Search publishing guides

Compare Search Intent Signals Across Eight Keywords

·

A short query can imply several goals or provide too little evidence to choose one. Review the matched phrases before deciding whether to write a guide, a comparison or a location page.

Search Intent Classifier for Keywords

Check the eight keywords with Global selected

Choose Load worked example and Classify keyword intent. Global shows eight keywords with one mixed and two unknown. The line 6 pottery query gets three local points for near me and two commercial points for best, so it is mixed.

Open keyword-intent.json and inspect line, signals and scores. Those fields show which words contributed to the label. Strength describes the pattern of rules that matched.

Check the eight keywords with Global selected
Keyword cueExample labelRule points or coverage
how toInformational3
loginNavigational4
best and reviewsCommercial4
buyTransactional3
near meLocal3
best and near meMixedCommercial 2 and local 3
postcodeUnknownNo Global match
seed packetUnknownNo matching rule

Open Search Intent Classifier and load the eight keywords

Switch to UK vocabulary and flag rows for review

Choose United Kingdom and enable Strict mode, then classify again. Postcode adds three local points on line 7. The result now shows one mixed, one unknown and two flagged for review. The split pottery query and seed packet are the two flagged lines.

Clear Use question signals to test the role of how to. It becomes unknown on line 1. Clearing Use modifier signals drops the regional phrases along with the general modifier matches.

Open Search Intent Classifier and load the eight keywords

Check the reader goal before commissioning a page

Use the labels to decide what to look at next. The line 6 query may benefit from a local class page with enough detail for a reader to compare options. A tool label alone cannot establish that page’s merits or actual search audience.

Review the task a reader wants to complete and check relevant search evidence when available. The scores come from fixed English wording rules and are not probability estimates.

Google Search Central, evaluating people-first content

Common questions

What does a strong label mean?

At least three points support one label, with a gap of at least two points over the next score.

What happens to weak and mixed labels in strict mode?

They keep their labels and receive a separate review flag, as do terms with no matched rules.

Sources and example

Put it into practice.

Related guides