When to build a keyword research target audience from search data

What to take away

Why does search data alone not give you a target audience?

A keyword tells you the question. It does not tell you who asked, what they already know, or what happens next. Two people searching the same phrase can sit at opposite ends of a decision.

That gap has widened. A segment that is only a keyword gives a writer nothing to plan against, because there is no described person behind the search.

Start with what you already hold: sales calls, support tickets, site search. Use keyword demand to fill the holes, not to replace that picture.

Write one sentence per source about who was speaking and what they wanted. Those sentences become the first evidence column in your segment notes.

How do you collect demand for a target audience?

Pull every query that names a need, a problem or a job to be done, then group them by what the searcher wants. Ignore volume for now; the aim is coverage.

Use Google Keyword Planner to explore keyword demand and Google Search Console to see queries bringing people to your pages. Compare these with your site's search terms and competitor pages; record each source and date, and use comparable demand measures when sizing needs.

If the list is thin, ask customers or prospects what they were trying to do, using neutral questions that do not suggest an answer. Record when and where responses came from, and use them to test a search-based need rather than substitute for search data.

Tag each query with the need it names and the stage it sits at as you collect. Untagged queries drift into a list nobody can plan from.

Which segments belong in the audience?

Sort the grouped queries into needs: learn, compare, buy, fix, maintain. Each need becomes a candidate segment. Move it into a content plan when related queries repeat the same need, stage and outcome, and at least one independent source, such as a sales call, support ticket or site-search record, supports that pattern. If it appears only in keyword data, mark it provisional and gather corroboration first.

Size each need using demand figures from the same keyword source and on the same basis. Treat the result as relative search demand, not a count of people: query volume alone cannot tell you how many distinct people are in a segment.

For competitive work, compare the pages ranking for each query with your own pages. Note what need each competitor page serves, but do not treat a content gap alone as proof that the need belongs in your audience.

Keep the count small. Four to six segments is enough for most England-focused programmes.

Give every segment a plain working name, such as a sole trader comparing software before a renewal. Write it in the words your sales team already uses.

How do you check the segments against real behaviour?

Test each segment against independent evidence before you commit. Record the source name, date and geography, and check that it supports the claim you are making; evidence about channel habits, for example, supports a channel claim, not a need by itself.

If a segment's channel habits contradict your plan, change the plan rather than the source.

Then set a review date. A segment built in September 2026 should be revisited as soon as your search data or your market moves.

Record the source name, date and geography beside each segment so a colleague can repeat the check and understand what the evidence does, and does not, support.

What does a finished segment look like?

The left column is a label; the right is a brief you can hand to a writer without further explanation. For this example, the brief could lead to a software comparison page for the comparison need and a renewal checklist for the renewal need, if query evidence supports both.

For a related audience research method, see the audience research guide for England.

Common questions

How many keywords do I need per segment?

There is no fixed number. A segment is ready when further keywords stop changing the need, the stage or the outcome.

Can I build a target audience without survey data?

Yes, but the segment will be weaker. Search data shows demand, rarely motivation. Label it provisional, set an earlier review date and note the gap in the segment file.

Should segments be based on job title or need?

Need. Job titles shift with restructuring, while the underlying problem stays stable. Use a title as a descriptor inside the segment, not as the segment itself.

How often should I refresh the segments?

At least twice a year, and sooner if your market, product range or main channels change. Tie each review to a dated source.