What a good Google Ads audit must contain

An audit is a basis for decisions

An audit is good if afterwards you know what to do in which order, what it is likely to bring in euros and what to leave alone. A ticked-off checklist does not achieve that. It shows where the account departs from the textbook, but says nothing about whether the departure costs money.

A finding therefore has four parts:

• Evidence: a report, period and figure that you can check yourself.
• Euro range: the effect on contribution margin, not on reported conversions.
• Effort and risk: what the change costs and what it could break.
• Decision: a clear proposal that you can accept or reject.

Without the euro range, it remains an observation. "Ad group with only one ad" or "missing assets" are such observations, as long as nobody shows what costs depend on them.

In substance, the audit has to answer five questions:

1. What drives the bids, and does it match the shop system?
2. Which campaigns and parts of the range make money after ad spend and which do not, including within PMax and the feed?
3. Which part of the reported revenue would probably have come without an ad, for example on brand terms, and how could that be tested?
4. What will be changed, in what order and with what effort?
5. What stays untouched, and under what condition does that change?

Measurement first, or no ranking holds

If conversion measurement is wrong, every ranking in euros is wrong too. That is why the audit starts there. Primary conversion actions feed the Conversions column and bidding, provided their standard goal is used for bidding; secondary actions only count in a custom goal. So the check is which actions actually steer and whether their values hold up when you reconcile them with the shop system.

A worked example shows how much this matters. Suppose a shop spends €20,000 a month, Google Ads reports €100,000 of conversion value, and the contribution margin rate before advertising is 30%. On paper, €10,000 remains. But if the reported value includes 19% VAT, it is around €84,000 net, €25,200 of contribution margin and €5,200 after advertising. The account earns half as much as you thought, without anything having changed in a single campaign.

If the audit finds an error like this, the rest of the prioritisation is provisional, and the report has to say so. An auditor who still delivers a precise ranking is calculating on a basis they have just disproved themselves.

How a finding becomes a euro figure

The calculation is simple: affected costs or revenue, times the share that can realistically be changed, times the consequence for contribution margin. Scale it up to a year and deduct the effort. For the line between profit and loss you need the break-even ROAS with all costs included.

The same worked example, this time with clean net values and invented numbers:

Two rows carry the result; the other two do not belong in the report. My rule of thumb: a finding stays in only if the lower end of the range covers the effort at least three times, because estimates tend to be too optimistic. That is a working rule, not a measured value.

Ranges are more honest here than point values. Anyone who states an effect to the exact euro is feigning precision. The calculation also has a limit: risks without an ongoing effect, such as unresolved ownership of the account, fit poorly into this table. They belong in a separate block, together with the revenue that would be affected if things went wrong.

The do-not-touch list is the best quality test

Whoever sells an audit earns money by finding something. A US agency describes the conflict itself: an auditor paid to find problems is inclined to overstate them, one who wants to keep the client is inclined to play them down. Some providers attach an offer for ongoing management straight to the report, or credit the price against their follow-on programme. There is nothing wrong with that; I sell ongoing management myself and have the same conflict. You should simply know that an audit is also a sales tool.

The do-not-touch list is the only part of the report the auditor earns nothing from. That is why I consider it the best test of whether someone has done the maths. It states specifically what stays, why, and when that changes. For example: "This campaign looks untidy but makes money after ad spend, so no restructuring until a test shows otherwise." Or: "The account structure stays; restructuring costs days and solves none of the documented findings."

Its value lies in protection. Every intervention in a working campaign can cost revenue, and that damage appears on no list of faults. If the list is missing entirely, either nothing that works was found, or nobody looked for it. The exception is an account with broken measurement: then it is hard to show what works, and the list rightly stays short and provisional.

What is an automated score worth?

As an assessment of an account, very little. According to Google, the optimisation score is an estimate between 0 and 100% that changes when you apply or dismiss recommendations. You can also reach 100% by dismissing every recommendation. According to Google it plays no part in Quality Score. The figure therefore says nothing about profit.

One practitioner reports an account with a score of 99.1% in which 18.7% of search term spend produced no conversion, while the cleanest account in his sample scored 48.8%. These are individual cases, but they show the direction. In addition, Google requires 70% for Partner status in the registered manager account. Anyone who is a Google Partner, including me, therefore has a stake of their own in this number. It does not belong in an audit as a yardstick.

The recommendations behind it are useful as a list of suggestions that someone checks one by one against the margin. More on this under optimisation score. In my view the same applies to the scores of external audit tools: they find deviations quickly but know neither your margin nor your returns.

How long does an audit take, and what does it cost?

There is no reliable market price for the DACH region (Germany, Austria and Switzerland), only individual providers' figures (as of November 2026): a British specialist charges £297, a Greek agency €300 plus VAT, both for around 30 checkpoints in five working days. A US agency estimates its market at $250 to $500 for small accounts and $2,500 or more for large ones. That says more about the scope of the service than about its value.

On duration, my view is this: the account itself is quickly read, because according to Google read-only access lets you browse linked accounts and run reports, and an outside specialist needs no more than that. The time goes into reconciling with shop data and margins. Once those are available, I consider days normal for a mid-sized shop, not weeks. If an audit arrives after an hour without a single question about your margins, it is an export.

You can work out a justifiable price yourself. Suppose your account spends €1,000 a month and an audit realistically lifts contribution margin by €100 a month. That is €1,200 in the first year. If you apply the factor of three from above, at most €400 is left for the audit. At this size a low-cost checklist can be enough, as long as it takes the measurement seriously. Calculating each finding in euros only pays off once enough money moves through the account for it to bring in more than it costs.

Four questions for anyone offering you an audit

• Do I get a euro range and the calculation for each finding?
• What data do you need apart from account access? Anyone who does not ask about margin and shop revenue cannot calculate on contribution margin.
• Does the report include a list of what should not be touched?
• Can I implement the report without hiring you afterwards?