What a Google Ads Health Check Actually Looks At

by Ben Baker | Oct 29, 2026 | General
Laptop on a wooden workspace displaying a stylized Google Ads dashboard with performance line charts and campaign metrics.

Money leaves the account every month. Some of it brings in work. The rest sits in a grey area, and the reporting tab isn't much help beyond clicks and a cost per click that wanders about for reasons nobody's explained. That's the normal state of an account that's been running a while without anyone opening the bonnet.

What follows are the checks a review runs, roughly in the order you'd do them. Where the searches really come from, how loosely the keywords are matching, whether the conversions mean anything, where the clicks land, and which settings have drifted since launch. Treat it as a Google Ads account audit checklist you can work through yourself.

Start with the search terms report

Every account holds two lists, and people confuse them constantly. The keyword list is what you told Google you wanted to appear for. The search terms report is what people actually typed before your ad showed up. One's intention, the other's reality, and most Google Ads wasted spend hides in the gap.

Pull the report, set the date range to the last 90 days so there's enough volume to judge, and sort by cost rather than clicks. A single term that quietly ate a chunk of budget matters more than one that collected a handful of cheap clicks.

Then read down the list and put every term into one of three piles. Relevant and converting, leave alone. Relevant but not converting, note it and come back once you've looked at the landing page. Irrelevant, which is the pile that pays for the afternoon.

Irrelevant usually looks like the same few things. Job hunters, when words like jobs or salary or apprenticeship turn up next to your service. Students and DIYers searching how to, or template, or free. The wrong product entirely, which happens when a word in your trade means something else in somebody else's. Wrong location, where a suburb or a state you don't travel to appears. Competitor brand names, which might well be deliberate, but should be a decision rather than an accident.

And if the report comes back clean, that's a finding too. Usually means the match types are tight, which is next.

Building a negative keyword list

A negative keyword tells Google not to show your ad when a search contains that word. It's the main tool for closing the gap the search terms report just exposed.

Two levels matter here. Campaign-level negatives apply to that campaign only, which suits a word that's a problem for one service and perfectly fine for another. Account-level negatives live in a shared list and apply across everything, which suits the words that'll never be worth a click whatever you're selling.

A sensible starting list for an Australian service business: free, cheap, DIY, how to, jobs, careers, salary, apprenticeship, course, training, wholesale, second hand, plus the names of any states or suburbs you don't service. Add to it monthly instead of building it once and forgetting about it, because the search terms report refills.

Match types, and what they changed into

Match types decide how loosely Google gets to interpret your keyword.

Exact match, written in square brackets, once meant exactly that phrase. These days it takes in close variants too: plurals, misspellings, reordered words, and searches Google reads as carrying the same intent.

Phrase match, in quote marks, shows your ad when the search carries the meaning of your keyword, with other words either side.

Broad match, written plain with no punctuation, hands the decision to Google. It treats the keyword as a topic rather than a string, and leans on your landing page, your other keywords and the person's recent activity to work out what's related.

Here's the bit that catches accounts built a few years back. Google has steadily widened what each type covers, so a keyword that was tightly controlled when somebody wrote it might now be matching well past anything they had in mind. Nobody touched it. The definition moved, and the keyword came along for the ride.

Check which match type each keyword sits on, then check that against the search terms report. If broad match is bringing in work you want, leave it be. If it's feeding the irrelevant pile, tighten to phrase or exact and watch the volume over the next fortnight.

Are the conversions real?

A conversion is whatever you told Google counts as success. That's the whole definition, and it's where plenty of accounts quietly come undone.

Count every click on a phone number, every visit to the contact page, every newsletter signup and every brochure download as a conversion, and the reporting looks terrific. Automated bidding then works hard to produce more of them, because that's what you asked for. What you end up with is an account optimising towards button clicks instead of customers.

First check is what's being counted at all. Open the conversion actions list and read it as a list of claims: did this event make the business money, or could it plausibly have led to money?

Second, the primary and secondary split. Only actions marked primary feed the bidding. Keep the ones representing real outcomes as primary, demote the rest to secondary where they still report but stop steering spend.

Third, form submissions and phone calls are different events, and each needs setting up in its own right. An account where most enquiries arrive by phone but only form fills get tracked is seeing a fraction of what's happening. Call tracking takes its own setup, whether that's a Google forwarding number, a call asset, or a third-party number fed back into the account.

The difference between a lead and an enquiry

A tracked form submission is evidence that a form got submitted. It isn't evidence that a human being wants to buy something.

Once a quarter, take the conversion count for a month and hold it against your actual enquiry inbox and call log for the same month. Where the dashboard runs well ahead of the inbox, the difference is usually spam, test submissions, duplicates, or the same person sending twice because the confirmation never loaded.

Knowing that ratio changes how you read every report after it, and it tells you whether the cost per conversion on screen is a useful number or a comforting one.

Landing pages and the message match

The ad's only half the transaction. The page it lands on has to finish the thought.

Message match is the plain version. If the ad promises emergency plumbing in Parramatta, the page should say emergency plumbing in Parramatta near the top, in words somebody reads before scrolling. Send that click to the home page instead, and you're paying for it twice.

Things worth checking on every landing page. Does the headline repeat the promise in the ad? Is the form or the phone number visible without scrolling on a phone? Does the page load quickly on mobile data rather than office wifi? Is there one obvious next step rather than six? And does it answer the obvious objection, which is usually price, availability, or whether you cover that suburb at all?

This is also where paid and organic stop being separate projects. A page built to convert paid traffic is usually a better page full stop, and the work that goes into search engine optimisation, the headings, the load speed, the clarity about what you do and where you do it, lifts the paid performance of the same page. If you're paying for both, they shouldn't pull in opposite directions.

Budget, bidding and the quiet settings

Last pass is the settings nobody revisits after launch.

Automated bidding. Strategies like maximise conversions, with or without a target cost per action, are only as good as the conversion data underneath them. Get the conversion actions wrong, and the bidding efficiently buys the wrong thing. Check which strategy each campaign runs on, and whether it had real data to learn from.

Location targeting. Check the locations, and the setting beneath them that decides whether you're reaching people in your area or people anywhere who show an interest in it. The second one quietly serves ads interstate.

Search partners. Puts your ads on sites other than Google, and search campaigns generally include it unless somebody's turned it off. Check its performance separately, because it's often where the cheap clicks that never convert come from.

Display expansion. A search campaign can be set to spend leftover budget on the Display Network. Cheaper clicks, lower intent, easy to leave switched on without noticing.

While you're in there, glance at Google Ads quality score. It's Google's estimate of how relevant your keyword, ad and landing page are to each other, out of ten. Not worth chasing on its own, but a column of low scores points back at the message match problem.

Final word

Most reviews turn up the same three things. A search terms report full of searches nobody would have chosen, conversion actions counting events that were never going to pay an invoice, and a setting left on since launch that's been spending ever since. None of it calls for a rebuild, just an afternoon and somebody willing to read the account rather than the summary.

If you would rather someone ran through this alongside you, TFCS can take a look at your Google Ads account as part of our search engine marketing work and tell you what it finds in plain English. 

In need of a quality website to boost your business?