Google Ads Account Audits

What you are buying is a diagnosis, not a hand on the account

An audit is a fixed-scope diagnostic with a defined end. Read access to the account is enough to run it. Nothing is paused, nothing is rebuilt, and no bid is changed, unless you separately ask for that work.

What you get at the end is a document: the findings, ordered by how much money each one is costing you, with the evidence attached to each so you can check the reasoning yourself. It is yours. Give it to your in-house team, your current agency, or the next one. Nothing in it depends on me being the person who implements it.

That separation matters more in paid search than almost anywhere else, because the party auditing the account is very often the party who built it, or the platform that profits from the spend. An audit conducted by whoever will be paid to fix what it finds has an obvious tension in it. An audit conducted by the platform selling the media has a different one.

The ongoing management, structure work and testing is a separate, retained engagement. It is deliberately not bundled here.

Conversion tracking is examined before anything else, and it is usually where the problem is

Say this plainly: the most expensive problem in most accounts is not the bidding. It is that the conversion data the bidding runs on is wrong.

Automated bidding optimizes toward whatever you have declared a conversion. It has no way to know whether that thing has value. So if the conversion action double-counts, fires on the wrong event, or was never verified after a change to the site, then every bid, every reported return, and every decision anyone has made from the reporting is built on a bad number — and the more budget the system gets, the more efficiently it spends into the error.

What is checked:

• Whether each conversion action corresponds to an outcome you would pay for, rather than a page view, a form open, a visit to a contact page, or a phone-number click nobody completed.
• Whether the same event is both hardcoded on the site and imported from analytics, producing two conversions from one form fill.
• Which actions are set as primary and therefore feeding bidding, versus secondary and merely observed. Accounts frequently bid on something the business does not care about.
• Attribution and conversion window settings, and whether they match how the business actually buys.
• When tracking was last verified end to end, and whether a migration, consent banner or checkout change has happened since.
• For lead generation, whether anything connects the reported conversion to a qualified lead, or the account is optimizing toward form fills of unknown quality.

If tracking fails this pass, it is the first finding and everything downstream is reported with that caveat attached, because it has to be.

Account and campaign structure, and where budget competes with itself

Structure is examined for what it does to money, not for whether it is tidy.

Campaign segmentation is judged against budget control: campaigns split for reporting convenience rather than for spend control usually mean nothing can be funded independently. Brand and non-brand mixed into one campaign produce a blended cost per conversion that describes neither.

Then match type distribution and the sprawl it creates. Match type sprawl is the condition where the same query is eligible across multiple campaigns and multiple match types at once. The account bids against itself, attribution muddies, and no single campaign's numbers describe what happened.

Ad group coherence gets its own check, because Google uses other keywords in the same ad group to infer intent when matching. An ad group holding unrelated keywords is not just untidy — it changes how every keyword in that group matches.

Finally, bidding strategy against available data: whether the account produces enough conversions for an automated strategy to learn from, or whether the strategy is optimizing noise and producing large swings in cost per conversion that nobody can explain.

Search terms versus keywords — the finding measurable in dollars

The keywords are what you bought. The search terms report is what you actually got. The gap between them is the single most persuasive thing an audit produces, because it can be stated as an amount of money against queries with zero conversions and no plausible commercial intent.

The report is read for spend concentration, for query patterns rather than individual terms, and for the specific categories that recur: research intent where you are selling, job seekers, existing customers looking for support, competitor names bid on by accident, and the long tail of near-misses that broad match brings in.

Negative keyword coverage is examined alongside it, at all three levels — account, campaign, and ad group. What matters is not whether negative lists exist but whether they are maintained. Most accounts have a list written at launch and untouched since, which was accurate for a market that has moved on. Where negatives are applied is also checked, since a negative at the wrong level either fails to block or blocks traffic that a different campaign needed.

Performance Max: asset groups, search themes and the exclusions nobody sets

Performance Max runs one campaign across YouTube, Display, Search, Discover, Gmail and Maps, with creative organized into asset groups — Google's term for a collection of creatives centered on a theme or audience. Audience signals are inputs to Google's own intent modeling, not targeting in the older sense.

The configuration audit covers asset group construction and whether the themes are distinct enough to report on separately, search themes, final URL expansion, and brand exclusions.

Brand exclusions get particular attention because of Google's documented priority rule: exact match keywords in Search campaigns take priority over Performance Max, while phrase and broad match keywords are prioritized comparably to Performance Max search themes. In an account built on phrase and broad match — which is most modern accounts — Performance Max is therefore competing for branded and high-intent queries that were converting anyway, and reporting those conversions as its own. It then appears to be the best campaign in the account.

The test is measurable: branded query volume and cost before and after Performance Max launched, and what happens when brand exclusions are applied. Reference: Google's Performance Max documentation. Where Shopping is involved, feed health is examined as part of the same pass, since a poor feed limits everything downstream of it.

Budget pacing, geography and the hours nobody has revisited

Pacing is examined for the two opposite failures: campaigns that exhaust the daily budget before the hours that actually convert, and campaigns underspending against a cap set for a season that ended. Both are invisible in a monthly total and obvious in an hour-by-hour view.

Geographic targeting is checked for what the setting actually does, which is not always what the person who set it believed. Location settings distinguish between people in a place and people showing interest in a place, and the difference routinely puts spend into countries a business does not serve. Bid adjustments and exclusions are read against where conversions genuinely come from rather than where the business would like them to come from.

Ad scheduling is reviewed on the same basis. For any business where the conversion is a phone call, the schedule against staffed hours is a real finding and frequently an untouched one.

Also in scope: responsive search ad asset coverage, extensions and assets, landing page relevance and load speed, and the Quality Score components, read as diagnostic signals rather than as a score to be chased.

What lands on your desk, and how long it takes to produce

A written findings document. Each finding states what was observed, the evidence behind it, what it is costing or risking, and what to change — specified precisely enough for whoever manages the account to execute without further conversation. Findings are ordered by recoverable spend and risk, not by how easy they are.

It also states what was checked and found sound, which matters. An audit that reports only problems is not describing the account; it is building a case.

One to three weeks from access being granted is typical. What moves it: the number of accounts and markets, the number of active campaigns, whether Shopping and a product feed are involved, whether conversion tracking has to be traced through a tag manager and analytics property to be understood, and how much historical data exists to read patterns against. A single Search campaign with clean tracking is a short job. Six accounts across four markets with Performance Max and a feed is not.

When you do not need an audit, and how you will know this one was worth buying

You do not need one if the account launched last month — there is not enough data yet, and the honest advice is to wait until the search terms report has something in it. You do not need one if you already have a specific, evidenced diagnosis and simply need someone to act on it; that is management, not diagnosis. And you do not need a paid audit to discover that you are not tracking conversions at all. That you can establish in ten minutes yourself, and the answer is the same either way: fix that first.

You will know the audit earned its cost if you can act on it without me. Every finding should be executable by your current team, and the wasted-spend findings should be verifiable within days of the negatives going in. The tracking findings are the ones to check hardest, because they are the ones that change what every other number in your reporting has meant.

A note on free audits. Most are a sales artifact, generated by a tool, weighted toward findings that require the provider's help. Some are useful. The test is whether the document reasons from the evidence in your account or reads as though it could have been produced for any account.

Frequently Asked Questions

Published 2026-08-11