The One Hour Google Ads Audit That Finds the Waste

Start with what you actually bought

Every audit begins in the same place, and most people never open it: the search terms report. Not your keywords. The real queries people typed before they clicked.

Sort by cost, descending, last 90 days. Read the top fifty rows. In almost every small account we look at, somewhere between a fifth and a third of that spend went to searches the owner would never have chosen: job seekers, students, free-tool hunters, DIY researchers, and competitors' brand names typed by their own staff.

Note that Google omits low-volume queries from this report for privacy reasons, so what you see is a subset. The waste you can see is a lower bound on the waste that exists.

The six checks, in order of money found

1. Negative keywords. Build one shared negative list from what you just read, and add the perennials: free, jobs, salary, course, tutorial, template, DIY, cheap, reviews of competitors. Apply it at account level, not campaign by campaign, so it keeps working when someone adds a campaign next quarter.

2. Match types. Broad match plus Smart Bidding will find volume, and it will find it in places you did not intend. If your budget is under a few thousand a month, phrase and exact are usually where you should be living, with broad reserved for a single, contained, closely-watched campaign.

3. Conversion tracking. Look at what is set to primary. If page views, clicks on the phone number, and form submits are all counted as conversions and all counted equally, then Smart Bidding is optimising toward whichever is easiest to produce, which is never the one worth money. One primary conversion that means revenue. Everything else secondary.

4. Performance Max, honestly. PMax will spend the budget you give it and report a good-looking ROAS, partly because it harvests demand your brand already created. Split brand terms out into their own search campaign so you can see what PMax finds that you did not already have. Reporting has improved through 2026, but placement-level detail is still thinner than a search campaign's.

5. Geography and schedule. Check the locations setting is Presence, not Presence or interest, unless you genuinely want people merely reading about your city. Then check the hours: if nobody answers the phone at 9pm, and speed of response is what decides the outcome, you are buying calls you will lose.

6. The landing page. The most expensive mismatch in paid search is an ad for a specific service pointing at the home page. The click was for one question; the page answers twelve. This is where most of the remaining budget leaks, and it is a conversion problem, not a bidding one.

The two structural mistakes underneath

Bidding on demand you already own. Brand campaigns look magnificent because people searching your name were going to arrive anyway. Some defence against competitors bidding on your name is legitimate. Counting it as acquisition performance is not, and it is the most common reason a reported ROAS falls apart under examination.

Judging paid on last click alone. Paid gets full credit because it carries tracking. The channels that created the demand carry none, which distorts the whole budget over time for the reasons set out here. Put self-reported source on the enquiry form and read the two together.

What good looks like after the hour

You should finish with: one shared negative list, one primary conversion action, brand separated from non-brand, a location setting that matches reality, and a note of which ad groups point at the wrong page.

That is usually enough to remove a meaningful slice of spend without losing enquiries, which is the only kind of optimisation that shows up in the bank rather than the dashboard. Then, and only then, is it worth arguing about whether the budget belongs on Google or Meta at all, or whether the creative has simply been running too long.

Frequently asked

How often should we run this? The search terms review monthly. The full pass quarterly, and immediately after any structural change.

Our agency says the account is optimised. How do we check? Ask them for the last 90 days of search terms sorted by cost, and read it yourself. The answer to whether an account is well run is in that one export.

Can Fixora audit ours? Yes: the six checks above, written up with the specific rows costing you money. Give us read access and get the findings within 48 hours.