Debrief · 11 October 2026 · 6 minutes

Google Ads audit: the checks that find money, and the one that finds the truth

Around 40% of digital ad money is wasted, on one agency's count. Figure out where you're at.

The count is one auditor's sample of Australian advertisers, measured with its own tool. Read it as a signal, not a census.

44%of audited digital spend wasted across 37 Australian companies in one quarter. Next&Co, Q2 2024, via Mediaweek

Your own number is in your own account, and the first reading can flatter. In the September 2026 snapshot, CobberBank's Google Ads account reads A$118.06 per application against a target of A$150. Take brand search out and the rest runs at A$161.20.

Finding that is the job of a Google Ads audit, a PPC audit or a paid media audit. The name changes. The list barely does.

Eleven of the checks find money: spend that could be working harder. The twelfth finds the truth: whether the account's numbers agree with anything outside the account. Most of the value sits in the twelfth. It is also the one a checklist tends to stop short of.

The eleven that find money

1. What counts as a conversion

Open the conversion settings before anything else. List every action marked as primary, because those are what the Conversions column counts and what the bidding chases. Look for two actions counting the same thing. Look for soft actions, like a page view, counted beside hard ones. CobberBank counts one action, an application. That is the clean answer.

2. What a conversion is worth

Conversion value is often an assumption. In the September 2026 snapshot, CobberBank's account shows A$5,033,050 of value. That is 9,151 applications at an assumed A$550 each. It is not revenue. Know which of your values are measured and which are estimated before anyone divides by them.

3. Brand and non-brand, apart

People who search your name already know you. They convert cheaply. Blended with everything else, they flatter the account. That is the split in the opening. In the September 2026 snapshot, brand search brings 49.3% of CobberBank's conversions at A$73.75 each. One account, two different stories.

4. Search terms and negatives

Keywords are what you bid on. Search terms are what you paid for. Read the search terms report for queries that will never become customers and add them as negatives. Then check that the negative lists are not blocking queries you want. A check that cannot run should say so. The demo room holds no search term data for CobberBank, so its negative keyword rule reports missing data instead of a finding.

5. Match types and structure

Broad match with automated bidding can work well. It needs clean conversion data and regular attention. Check for the same keyword in several ad groups, for campaigns bidding against each other and for ad groups so wide that one ad has to answer fifty intents.

6. Performance Max

Left alone, Performance Max can buy your brand searches and report them as its own wins. The demo room flags this for CobberBank. In the September 2026 snapshot, Performance Max holds 20.5% of Google spend at A$173.24 per application. Brand search converts at A$73.75 and has no protected budget. The suggested action is to add brand exclusions, protect the brand budget and then read the Performance Max numbers again.

7. Bidding strategy and targets

Check each campaign's bidding strategy against its goal. Check each target against what the campaign delivers. A target nobody has revisited in a year is a setting, not a strategy. Large, frequent changes to targets and budgets send the bidding back into its learning period.

8. Budgets and pacing

Compare spend with plan, month by month. In the September 2026 snapshot, CobberBank planned A$959,040 and spent A$1,080,367, which is 12.7% over. Caught in the second week, an overrun is a decision. Caught at month end, it is an explanation.

9. Settings that leak

Four defaults deserve a look. Location targeting set to "presence or interest" reaches people who are only interested in your area. Search campaigns opted in to search partners and the Display Network spend outside Google search. Ad schedules and device adjustments may be untouched since launch. Auto-applied recommendations may be changing the account without anyone approving them.

10. Ads, assets and landing pages

Check that every ad group has current ads, no disapprovals and a full set of assets. Then follow the click. In the September 2026 snapshot, CobberBank's brand search turns 58.9% of clicks into engaged visits, meaning visits that stayed. Its Demand Gen campaign turns 14.9%. That campaign bought 138,702 clicks and 126 applications. It may be doing the job it was given. It should be a decision and not a surprise.

11. Tags and names

Check that auto-tagging is on, that the conversion tag fires once, that Consent Mode is in place and that enhanced conversions are set up. Then check that analytics can read your campaign names. In the September 2026 snapshot, GA4 can name one of CobberBank's five Google campaigns. That campaign carries 30.8% of the spend. The rest arrives under a label the ad account does not use.

The twelfth: do the numbers agree

Every check above divides by the account's own conversion count. None of them tests it.

Pull one month from three places: Google Ads, your analytics and your CRM. Use the same dates and compare like with like. Here is paid search in the September 2026 snapshot. CobberBank's two search campaigns report 7,114 conversions from 170,298 clicks. GA4's Paid Search channel reports 3,284 conversions from 116,897 sessions. That leaves 53,401 clicks with no matching paid search session. The third count is the CRM's, the one finance believes.

When the three disagree, the cause is usually ordinary. A click identifier was lost in a redirect. Two people defined a conversion two ways. The attribution windows differ. Across every paid channel in the September 2026 snapshot, GA4 credits one in twenty of the sessions that arrived on a paid click to a paid channel. The five usual causes and the fix for each are in Why GA4 and Google Ads never agree, and which one to believe.

Until the three counts agree or the gap has a written reason, treat every number from checks 1 to 11 as unreconciled. A cost per conversion is only as good as the conversion count beneath it.

What a free audit is for

Many agencies offer a Google Ads audit at no cost. Take it, and know what it is. A free audit is a proposal. Its job is to win the media account, which is a fair thing for an agency to want. So it concentrates on the checks that show what a new manager would change. Those are the eleven. A good one does them well.

The twelfth sits outside that brief. It needs access to analytics and the CRM. It takes weeks, not hours. Its finding is no reason to change who manages the media. So read a free audit for the eleven. Commission the twelfth separately, from someone who does not manage the media.

Where a Number Check fits

NR's Number Check is the twelfth check, run across the last twelve months and every source you report from. It is narrower than a marketing audit. It looks at the numbers, not the people. It is read-only, it involves no media management and you keep the pack whether or not you continue.