How to audit a Google Ads account

Most audits are delivered as a long list of observations sorted by whatever the auditor noticed first. That produces a document nobody acts on, because the reader cannot tell which three findings matter and which forty are noise. The fix is to audit in dependency order, so each pass only asks questions the previous pass has made answerable.

Start with conversion tracking, not with keywords

Every other number in the account is derived from the conversion data. If that data is wrong, the cost per acquisition is wrong, the campaign comparison is wrong, and Smart Bidding has been optimising toward the wrong thing for as long as the error has existed. Checking it takes twenty minutes and it changes the meaning of everything else in the audit.

What to check

• Open the conversion actions list and count how many are set to Primary. Every Primary action is something bidding optimises toward. Ask what each one is worth, and whether it should really be steering spend.
• Look for duplicates. A common pattern is the same lead counted once by a Google Ads tag and again by a GA4 import, which silently halves the reported cost per acquisition.
• Check the counting setting. A conversion set to count Every rather than One will multiply on a thank-you page that visitors reload or return to.
• Confirm enhanced conversions is on. Without it the platform is learning from a smaller and non-random subset of your real conversions.
• Check the conversion window against your actual sales cycle. A thirty-day window on a ninety-day cycle systematically under-credits the campaigns that start conversations.

Then do the test that most audits skip: submit a real form or complete a real purchase, and confirm the conversion appears where it should, once, with the right value. A tag that fires in preview mode has proved almost nothing about what happens on a real device.

Finally, compare the account's reported conversion count against your CRM for the same period. A gap is normal. A gap you cannot explain is the most important finding in the audit, and it usually points at marketing analytics services work rather than at anything inside Google Ads.

Find the spend that reached the wrong queries

With the measurement understood, the next pass looks for money that bought traffic you never wanted. This is the fastest source of recoverable budget in most accounts, and unlike structural work it needs no learning period to take effect.

• Pull the search terms report for the last ninety days, sorted by cost, and filter to terms with zero conversions. Read the top fifty by spend. In accounts using broad match without a strong negative list, the drift from the intended intent is usually obvious within the first ten rows.
• Look for competitor and job-seeker queries, which are two categories nearly every account pays for by accident.
• Check whether campaigns are bidding against each other on the same terms. Cross-campaign negatives are missing far more often than they are present.
• Review placement reports on Display and any Performance Max campaign that serves there. Mobile game and made-for-advertising placements are still a meaningful drain.
• Check the brand exclusion setting on Performance Max. Without it, PMax will bill you for traffic your brand campaign already owned at a lower cost.

Wasted spend is the only audit finding that pays back the same week you fix it. Everything else needs a learning period.

Ask whether the structure allows good bidding

Structure is not an aesthetic question. The only thing that matters is whether the way campaigns are split lets you control budget and lets the bidding algorithm gather enough signal in each container to learn from. Those two goals pull against each other, and the right answer is a compromise rather than a rule.

• Count conversions per campaign per month. A campaign generating a handful of conversions cannot support its own bidding target, and consolidating it into a sibling usually improves both.
• Check whether budgets are capped. A campaign limited by budget for most of the month is a strategy decision that was probably never made deliberately.
• Look at whether ad groups are tight enough that the ad text can honestly match the query. If one ad group covers six unrelated intents, no responsive search ad will match all of them.
• For Performance Max, check that asset groups map to something you would want reported separately, such as a product line or a service. A single asset group means a single opaque number.

Resist recommending a full rebuild by default. A rebuild discards the conversion history that Smart Bidding depends on, so performance drops while the new campaigns relearn. Rebuild when the structure genuinely prevents good bidding, and say so plainly when it does not.

Only now, look at the bidding

Bidding is the last pass because a bid strategy can only be judged once you know the conversion data is real, the traffic is roughly the right traffic, and the campaign has enough volume to learn.

• Check whether the target CPA or ROAS was derived from the business economics or picked to match recent performance. The second is common and it quietly locks the account at its current level.
• Read the change history. Targets changed every few days never complete a learning period, so the account spends its life relearning.
• Compare impression share lost to rank against impression share lost to budget. They point at completely different fixes and they are frequently confused.
• For any campaign switched to a new bid strategy in the last three weeks, note that its numbers are not yet readable, and say so in the audit rather than drawing a conclusion from them.

Write the audit so it can be acted on

A finding without an owner, an effort estimate, and an expected effect is trivia. Group findings by root cause rather than listing them individually, because thirty findings usually share three causes, and fixing the causes closes them in batches.

1. Lead with the measurement findings, because they change how everything else should be read.
2. Then the recoverable waste, with the amount quantified from the search terms report rather than estimated.
3. Then the structural changes, each marked with whether it requires a learning period.
4. Then the bidding changes, which should mostly be a consequence of the three passes above.
5. End with what you checked and found to be fine. An audit that only lists problems gives no sense of proportion.

If the account turns out to be measured correctly and still underperforming, the constraint is usually downstream of the click rather than inside the account, which makes conversion rate optimization the higher-leverage next step. Ongoing google ads management is what keeps the four passes from drifting back out of alignment.