“How to clean a steel bottle” and “buy 500 ml steel bottle” share a product but ask for different help. Grouping them under one broad label can hide the difference between a care article and a product page.
The Keyword Clustering tool suggests topic groups and likely search intent from supplied query text. Use it to organize a list, then review whether the queries in each group could be served by the same kind of page.
Prepare a traceable keyword list
Paste CSV with the exact header id,keyword into Keyword CSV: id,keyword. Use stable, unique IDs so you can join the classifications back to your original research.
This illustrative list contains four queries:
id,keyword
K1,how to clean a steel bottle
K2,remove smells from a steel bottle
K3,best steel bottle for commuting
K4,buy 500 ml steel bottle
The CSV accepts up to 50 rows within 9,000 characters. Each keyword can contain up to 200 characters. Long inputs can need fewer rows to fit the result. Quote CSV cells that contain commas or line breaks.
Explain the subject without inventing data
Topic and audience context is optional and accepts 1,000 characters. For the example, use “Content planning for a shop selling steel drinking bottles.” That context clarifies the subject without claiming that a query is popular.
Select Group keywords and complete browser verification when prompted. Use query data you have permission to send to the site's configured AI providers; remove customer identifiers or private project details that are unnecessary for classification.
Inspect assignments before cluster totals
A reasonable illustrative grouping is:
| IDs | Group | Likely intent |
|---|---|---|
| K1, K2 | Bottle care | Informational |
| K3 | Commuting bottle comparison | Commercial |
| K4 | Purchase a 500 ml bottle | Transactional |
The first pair seeks instructions. The comparison query suggests a buying decision still being evaluated. The last explicitly asks to buy. These labels describe a reading of the words, rather than observed search behavior.
Read the Reason for each assignment. If a care query lands in the purchase group because both mention a bottle, move it in your planning notes or revise the list and request another grouping. Check that every original ID appears exactly once.
Mixed and uncertain are legitimate intent labels. Use mixed when a group truly combines tasks, and retain uncertainty when the wording does not support a confident classification. Do not force an ambiguous query into a sales category to make the list look tidy.
Use the groups as planning proposals
The example has two care rows, one comparison row and one purchase row: 2 + 1 + 1 = 4 supplied queries. That count does not establish that care has twice the search demand of purchasing.
The tool does not fetch search results or measure volume, difficulty, traffic or rankings. Before assigning pages, compare the proposed groups with your existing content and obtain any separate research your decision needs. One group may fit a section of an existing page; another may need a distinct page because the reader is trying to complete a different task.