hears from the software. They hear from an account manager who knows exactly where the month stands, and that is the whole point.