Is an ads agency worth it under ten thousand a month
The arithmetic that tells you whether a management fee can pay for itself yet — and the three cases where the honest answer is no.
Most of the pages answering this question are written by agencies, which makes them a strange place to look for a no. So here is the no first, then the arithmetic that tells you when it turns into a yes.
When is an ads agency not worth it?
Three cases, and they are common enough that we turn work away on them.
The first is thin margin. If you keep 20 per cent of a sale after everything, every pound of fee has to generate five pounds of new revenue before it has paid for itself. Not five pounds of revenue in total — five pounds of new revenue that would not have happened. That is a hard test, and most accounts under moderate spend fail it.
The second is a broken step further down. If your product page converts at a fraction of what it should, or your offer only sells to people who already know you, media buys you more visits to the same problem. Fix the page, then buy traffic to it. An agency that takes the retainer without saying this is selling you activity.
The third is not enough data for the platform to learn from. Meta's own delivery guidance is explicit that an ad set needs roughly fifty optimisation events in a rolling seven-day window before it leaves the learning phase and delivery stabilises. If your budget and your price mean five purchases a week, no operator on earth gets the algorithm out of learning. You are paying a specialist to wait.
What does the fee have to earn to break even?
This is a one-line calculation and almost nobody does it before signing. Take the monthly fee. Divide it by your gross margin. That is the extra revenue the arrangement has to produce every month just to arrive at zero.
A worked example, using your numbers not ours. A $3,000 monthly fee at 40 per cent gross margin needs $7,500 of additional monthly revenue to break even. At 25 per cent margin the same fee needs $12,000. If your whole business is doing $30,000 a month, you have just asked an agency to lift you 40 per cent before anybody profits.
| Monthly fee | Gross margin | Extra revenue needed to break even | Extra revenue needed to be clearly worth it |
|---|---|---|---|
| $1,500 | 25% | $6,000 | $12,000 |
| $1,500 | 50% | $3,000 | $6,000 |
| $3,000 | 25% | $12,000 | $24,000 |
| $3,000 | 50% | $6,000 | $12,000 |
| $5,000 | 25% | $20,000 | $40,000 |
| $5,000 | 50% | $10,000 | $20,000 |
The fourth column doubles the third, and that is deliberate. Breaking even on a fee is not a reason to pay it — you did the work of managing an agency and got nothing. A fee is worth paying when it returns roughly twice what it costs, because that is where the arrangement survives a bad quarter.
Now compare the break-even number to your ad spend. If you are spending $3,000 a month on media, the fee is asking for improvement of a size that the media itself cannot produce. That is the real reason floors exist. It is not snobbery about small clients. It is arithmetic that does not care how much anyone wants the deal.
What changes above the threshold?
Somewhere around $5,000 to $10,000 a month of spend, three things change at once, and they compound.
- The account produces enough conversions each week to clear Meta's learning threshold, so testing produces signal rather than noise.
- The fee becomes a smaller share of total media, so a modest percentage improvement in efficiency can actually cover it.
- The number of decisions per week exceeds what a founder can make well between everything else — which is the point where expertise starts saving money rather than costing it.
Against Stackmatix's 2026 market bands — 10 to 20 per cent of spend, or flat retainers running from roughly $1,500 to $10,000 a month — a $10,000 media budget carrying a $2,000 fee means the operator has to find a fifth more efficiency to be free. That is a plausible ask on a neglected account. On a $2,000 budget carrying the same fee, it is not.
The cost trend matters too. LocaliQ and WordStream's analysis of 1,180 United States campaigns run between April 2024 and June 2025 found the average Facebook cost per lead at $27.66, up 20.94 per cent year on year. Standing still gets more expensive every year. Below the threshold that trend is a reason to be careful; above it, it is a reason to have someone whose job is fighting it.
What about a freelancer instead?
This is the middle path and it is underrated, including by us. A good freelancer at a few hundred pounds a month will keep an account tidy, catch the obvious waste, and tell you when the budget is too small for what you want.
What you give up is real, though, and it is mostly creative volume. One person managing eight accounts does not produce ten new pieces of creative a month for any of them, and creative is where most of the remaining upside on Meta lives. You also carry the single-point risk: freelancers get busy, get a full-time offer, or go quiet in the week your campaign breaks.
The honest split is this. If the account mostly needs maintaining, hire a freelancer. If it needs new ideas produced at volume, that is an agency-shaped problem, and it is worth waiting until you can afford one rather than buying half of one.
How do you test an agency for one month?
You do not need a year to find out. You need one month structured so that the answer is unambiguous.
- Agree the single number before anything starts, and write it into the contract. One number, with a date.
- Give them the account and the budget you already spend — not more. A test funded by a budget increase cannot tell you who caused the change.
- Require the creative count in writing. Ten new pieces, or three, but a number, delivered in the month.
- Take a screenshot of the account's last ninety days on day zero. Agencies rarely change the past, but memories of the baseline get generous.
- Ask for one weekly note against the number, in writing, and read whether it reports problems as well as wins. A month of only good news is a reporting problem.
- At day thirty, compare against the baseline, not against the pitch.
One month will not prove a full return, especially on a considered purchase with a long buying cycle. It will prove the things that predict one: whether creative actually shipped, whether the account got cleaner, whether the person you met is the person doing the work, and whether they tell you the truth when a week goes badly.
So what is the answer?
Under roughly $5,000 a month of spend, on thin margin, with a page that does not convert: no, and any agency that tells you otherwise is telling you what you want to hear. Under that number with fat margin and a product that sells: a freelancer, and revisit in six months.
Above it, with margin that survives the arithmetic above and a real appetite to spend more: yes, on a flat fee, against a number you agreed, reported weekly, and cancellable when it stops working. Retainers should be losable. If cancelling yours would be awkward rather than obvious, the terms were written for the wrong person.