When Google Ads is not converting, the usual next move is to spend more or rewrite the ads. That puts the next pound into the same unknown. Run seven checks first. They cover what the account can see, what it bids towards, and what happens after the click.
Do them in this order. A later check is hard to trust if an earlier one is broken. Use the last 30 days as the window, unless you already know a shorter stretch is the one that changed.
1. Conversion tracking
If the account cannot see a lead, a sale or a call, every other number is a guess. Open the conversion actions and see which ones are recording. Then compare that list with what the business actually calls a result.
Submit a test through the form the ads use. Do the same for a click-to-call number if you pay for calls. Confirm the test shows in the conversion action, not only as a view of the thank-you page. That URL can load even when the form failed a step earlier.
- List the actions that are recording, and the ones that are silent.
- Send one test lead and one test call, and note where each appears.
- If you do not trust a conversion value, leave it out. Bidding can run without a made-up number.
2. Conversion goals used for bidding
Tracking can be on, and the campaign can still bid towards the wrong thing. Look at the goals included in the campaign, not only the list of actions in the account.
A campaign that treats a newsletter signup, a PDF download and a sales enquiry as the same result will buy the cheapest of the three. Often that is the one the business did not mean. Note which actions are primary for bidding, and which are only observed.
- Open each live campaign and record the goals it bids on.
- Separate observed actions from the ones that change the bid.
- Change one goal at a time. Judge it over the last 30 days, after it has had time to record.
3. Search terms and negatives
The keyword list is what you hoped to buy. The search terms report is what you bought. Export the search terms for the last 30 days and read them as a customer would.
Sort by cost. Mark terms that cannot lead to the offer, even when they share a word with a keyword you chose. Add those as negatives at the level where they leaked. Use the ad group if the leak is local. Use the campaign if it is wider.
- Export search terms for the last 30 days and sort by cost.
- Add negatives for queries that cannot become the offer.
- Read the negatives you already have. A wide negative can block the enquiry you are trying to buy.
4. Match types and keyword intent
Match type decides how far a search can wander from the keyword. A broad keyword for a specific service will buy research, jobs, definitions and competitor names. An exact keyword will not repair a page that talks about something else.
Group keywords by intent. Someone comparing suppliers is not the same as someone ready to book, and not the same as someone looking for a job or a login. If those sit in one ad group, they share one ad and one landing page. Split them, or accept that the ad is a compromise.
On the rows that cost the most, read the keyword beside the search term. If you cannot say why that search should hit that keyword, the match is too loose or the keyword is wrong. Phrase and exact still need negatives. They leak less. They do not leak never.
- Label each live keyword as ready to buy, comparing, or something else.
- Split ad groups where one ad is being asked to serve two intents.
- On the costly rows, write the sentence that links the search to the keyword. If you cannot, change the match or the keyword.
5. Ad-to-landing-page match
Open the ad that spent the money and the URL it used. The promise in the headline should be the first thing the page confirms. If the ad names a service, a place or a time frame, the page has to say the same thing without a hunt.
Check the path on a phone. The headline, the offer and the way to enquire should be visible without a long scroll. If the page opens with a company story and the form is at the bottom, the ad and the page are having two different conversations.
Where several ads point at one generic page, note which ad is spending. Give that ad a page that matches it, or narrow the claim. Check that an enquiry is possible from the screen the person actually sees.
- Pair the top-spending ad with its final URL and read them together.
- Repeat that on a phone, from the click through to the form.
- Give a heavy-spending ad its own page, or narrow the claim in the ad.
6. Landing page speed and form friction
A matched page can still lose the person. Treat speed and the form as two checks inside this one. You want to know if the page becomes usable, and how hard it is to send an enquiry.
Open the landing page on a phone, on a normal connection. You do not need a score to act. If the form is delayed by large images, a pile of tags, or a layout that jumps, fix that before you buy more clicks.
Then count the fields. The first reply needs a name and a way to respond, not the whole brief. If a field is there because a CRM wants it, and the first conversation does not, take it off the paid page.
Check the thank-you step. If the form says it sent and nothing arrives, or the button fails on a second tap, you will under-count enquiries. Fix the form, then confirm the test lead still records.
- Complete a test enquiry on a phone and note where you waited or mis-tapped.
- Remove fields the first conversation does not need.
- Confirm the thank-you step and the conversion action still agree.
7. Location, schedule and device settings
Spend can look like it is "not converting" because it is happening where or when you do not serve. Open locations, ad schedule and devices for the campaign.
On locations, confirm you are targeting people in the places you serve, unless you have a reason to pay for people researching that place from elsewhere. Read the location report for the last 30 days. If a town or country you do not serve has cost, exclude it. If a place you do serve has no impressions, the targeting or the bid is keeping it out.
On schedule, match the hours you buy to the hours someone will answer. If the business only takes calls on weekdays, overnight and weekend spend is buying clicks nobody will handle. Cover that time, or stop buying it. Check whether those hours produce any of the actions you kept in the second check.
On devices, treat a gap as a clue. If phones take the cost and the form is hard on a phone, this is the same problem as the previous check. Fix the page before you switch phones off. Switching the device off hides the fault.
- Read locations, schedule and devices for the last 30 days.
- Exclude places you do not serve, and hours nobody will answer.
- If phones spend and do not enquire, fix the mobile page before you cut the device.
These seven checks are a sequence. Tracking first, then the goal the bids use, then the queries, the match, the page, the friction, and the settings that decide who sees the ad. If you change several of them in one afternoon, you will not know which one mattered.
When the checks disagree, or the account is too tangled to read this way, the next step is a marketing performance audit. That review covers measurement, paid media and the pages they land on, before anyone opens a new campaign. The services and engagement models page sets out how that work is scoped if you want someone else in the account with you.