Multi-Dimension Query Classifier: Rule-Based Colab Notebook and Google Sheets Template

Most keyword tools give each query one search intent label and stop there. A single query says much more than that: whether the searcher is a beginner, whether they want a video or a template, whether location matters, whether they are trying to fix something. This free rule-based classifier reads every keyword in your export and adds a column for each of those dimensions, using regex rules you can open and change.

What it labels

• Query length in words
• Main intent: informational, commercial, transactional, navigational (more than one if several match)
• Query type: Do, Know or Go
• Geo relevance, device relevance and an explicit content format (video, infographic, template, document, webinar, podcast, eBook)
• Knowledge level (beginner, intermediate, expert) and user persona (student, professional, casual user, budget-conscious, affluent and more)
• Implicit problem (“how to fix”, “won’t work”) and context such as urgency or mood

Who it’s for

SEOs and content strategists who want richer intent labels on a keyword set without paying for another tool, and without needing to write Python. If you can upload a CSV, you can run it.

How to use it

1. Open the notebook in Google Colab and save a copy to your own Drive.
2. Export your keywords from Semrush, Ahrefs or Search Console as a CSV with a Keyword column.
3. Run the first code cell. It asks you to upload the CSV, then applies each labelling function to the Keyword column and shows the first rows.
4. Run the last cell to download updated_file.csv with all the new columns.
5. Edit the regex lists for your industry (your brand names, your product words) and run it again.

Prefer spreadsheets? The Google Sheets version does the same with formulas. Make a copy: it has a demo tab of about 3,950 “first time dog owner” keywords, a tab that shows each formula, and the notebook’s output for comparison.

For intent alone, there is also a refreshed rule-based classifier on GitHub that matches whole words, keeps every matching intent (multi-label), flags easy keywords (difficulty under 65) and money keywords (CPC of $0.01 or more), and prints the same rules as a Google Sheets IFS(REGEXMATCH()) formula and a Data Studio calculated field.

What you get

Your keyword export back as a CSV with ten extra columns you can pivot, filter and chart, plus rules you can audit line by line.

Limits

• Rules only see the words in the query. A keyword like “chatbot benefits” says nothing explicit about intent, so it stays unclassified.
• Navigational intent depends on brand names, so add yours and your competitors’.
• The lists are English-only and some Sheet formulas match parts of words; the GitHub version fixes this with word boundaries.

Read the full method comparison in Search-Intent Classification with Machine Learning, and go deeper in the Semantic ML-enabled Keyword Research course.