Search terms report

The search terms report lists what people actually typed into Google before your advert appeared. Your keywords are what you asked for. The report is the record of what those requests were interpreted as, and the distance between the two is where most wasted spend in a search account lives.

In short

• Keywords are a request. Search terms are what the request was read as.
• The report is deliberately incomplete, so the totals will not reconcile.
• Read it for patterns and block the pattern, not the individual search.
• The searches you want more of are in there too, and they are usually missed.

A keyword and a search term are not the same object

A keyword is an instruction you gave the platform. A search term is a real thing a real person typed. Between them sits match type, which decides how loosely the instruction may be read.

Nothing about that arrangement is hidden or unfair. It is how the system is designed to work, and it is the reason a keyword can be sensible and the resulting searches useless. A plumbing company bidding on a repair phrase can be shown to somebody looking for parts, somebody looking for a job, somebody writing a school assignment and somebody in a city the business does not serve. Each of those is a defensible interpretation of the phrase and none of them is a customer.

The report is the only place that distinction becomes visible. Without it you are reading performance figures for keywords and guessing at the searches underneath them.

What the report shows you

Each row is a search, with the advert performance that followed it.

• The term itself. What was typed, in the words the person used.
• The keyword it was matched to. Which of your instructions produced it.
• Impressions, clicks and cost. What the search consumed.
• Conversions. What it produced, where you are tracking that properly.
• Match type. How loose the connection was, which tells you where to tighten.

The last column is the one people skim past. Sorting by it groups the loose matches together, and loose matches are where the surprises are concentrated.

View the report at ad group level whenever you can. The account view tells you that money went somewhere odd. The ad group view tells you which group let it through, which is the difference between knowing there is a leak and knowing where the pipe is. It also makes the repair obvious, because a negative added at group level leaves the campaigns that were behaving alone.

What it does not show you

The report is a large sample rather than a complete record, and knowing that prevents a lot of confusion.

Searches that too few people made are withheld for privacy reasons, so the clicks and spend listed in the report total less than the campaign reports as a whole. The shortfall is bigger on small accounts and on accounts running loose match types, which is an uncomfortable combination: the accounts most likely to be attracting odd searches are the ones shown least of them. It also means you cannot audit your way to certainty. You can only remove the patterns you can see and keep looking.

Two further limits are worth holding in mind. Automated campaign types report themes rather than terms, so the same diagnosis is coarser there. And the report tells you what was searched, never why, so a term with no conversions may be a bad search or a good search meeting a bad landing page.

Timing is the third limit. Conversions are credited back to the day of the click and appear only once they happen, so a report pulled this morning understates the past week of it. Judging a term as worthless three days after it first appeared is how useful searches get blocked, and blocking is easy to do and easy to forget you did.

Reading it without losing a morning

Sort by cost, descending, and start at the top. The expensive rows are the ones worth your attention and there are rarely many of them.

• Look for repeated words, not repeated searches. Four different searches sharing one word are one problem, not four.
• Watch for the free, cheap and salary group. Words that signal somebody with no intention of buying appear constantly.
• Check the geography. Business names and place names from cities you do not serve are among the most common leaks.
• Separate the wrong service from the wrong person. Somebody wanting a service you do not offer may still be worth a different page. Somebody looking for employment is never worth anything.
• Mark the terms you want more of. These become keywords, not negatives.

Fifteen minutes a week on the top twenty rows will do more than an hour spent scrolling the whole list, because the tail is long, cheap and mostly harmless.

Turning what you find into changes

Every row leads to one of four actions, and deciding which is the entire skill.

• Block it. Irrelevant and unlikely to improve. Add the pattern to your negative keywords at the right level, campaign or account.
• Promote it. Relevant, performing, and currently buried inside a broad keyword. Give it its own keyword so the advert can address it.
• Tighten around it. The keyword itself is fine but the match type is letting too much through. Move it to phrase or exact.
• Fix the destination. The search was right and nothing happened after the click, which is a landing page problem wearing a keyword disguise.

Make the changes in one sitting and write down what you did. The next report will contain the consequences, and without a note you will be reading it without knowing what changed.

One caution about scale. An account carrying a long negative list eventually starts blocking things nobody remembers deciding to block, and the symptom is a campaign whose impressions fall steadily for no visible reason. Read the list through once or twice a year and remove the entries added for a promotion that finished.

Words you will hear

• Search term. What a person typed. Sometimes called a query.
• Keyword. What you asked to appear for.
• Negative keyword. A word or phrase that prevents your advert showing.
• Search theme. The grouped description automated campaigns report in place of individual terms.
• Query mining. The habit of reading the report for new keywords rather than only for things to block.

The report is free, it takes minutes, and it is the only honest account of what your budget bought. An account nobody reads it for is an account being run on the assumption that the keywords and the searches are the same thing, and they never have been.