Google Ads Audit Checklist, Worked Through on Real Accounts

Our Google Ads waste check gives you a score and a list of settings to look at. This post starts where that list stops. It takes the checks that most often turn up wasted money, shows what each one found in an account I audited this year, and walks through the fix inside Google Ads. If you have not scored your account yet, start with our Google Ads waste check. It is also where you can ask us to run a full audit for you.

The accounts below are anonymised. The numbers are the ones recorded at the time of each audit.

Example 1: A law firm paying for people looking for a government website

What the report showed. In September, the most expensive search term in a law firm's account was the name of a government reporting website, pulled in by one of the firm's phrase keywords. Roughly a third of their keywords were catching people who wanted a public website or a case status page. None of those people was going to call a lawyer.

The fix, step by step.

1. Open Insights and reports, then Search terms. Set the date range to the last 30 days and sort by cost.

2. Read the top 100 rows one at a time. For each, ask whether the searcher wanted a lawyer or only shared a word with your keyword.

3. Tick every navigational search and every definition search, such as “what is”, “meaning of” and “how to report”, and add them as negative keywords.

4. Note which of your keywords keep matching these searches, and pause them.

What changed. We paused dozens of keywords and added a long negative keyword list in a single sitting. The account came out far smaller, with every remaining keyword naming a lawyer or a live legal problem.

How to know it worked. Reopen the report two weeks later. The top rows by cost should now be people looking for legal help.

Example 2: Broad match on an account that had not earned it

What the report showed. The same firm had broad match keywords running without the conversion history to steer them. Broad match stretches a keyword to anything Google judges related, and without data that judgement is loose.

The fix, step by step.

1. In the keywords view, filter by match type to list every broad keyword.

2. Test each campaign against one rule: broad match stays only where the campaign has 30 or more conversions a month, runs Smart Bidding and has had its negatives reviewed in the last fortnight.

3. Where a campaign fails, add the keyword as phrase or exact, then pause the broad version.

What changed. The firm now runs no broad keywords at all, only exact and phrase.

How to know it worked. Tighter matching tends to show first as a better conversion rate on the same spend.

Example 3: An online store buying searches for rival stores

What the report showed. On an online store's account in Saudi Arabia, three phrase keywords were taking most of the non brand search budget for the period. The searches named rival stores and produced zero purchases. In the keyword report those keywords looked fine. Only the search terms showed the problem.

The fix, step by step.

1. In the search terms report, scan the top rows by cost for brand names that are not yours.

2. Add each rival name as a negative keyword, and keep the names in a shared negative keyword list so every Search campaign picks them up.

3. Check the list again each month, since new rivals appear.

Why it matters. Competitor searches look commercial, and they are. The searcher has simply already chosen someone else.

Example 4: The same retailer, bidding toward the wrong action

What the report showed. Every app purchase in that account was set to secondary, while a batch of engagement actions was set to primary. Smart Bidding was busy finding people who engaged, not people who bought, and the team was blaming the creative.

The fix, step by step.

1. Go to Goals, then Conversions, then Summary.

2. Open each engagement action and change it to secondary.

3. Make sure the purchase action is primary.

4. Open each campaign's own goal settings as well. A campaign can override the account default, so the account view does not always tell the whole story. On one audit the account view suggested bidding was chasing junk, while three of four campaigns were already set to purchases only.

What changed. The fix took two minutes. Spend stays the same, but the budget now points at buyers, so return on ad spend is the number to watch.

Example 5: A services account whose Performance Max ran on word game sites

What the report showed. On a services account we audited in the US, the Performance Max placements list ran to hundreds of websites and apps in a few months: word unscrambler sites, gematria calculators, people search directories, low income housing listings and job help forums. Between them they had spent a meaningful share of the budget and produced zero sales. They did produce leads, but the wrong ones: Spanish language enquiries from in app browser traffic that the front desk could not serve. The bidding was learning from those leads, and the account had no placement exclusions at all.

The fix, step by step.

1. Open the campaign, go to Insights and reports, and find the placements view.

2. Set the last 90 days and sort by cost.

3. Add every site and app that is clearly not your customer to an account level placement exclusion list, along with app categories that never convert for you.

4. While you are there, check whether Search Partners and Display expansion are on, and turn them off wherever the job is catching high intent search.

How to know it worked. For a clinic, the lead mix matters as much as the money. Front desk staff should stop chasing enquiries that were never going to book.

Example 6: A retailer whose ROAS looked several times better than it was

What the report showed. A Saudi retailer's app conversions were mapped through the wrong currency, which made return on ad spend look several times better than it was. On other accounts, every purchase was counted twice because two events were both set to primary.

The fix, step by step.

1. Place a real test order, or submit a real test lead.

2. Check that it lands once, with the right value and currency, in Google Ads and in your analytics.

3. If it lands twice, find the duplicate in Goals, Conversions, Summary and set one of the pair to secondary.

4. If the value or currency is wrong, fix the mapping before you trust any ROAS figure.

Why it matters. Every other check depends on the conversions column. If that column is wrong, you scale the wrong campaigns with full confidence.

Two smaller checks

Location. Open Campaign settings, Locations, Location options. If it reads “Presence or interest”, a local business is paying for people elsewhere who are only interested in its city. Switch to Presence, then read the locations report by user location rather than target location.

Device and hour. Segment 30 days by device and by hour, looking for spend with nothing to show for it. Do not cut on a hunch. If a segment converts now and then, lower the bid instead of excluding it.

After the cleanup

Leave the account alone for two weeks so Smart Bidding can learn from the cleaner signal. Then run the checks again, starting with search terms and placements, and compare conversion rate and cost per lead with the weeks before.