Field note · Traps

The retainer trap, in one diagram.

22 janvier 2026 · 7 min · Un-gated, no email wall, no download form · For anyone signing one this quarter

Open your last bank statement and look at the retainer line. Notice how it's the only invoice with no “what did we get this month?” attached. A retainer is the cleanest invoice a services firm can write and the slowest invoice a buyer can stop. That asymmetry is the entire diagram. Once a month a number leaves your account; once a month, on average, nothing of equal value arrives, because nothing was named that had to. The trap is geometric, not moral: no one set out to mistreat you, the contract just rewards both sides for not naming what's true. Here's what the math looks like before you sign.

There are two reasons a vendor will offer you a retainer: they want predictable revenue, and you want to feel covered. Both are reasonable. The trap is what neither of you puts in the contract: a definition of done. With a project, “done” is the deliverable and the invoice closes when it ships. With a retainer, “done” is the end of the month, and the next month opens with the same number on the same calendar reminder. The work has nothing to push against, so it expands or shrinks to fit the month, usually shrinks, because the urgency is on your side and the calendar is on theirs.

We see this with fractional CTOs, with web maintenance, with “strategic” marketing retainers, with IT support, with copywriters, with so-called managed-services. The pattern is uniform: a monthly number, an ambiguous scope, and a relationship that requires effort to end. The effort is the trap. Most operators won't spend two weeks of energy to save four thousand a month they've already mentally written off. So the four thousand keeps leaving, for years, with no exit conversation that's pleasant enough to start.

What a retainer actually is

A retainer is not a service. It is a call option you've paid for. You're paying every month for the right to call on someone, who is in turn committing to be available when you call. If you don't call, the money is still gone, that's the option you bought. If you do call, the work usually comes out of a notional “bucket of hours” that resets monthly, and the vendor is incentivised to make calls feel mildly inconvenient so the bucket doesn't overflow.

There are markets where call options are honest and valuable: legal counsel for a regulated business, on-call security response, an insurance policy. In each of those, the option exists because the thing you might need is genuinely emergent, expensive when it arrives, and impossible to procure quickly. Software and marketing are mostly not emergent. The decision to redo a website is a decision you can see coming a quarter out. The decision to migrate a CRM is a six-month conversation. Buying call options against work that has lead time is paying for a kind of insurance you didn't actually need.

The compounding math

Take a retainer of five thousand a month. That's sixty thousand a year. Three years in, by which point neither side has a strong narrative for why the rate hasn't gone up, you're a hundred and eighty thousand into a relationship whose deliverables would, if listed on one page, fit on one page. The cost is real. The harder cost is the second-order one: every problem in your business has, by default, been routed to the retainer, which means the muscles inside your business that would have learned to solve those problems have atrophied for three years.

When the retainer ends, and it will, because the comfort eventually outweighs the work, the company has to re-learn the things the retainer was doing on its behalf. The bigger the retainer was, the more those skills decayed. So leaving a retainer is more expensive than it looks; it isn't just the money you've spent, it's the operational competence you didn't build. Vendors don't sell this part. They don't need to. It's a load-bearing pillar of the lock-in.

Now layer the rate creep. Most retainers carry a clause about annual increases tied to some index. The index always wins. The rate compounds at three or four percent a year. Over a five-year retainer, you've paid roughly the original number multiplied by five-point-five. The vendor has paid down a sales-and-marketing cost they spent once, in year one, and the rest is margin. The deal doesn't look like that on the day you sign. It looks like that on the day, three years later, when you finally do the arithmetic.

The diagram, then, is two lines on the same chart. Yours: a flat horizontal payment with a small upward creep. Theirs: a sharp initial cost in winning your business, then a long flat plateau of low-effort delivery. You're paying for the slope that already happened. They're collecting the rent on it for as long as the conversation about ending it is harder than the cost of continuing.

When retainers are legitimate

Some work is genuinely retainer-shaped. On-call security incident response, where the team has to know your stack cold before the bad day arrives. Managed IT for a business too small to staff a sysadmin, where the day-to-day really is uniform across months. Outsourced bookkeeping, where the work is volumetric and predictable. Compliance/audit engagement, where the cadence is fixed. In each of those, the deliverable per month is genuinely consistent, the response time is the product, and the retainer is the honest contract.

Outside that set, retainers exist because nobody on either side had a sharper alternative on the day they signed. Software development, design, marketing, advisory, almost all of these are better expressed as named projects with a number and a finish line. If you can't name the project, the retainer isn't the right answer, it's the answer you give when the right answer doesn't exist yet.

Lines to use to escape

If you're already in one and want to leave cleanly, the line is: “Let's price the next three months as a fixed-scope project instead.” It tests both sides. If the vendor can write down what they'll do for the same money with a deliverable attached, your retainer was overpaying for a project. If they can't, your retainer was overpaying for nothing.

If you're about to sign and want to avoid the trap entirely, the line is: “Send me the first three deliverables in a project format. We'll do those, then talk about the next quarter.” This is what the relationship would look like if it were honest. Most vendors will pivot to it. The ones that won't are telling you the call-option was the product.

If a fractional CTO is on the table, the line is: “What's the named decision I'm hiring you to lead this quarter?” If there's a clear answer (a vendor selection, a re-architecture call, an audit), pay them a flat fee to lead it and re-evaluate at the end. If there isn't, you don't need a fractional CTO yet. You need a few hours of someone's attention on the actual question.

None of this is anti-vendor. We send work to people on retainers all the time when the retainer fits the shape of the work. The point is that the retainer should be the conclusion of a conversation about the work, not the conversation itself. Most aren't. Most retainers were the conversation, and the work was negotiated to fit the monthly bill that was already agreed.

← All field notes
No email wall, no download form, nothing to unlock. Forward this to the colleague who signs the invoices. If your retainer is the legitimate kind, you'll feel it within a paragraph. If it isn't, you'll feel that too.