The decision took about four minutes. We were on a monthly call, being walked through a deck, and someone said the phrase “we’re seeing some really encouraging signals.” Nobody in the room could name which ad had spent the most money that month. That was it. Not a blow-up, not a bad quarter — just the quiet realisation that the people being paid to run the account had a worse picture of it than we did, and we were only looking on Sundays.
What we did next was the mistake, and it’s the reason this page exists. We started interviewing replacements. Within a week we had three proposals, all of them good-looking, none of them written by anyone who had opened our ad account. We were about to hand the same job to a new supplier using the same brief, based on the same understanding of the problem — which is to say, none.
You cannot brief a replacement on a problem you have not diagnosed. You can only describe the symptom you noticed, which is usually “it stopped working.”
So we stopped, and we read the account properly first. Not a dashboard skim — a proper teardown, the kind you do when nobody is going to be embarrassed by the answer because everybody involved has already left.
What was actually in there
Four things, and none of them were the thing we thought was wrong.
The four findings below are our own account, on a retainer we ended — operator record, not a client case. Every one of them has since turned up in accounts we have read for other people, which is why they are worth your time rather than just ours.
Almost nothing was being produced. There were eleven ads live. Not eleven that month — eleven, total, carrying the whole account. A winning ad has a working life measured in days at real spend, and eleven ads cannot outrun that. The account wasn’t underperforming because of bad strategy. It was underperforming because the supply of new creative had quietly fallen to roughly nothing and nobody had put a number on it.
Old creative was still spending. Two ads from the previous November were still running in July, still promising a deadline that had passed eight months earlier. They were not flagged, because nothing was checking. Every month they had appeared inside an aggregate that looked fine.
Money was in the wrong channel. One channel was returning roughly twice what the other was, on less than half the budget. The reallocation was not a strategic insight. It was arithmetic, available in the platform, to anyone who looked at both accounts on the same afternoon.
The cheapest revenue was switched off. Automated email flows — the ones that fire when somebody abandons a checkout — were built, written, and sitting in draft. Not missing. Built. Somebody had done the work and never pressed the button, and the monthly report had no line where that would have shown up.
The uncomfortable bit
None of those four findings required a genius. They required somebody to spend an afternoon in the account with no incentive to make it look good. That is a real distinction, and it is the one that explains most of what goes wrong in this relationship.
The agency was not lying to us. The monthly report was accurate. It just answered a different question than the one that mattered. It reported on what had happened rather than on what was structurally about to keep happening, and there is no line in a performance summary called “the flows are built but off.”
The problem is almost never the strategy. It is the throughput — and throughput is invisible in every report that measures results instead of output.
Creative supply is the variable. Almost everything else is downstream of it.
Here is the part the category does not lead with, because it is unflattering to how agencies are staffed. At real spend, a winning ad fatigues in a matter of days. That is not a controversial claim; it is the reason your best ad from March is invisible in April. Which means the number that decides your ceiling is not your targeting, your bidding or your budget. It is how many genuinely different ads you can put in front of the algorithm per week.
A human creative team is expensive per asset and slow per cycle, so a retainer sized to a founder’s budget buys a handful of assets a month. That is a supply problem dressed up as a performance problem — and no amount of strategy fixes a supply problem.
So we built the other thing. An engine that drafts at volume, scores every concept against a fixed rubric before anything earns production, and kills its own work cheaply. Senior operators direct it and sign off everything that ships — the engine never decides what a client sees. We ran it on our own brand first, with our own money, which is how we know what it does and also how we know what it does not do.
What we would tell you to do before you hire anyone
Get the account read first. Not by the people pitching to run it — by anyone whose read you can check. The read is worth more than the shortlist, for three reasons.
It tells you what you are actually buying. “We need better creative” and “we need eight times more creative” are different purchases at different prices, and only one of them is on most rate cards.
It gives you something to brief with. A supplier handed a diagnosis performs differently from a supplier handed a symptom, and you can tell within one cycle which one you got.
And it gives you a scoring rubric for the pitches. When someone walks you through their plan, you will know within about ninety seconds whether they have looked at your account or at their template.
The honest objections
Fair, and worth saying plainly: we sell a managed service, so we benefit if you buy one. What we can do is make the argument checkable. Our prices are published on the site rather than gated behind a call. The audit is free and requires no call. And the report you get names the things you can fix yourself — several of which cost nothing and involve no supplier at all.
If the honest read is that you should stay where you are, the report says so. That has happened.
It is a sales instrument. We are not going to pretend otherwise — we give it away because a good one argues for us better than a deck does. What makes it not a bait-and-switch is that it is complete and it is yours: the findings, the arithmetic, a 30-day plan and a batch of ad concepts built for your brand, whatever you decide afterwards.
No call is required to receive it, and none is booked on your behalf.
Below roughly £3k a month on ads, the honest answer is that no managed service is worth it and we will tell you that rather than sell you one. The band the work is built for is £5k–50k a month, founder-led, without an in-house creative team. Between £3k and £5k it depends entirely on margin — ask us and we will tell you which side of the line you are on. (Bands are category norms in GBP, not a measurement of your account.)
Above that band there is usually already an internal team, and the useful question changes from “who runs this” to “what is the team missing”. Still worth the audit; possibly not worth the retainer.
Not yet, and we would rather say so than borrow logos. We are new as a service. What we have instead is our own account — the 247 winners, the +32% ROAS, the money that was ours when it worked and ours when it did not — plus one client case under NDA where cost per lead went from €22.14 to €4.56.
If proof from a comparable brand is what you need to move, an established agency is genuinely the safer choice and you should take it. We would rather lose the deal than manufacture the evidence.
Just fired your agency?
Get the account read before you hire the next one. Every pound of wasted spend found and named, in 24 hours, free, no call.
You get the full teardown — where the money is going, what the pages are costing you, what your competitors are running — plus a batch of ad concepts built for your brand and scored before they were shown to you, and a 30-day plan in the order the money moves.
It takes about two minutes to start. A senior operator signs off every report before it reaches you, and the whole thing is yours whatever you do next.
Get your free Crank Audit2 minutes · findings in 24 hours · no call required