Google Ads getting irrelevant searches? How to review what you are paying for

If Google Ads is getting irrelevant searches, start with the search terms report, not a wholesale change to your keywords. Review the searches that generated cost, separate genuinely unwanted intent from relevant searches that have not converted, then add negatives at the right match type and scope. My recommendation is to block what your business cannot serve before deciding what simply needs better performance.

This guide is for an existing account where clicks are arriving but the searches, or the resulting enquiries, look wrong. It covers Search campaigns and the important differences for Performance Max. You do not need another account launch checklist. You need to know what to exclude, what to keep and what the report cannot tell you.

Start with the searches, not the keyword list

A keyword is what you target. A search term is what someone searched. They are not interchangeable. Positive exact match can include close variants with the same meaning or intent, so switching everything to exact match is not a promise of word-for-word traffic. Our guide to Google Ads keyword match types explains that wider distinction.

In Google Ads, open "Campaigns", then "Insights and reports", then "Search terms". Select the campaign you are investigating and a date range that makes sense for its sales cycle. I would use a period with enough activity to see patterns, while keeping very recent clicks separate if their enquiries or sales have not had time to appear.

Show cost, clicks, conversions and, where useful, conversion value. For keyword-based Search campaigns, add the "Keyword" column, which is not shown by default. Keep the campaign, ad group and "Match type" visible too. Sort by cost first, then download the report if grouping similar searches is easier in a spreadsheet.

The report's "Match type" describes how the search related to the matched keyword. It is not necessarily that keyword's configured match type. A broad keyword can produce a row labelled "Exact". Check the keyword itself before blaming, or congratulating, a setting.

Make three decisions before adding negatives

I would group the visible searches by intent, rather than judge every spelling as a separate problem:

• Exclude: The person wants something you do not offer. Recruitment searches in a customer acquisition campaign, an unsupported product or a service you never provide are candidates.
• Keep: The search describes a customer need you can meet. Different wording is not automatically irrelevant.
• Investigate: The search could be useful, but its economics or lead quality are unclear. Check the landing page, conversion records and actual enquiries before deciding.

A search for a service you do not sell does not become useful because it recorded a form submission. Equally, zero conversions do not prove a relevant search is wrong. Tracking may be incomplete, the buying decision may take longer, or the landing page may fail to explain the offer. There is no universal number of clicks after which every search deserves a negative.

For lead generation, compare the report with the team's qualification notes where attribution allows. "Conversions" and "people we can actually help" are not always the same number. Our guide to Google Ads metrics and lead quality explains how to connect the account numbers with commercial outcomes.

Read the whole query before excluding a word. "Free" might describe a price shopper you do not want, or a free quote you genuinely offer. "Jobs" could mean vacancies or completed jobs. A downloaded negative list has not read your service menu.

Check how much cost the visible queries explain

The search terms report is not a complete transcript. Google omits some searches with insufficient query activity to protect privacy. Do not assume every paid click will have an individually visible query.

For a like-for-like Search review, compare the cost on the visible query rows with the corresponding Search cost for the same campaign, dates and network scope. Visible query cost divided by that matching total gives the share of cost you can inspect individually. Do not mix an account total with one campaign's rows, or add summary rows to the individual rows and count the same cost twice.

Search terms insights can help with the remaining picture. Its categories take omitted queries into account in aggregate, but do not reveal every hidden query. Categories are generated labels, not necessarily searches that triggered an ad. A label containing an excluded word is therefore not proof that your negative failed.

The gap is unitemised cost, not confirmed waste. Use the available themes and business outcomes to investigate it, and say plainly where query-level evidence stops.

Choose a negative that blocks the problem, not the customer

Negative match types behave differently from positive match types. For Search:

• Negative broad: All the negative's terms must be present, in any order. A negative broad keyword repair course blocks "repair course Brisbane" and "course for repair".
• Negative phrase: The terms must appear in the same order. "repair course" blocks "repair course Brisbane", but not "course for repair".
• Negative exact: The terms must appear in the same order without extra words. [repair course] does not block "repair course Brisbane".

These are word-based exclusions, not a way to exclude an entire meaning. Google says negatives automatically account for casing and misspellings, but synonyms and singular or plural versions need separate consideration. Excluding "course" is not the same as excluding "courses" or "training".

Choose the narrowest scope that solves the actual problem. An ad group negative can keep an unwanted service out of one Search ad group without blocking another that sells it. Use campaign scope when the exclusion applies throughout that campaign. Reserve account-level negatives for searches you do not want anywhere across the account's relevant Search and Shopping inventory.

If only one query is clearly wrong and the surrounding intent is useful, a negative exact keyword is a cautious option. If a whole phrase reliably identifies an unwanted service, a phrase exclusion may be more useful. Check what legitimate searches the proposed negative could also block before saving it.

Performance Max needs a different reading

Performance Max now provides actual search term reporting, including terms that did not convert. In the Search terms report, choose "Search terms and landing pages for Performance Max" from the dropdown to review queries alongside their landing pages and ad formats. Advice that PMax only exposes broad themes is out of date.

You can apply negative keywords at campaign or account level, but they affect Search and Shopping inventory only. They do not exclude unwanted Display or YouTube content. Those require the appropriate content suitability controls.

Do not divide visible query cost by the whole PMax campaign's spend and call the difference hidden search waste. PMax serves across channels. Keep any cost comparison within matching inventory, and assess campaign outcomes separately. Store visits and store sales conversions are not available in the search terms report, another limitation if those matter to your business.

A blank matched-keyword field is not automatically an error either. Dynamic Search Ads and Shopping do not use keywords in the same way as keyword-based Search campaigns. Review the targeting and landing page that actually served, rather than searching for a keyword that does not exist.

Worked example: identify waste without inventing savings

Hypothetical example, not a Nexus client result. Assume an Australian repair business spends AU$1,200 on one Search campaign during the review period. Campaign, date and network scopes match. Visible individual query rows total AU$960. Tracking is working, the qualification records can be linked to these searches, and the example's outcomes have had time to arrive.

The visible rows contain AU$720 from 90 clicks on relevant repair searches, producing six qualified leads. Recruitment searches account for AU$96 from 12 clicks. Searches about repair costs account for AU$144 from 18 clicks, with no qualified leads yet. These three groups total AU$960.

You can inspect 80% of the campaign's cost individually: AU$960 divided by AU$1,200. The remaining AU$240 is unitemised. The AU$96 recruitment group is 8% of total campaign spend, or 10% of visible query cost. Those percentages describe different denominators; neither means 10% of all spend is proven waste.

I would exclude the clearly identified recruitment intent from this customer acquisition campaign. I would investigate the cost-research group rather than reflexively block "cost". The relevant group's AU$720 divided by six qualified leads is AU$120 per qualified lead, but that is the group's cost, not the whole account's acquisition cost.

AU$96 is historical cost on unwanted intent. It is not a guaranteed future saving. Removing those searches may change where the campaign spends; it does not guarantee the replacement clicks will produce six more leads.

Your next review should end with a decision log

Record the query or intent, its cost, why you excluded or retained it, the negative match type and scope, and the change date. At the next review, check new queries and qualified outcomes, allowing for your sales cycle. Repeated irrelevant themes deserve a targeting review, not an endlessly growing list of isolated exclusions.

If you cannot confidently separate bad targeting from weak measurement, Nexus Google Ads consultancy can help you review the account and prioritise changes while your team keeps running it. You supply all final ad creatives and retain responsibility for comments, messages and customer follow-up. The useful outcome is a clearer decision about paid traffic, not another report nobody acts on.