Reading your search terms report

Owners skip this because the interface buries it and because the report is boring. It is also the highest return twenty minutes in paid search, and nothing else in the account comes close.

The gap it reveals is always bigger than expected. You bid on a service you sell. Google matches you to questions, comparisons, job listings, and homework assignments that share a word with it. You do not find that out from a performance summary. You find it from the raw list.

Where is the report and what am I looking at?

In Google Ads it sits under Insights and reports, listed as Search terms. Set the date range to the last thirty days. Add the columns for cost, clicks, and conversions if they are not showing.

Then sort by cost, highest first. This ordering matters more than it sounds. Sorting by clicks shows you popularity. Sorting by cost shows you where the money went, and money is what you are auditing.

Google's own documentation is explicit that the report is not complete. Search terms without enough query activity are omitted to keep with Google's data privacy standards. Those searches spent your money and you will never see them written out. The newer Search terms insights view groups some of them into themes and subthemes, or aggregates them as other queries, without exposing the queries themselves.

Read each term as a sentence a stranger said out loud. Not as a keyword. "Cheapest emergency plumber" and "emergency plumber near me" are the same keyword and completely different customers.

How do I decide which terms are buyers?

Sort each one into four buckets as you read. It goes faster than it sounds.

Buying. They named the service and signaled urgency or location. These are what you are paying for. Note which ones convert and consider giving them their own tighter ad group.

Browsing. How, why, cost of, versus, DIY, tutorial, what is. Real people with a real problem who are not hiring today. Some of them will hire later, which is a remarketing question rather than a search question, though remarketing behaves differently when traffic is low and it is not a cure for a browsing-heavy account.

Wrong. Jobs, salary, parts, supply, free, a service you do not offer, a brand you do not carry, a city you do not serve. Cut without hesitation.

Ambiguous. You genuinely cannot tell. Leave them, watch them for another month, and revisit. If they have spent real money with nothing to show after two cycles, cut them.

The discipline is doing this in one pass without stopping to optimize bids. Read, classify, act. Bid changes are a different session.

What does the arithmetic look like on two real terms?

This is where the report stops being an audit and starts being a plan. Take two terms out of the same campaign.

Term A, a buying term. Spent $600 last month. At LocaliQ's 2026 average cost per click of $8.33 for Home and Home Improvement, that is 72 clicks (derived). It produced 8 leads, so its own conversion rate is 11.1%derived and its cost per lead is $75 derived rather than a category average.

Term B, a browsing term. Spent $420. That is 50 clicks (derived). It produced 1 lead, at a cost per lead of $420 derived.

Now look at what the summary screen shows you. Combined, those two terms spent $1,020 and produced 9 leads, a blended cost per lead of $113 derived. That is close enough to LocaliQ's published $90.92 cost per lead for the category that a monthly report would call it normal. The blend is hiding one term that pays and one term that does not.

Now redeploy. Kill term B and move its $420 to term A. That buys 50 more clicks (derived) at term A's own conversion rate of 11.1%derived, which is 5 or 6 more leads (derived). The same $1,020 now produces about 14 leads instead of 9, a lift of roughly 51%derived, with no change to the budget, the bids, the ads or the page.

Those LocaliQ figures are medians across a campaign sample, LocaliQ sells advertising management, and the 2026 edition does not publish its sample size. The 2025 edition disclosed 16,446 US campaigns. Use them as placeholders and substitute your own cost per click, which is in the same report you are already reading.

Run this on your top five terms by cost. Two of them usually carry the account and one of them is usually quietly eating it.

What do I do with what I find?

Three actions, in this order.

Add the wrong terms as negative keywords. Phrase match for multi-word ideas, broad for single junk words. Add the singular and the plural of each, because Google's documentation states that negative keywords do not match close variants: excluding "flowers" will not stop "flower." This is the immediate saving and you will see it in the next week's spend.

Promote the buying terms. If a term converts and it is buried inside a loose ad group, pull it out. Give it its own group with an ad that repeats the exact words the searcher used and a landing page that answers exactly that request. Matching those three is most of what people mean by relevance, and Google's own Quality Score documentation describes exactly those three components: expected click-through rate, ad relevance and landing page experience.

Then look at the terms that convert to leads and check whether those leads become customers. Cheap leads from vague searches can look like a success in the account and lose you money in the business, because cost per lead and cost per customer move independently. The report tells you what was cheap. Your books tell you what was profitable.

Where does this report mislead me?

In four specific ways, and knowing them keeps you from cutting something that was working.

Conversion lag. A conversion is credited to the date of the click, not the date it happened. A term you judge dead on the third of the month may still be collecting. Google notes this directly as a reason the search terms report and search terms insights can disagree.

Theme labels are stale by design. Google generates the labels in search terms insights from the last 56 days of data, which means they can include terms that are no longer serving or have already been excluded. Do not read a theme as a live description of your account.

The invisible share. Because low-volume queries are omitted, you are auditing the visible portion of your spend and treating conclusions from it as if they covered all of it. They do not.

Small numbers. A term with 40 clicks and no conversions has not proved anything. At an 8.05% category conversion rate you would expect about 3 leads from 40 clicks (derived), and getting zero is an ordinary outcome rather than evidence. Cut on cost and obvious intent, not on a thin conversion count.

The honest summary is that this report is very good at showing you searches you should never have bought and only moderately good at telling you which of the remaining ones deserve more money. Use it for the first job with confidence and for the second with patience.

What does the report tell me about my website?

More than most site audits do, if you read it in the right direction.

Every high-cost term with clicks and no conversions is a page failing a specific question. The searcher told you what they wanted. Your page did not answer it fast enough. That is a content problem with an exact brief attached.

If several expensive terms name a service you offer but never wrote a page about, you found your next page. If several name a neighborhood, you found your next geography. The report writes your website roadmap for free.

For service businesses the failures cluster in the same place. The visitor lands, cannot immediately confirm you do their specific job in their specific area, and leaves. What a contractor website needs above the fold is the fix for most of that population.

When the terms look right and the clicks still do not convert, the leak is usually the form. You can find where people quit without buying anything, and form abandonment analysis without expensive tools covers how.

And if the buying terms are dominated by urgent, immediate language, your format may be wrong entirely. People in a hurry want a phone, not a form, which is why call-only campaigns exist for service businesses and why sending that traffic to a contact page wastes it.

What to do this week

Block twenty minutes. Open the search terms report, last thirty days, sorted by cost. Read the top fifty terms and write each one into one of the four buckets on paper.

Add every wrong term as a negative before you close the tab, singular and plural. If you leave it for later, it does not happen.

Then run the two-term arithmetic above on your highest-cost buying term and your highest-cost browsing term. Write both cost-per-lead figures on the same line. That comparison, not the account average, is what tells you where next month's money goes.

Last, open the landing page for your best buying term on your phone. If it does not answer that exact search in the first screen, you have found next week's work.

Put a recurring monthly reminder on the calendar. Weekly for the first month of any new campaign.

Questions about what your account is actually buying?

Email me at eric@seod.com with your top fifty search terms by cost from the last thirty days. Export or screenshot, either works. I will mark each one as buying, browsing, or wrong, and send it back with the negatives I would add first and the two terms I would move money between.

It takes me about fifteen minutes and you keep the list whether or not we ever work together. If the account is already tight, I will tell you that and save us both the meeting.

Or keep reading more on paid search for local businesses.