Audit your own ad account: the nine checks, in order

Most ad-account audits you can buy are a slide deck of screenshots you could have taken yourself. The value was never in the looking — it's in knowing which order to look in, because the first three checks find most of the money and the last three are where people usually start.

So here's our order, with what a bad answer looks like at each step. It takes an afternoon. If you'd rather we ran it, the audit is free — but nothing below is held back to make that more likely.

1. Does your conversion event fire once, at the right moment?

Before any judgement about performance, establish whether the numbers describe reality. Trigger a real purchase or lead on your own site and watch what the platform records. You are looking for one event, at the right step, with the right value attached.

Bad answers: the event fires on page load rather than on completion; it fires twice because a tag manager and a native pixel are both installed; the value is missing, so the platform optimises toward cheap conversions rather than valuable ones. Every one of these is common, and every one invalidates everything downstream — including the last six months of decisions you made from the dashboard.

2. Do platform-reported results reconcile with your back end?

Take last month. Compare platform-attributed conversions with what your commerce backend actually recorded. They will not match — attribution windows and modelling guarantee that — but you're looking for the size and the direction of the gap, and whether it's stable.

Bad answer: platform-reported revenue is a large multiple of what the business banked, and nobody knows the ratio. If you don't know your gap, you don't know your real cost per acquisition, and every ROAS target you've set is arbitrary.

3. Is spend going where you think it's going?

Sort every campaign, ad set and ad by spend, descending, for the last 30 days. Read the top ten rows out loud.

Bad answers: a retargeting campaign consuming prospecting-sized budget against an audience of a few thousand people; a paused-strategy campaign still running because nobody turned it off; one ad set quietly taking 60% of a budget you believed was split evenly. This check takes four minutes and finds real money more often than any creative critique.

4. Are you bidding against yourself?

Look for the same or heavily overlapping audiences running in parallel ad sets. Fragmenting one audience across many ad sets splits the learning, multiplies the cost of every test, and puts your own budget on both sides of the auction.

Bad answer: a dozen ad sets, each with a small budget, each never leaving the learning phase. Consolidation is usually the single highest-value structural change available to a £5k–50k/mo account.

5. Does anything ever exit the learning phase?

Check how many of your ad sets have accumulated enough conversions in a week to stabilise. If most haven't, your budget is spread too thin for the platform to optimise at all, and you're paying learning-phase prices permanently.

6. Is your retargeting audience big enough to be worth a campaign?

Retargeting flatters itself in reporting — it converts people who were already going to convert — and small audiences fatigue viciously because the same people see the same ad repeatedly.

Bad answer: significant budget against an audience of a few thousand, with frequency climbing week over week. Check frequency alongside audience size; the two together tell you whether you're building demand or just annoying a small group of people expensively.

7. How many distinct creatives actually carried spend?

Not how many exist — how many took meaningful spend last month. Most accounts discover the answer is three or four, regardless of how many were uploaded.

Then do the replacement-rate arithmetic: the creatives you need per week is the number you want live divided by their effective life in weeks. If your production is below that number, everything else in this list is treating symptoms.

8. Are you testing angles, or testing decoration?

Line up your last ten creatives and ask what claim each one makes. If eight of them make the same claim in different colours, you have not run ten tests — you've run one test ten times.

This is where our own scoring taught us something uncomfortable. When we scored our fifteen launch concepts, hook and brand fit barely moved the overall score — they were near-constant, because a house style makes them constant. What separated the best from the worst was clarity and offer: whether a stranger could tell what was being sold and what they'd get. Decoration varies easily and matters least.

9. Would a stranger understand the offer in three seconds?

The last check is the cheapest and the one most often skipped. Show your top-spending ad to someone outside your company for three seconds, then ask them what the company sells and what they'd get if they clicked.

Bad answer: they describe the mood. Our own worst-scoring concept scored 45 out of 100 with a perfectly respectable hook and a clarity score of 40 — visually strong, and it said nothing. That ad is still published on our lab page with its score attached, because an ad that fails this test looks exactly like an ad that's working right up until you read it.

Work top-down and don't skip. Measurement first, because bad data makes every later judgement wrong in ways you can't see. Structure second, because it's usually the biggest single lever and costs nothing to change. Creative last — not because it matters least, but because fixing creative on top of broken measurement means you'll never know whether the fix worked.