Search for a Google Ads audit checklist and you will find a lot of numbered lists. They are mostly the same list. Check your search terms. Check your negatives. Check your conversion tracking. All true, all unarguable, and none of it tells you whether your account passes.

The gap is that a tip is not a control. A control has to say what a pass looks like.

A control is a checkable statement

"Check your conversion tracking" is advice. This is a control:

Conversion actions defined. Pass: at least one primary conversion action is configured. Fail: no active conversion actions.

The difference matters because the second one can be wrong. You can point it at an account and get an answer you did not choose. That is the whole property that makes an audit worth running rather than reading.

Every control we run states its criteria that way, and all 161 of them are published with the thresholds that decide the verdict. Not because the list is secret sauce, but because a checklist you cannot check yourself is just an opinion with a number on it.

Severity is not the same as difficulty

The instinct is to rank findings by how hard they are to fix. That produces a list where "add three negative keywords" outranks "your conversions are double-counting", because one takes a minute and the other takes an afternoon.

Weight by what the finding costs instead. A conversion tracking fault distorts every other number in the account: the bid strategy is optimising toward a signal that is wrong, so the budget, the keyword decisions and the creative verdicts are all downstream of it. That is why conversion tracking carries the heaviest category weight in our scoring and why a critical failure there drags the score more than a dozen medium findings elsewhere.

The honest answer is sometimes "we cannot tell"

This is the part most checklists skip.

Some controls cannot be answered from the data you have. Server-side event deduplication, enhanced match quality, aggregated event measurement: none of these are readable from the Google Ads or Meta APIs. If an auditor marks them as passing, the auditor is guessing. If it marks them as failing, it is punishing you for an API limitation.

So the fourth verdict is N/A, and an N/A control leaves the denominator entirely. It does not quietly count as a pass, and it does not count as a failure. The report says which controls the data could not answer, and why.

An audit that always returns a full score on partial data is telling you something about the audit, not about the account.

Where to start

If you want the list, it is here, grouped by category and marked with severity. If you would rather have it run against a real account, that is what this product does, and the free plan covers one advertiser once a month with every one of the 161 controls.