Search Console Regex Builder

This Search Console regex builder writes the regular expressions you need to filter the Performance report: question queries, non-brand queries, long-tail queries, search intent, folders, URL parameters, and more. Turn on the patterns you want, choose whether each one keeps or removes rows, and copy a regex that’s checked for RE2 errors. Then drop in your Search Console export to see exactly which queries or pages the filter keeps and their share of your clicks. The same filter is ready for Looker Studio, the Search Console API, and BigQuery.

How to use the Search Console regex builder

• Choose Queries or Pages. Each has its own set of ready-made patterns.
• Turn patterns on. Set each one to Include to keep matching rows or Exclude to remove them, and adjust its options. For example, include Questions and exclude your brand name.
• Copy the regex. The Matches regex and Doesn’t match regex boxes update as you go, with a check that RE2 will accept them.
• Test it. Drop in the .zip, Excel, or CSV file from the Performance report’s Export button, or paste a list, to see the rows it keeps and leaves out.
• Use it anywhere. Follow the steps for Search Console, or copy the Looker Studio formula, API request, or BigQuery SQL.

How Google Search Console regex works

Google’s guide to advanced filtering in the Performance report explains the rules. To filter by regex, add a filter for Query or Page, choose Custom (regex), and pick Matches regex or Doesn’t match regex. Search Console uses RE2 syntax, and three details matter:

• Partial match. The pattern can match anywhere in the query or URL unless you use ^ for the start or $ for the end.
• Letter case is ignored. Search Console regex isn’t case-sensitive by default. Start the regex with (?-i) to match case exactly.
• Several values. Separate them with a vertical bar, as in coffee|espresso.

RE2 guarantees fast, predictable matching, but it leaves out some features, such as lookaheads and backreferences. That’s why a GSC regex copied from a JavaScript or PCRE tutorial is often rejected.

17 ready-made Search Console regex filters

Filter	What it finds
Questions	Queries with who, what, how, why, can, does, and other question words you choose, plus queries with a question mark
Brand names	Your brand and product names, with spacing and hyphen variants and misspellings. Exclude it for non-brand queries.
Query length	Long-tail queries of 5 or more words, or short-head queries of 1 or 2
Search intent	Comparing (best, vs, price), ready-to-act (buy, order), learning (how to, guide), and login or contact queries
Local searches	near me, open now, and the places you serve
Topic words	Queries with any or all of your words, in any order
Exact queries, years and months	A list of target keywords, and queries with 2026 or a month name
Folders, URL list, subdomains	Pages in /blog/ or other folders, a list of URLs, or one subdomain in a Domain property
Parameters, file types, depth	URLs with ?utm_ or other parameters, PDFs and images, and pages by number of folders
Language folders, archives, URL words	/fr/ and /en-gb/ folders, pagination, tag, category, author, and feed URLs, and words in the slug

You can also write your own regex. It’s checked the same way and combined with the others.

Regex for question queries

With the default question words, the Search Console regex for question queries is:

\b(should|where|which|what|when|does|will|how|why|who|can|are|do|is)\b|\?

The \b marks a word boundary, so “how” matches “how to descale” but not “showroom”. Question words in other languages, such as cómo or warum, can be added too; words with accented letters use spaces as boundaries, because RE2’s \b only recognizes ASCII letters. Question queries are a quick way to find content ideas and sections that need a direct answer.

Non-brand queries and branded vs non-branded traffic

To see non-brand queries, set Brand names to Exclude and add your brand, product names, and misspellings. Names of two or more words also match without spaces or with hyphens, so “tools instant” catches toolsinstant and tools-instant. Comparing branded vs non-branded queries shows whether growth comes from people who already know you or from new searches. The tester’s share of clicks gives you that split straight away.

Google anonymizes some rare queries to protect privacy. Those rows have no query text, so no query filter can include them.

Regex for long-tail queries

Query length counts words. For five or more words, the Search Console regex is ^\S+(\s+\S+){4,}$: a word, then four or more spaces and words, from the start of the query to the end. Long, specific queries often show what people really want to know.

Page filters for folders, parameters, and thin pages

A Search Console regex for pages matches the full URL, including https:// and the host. A folder filter for /blog/ becomes ^https?://[^/]+/blog(/|\?|#|$), which matches every page in the folder on any host but not /blogger/ or /shop/blog/. Use Archives to exclude pagination, tag, and category pages from a content report, or Parameters to find URLs with tracking parameters that Google shows in search.

Test your regex on real Search Console data

A Search Console regex that looks right can still miss queries or catch the wrong ones. Export the Performance report, drop the file into the tester, and see:

• How many rows the filter keeps, and their share of clicks and impressions
• CTR and impression-weighted average position for kept rows and for the rest
• Every kept row with the matching part highlighted, and every row left out

Search the rows, download the kept ones as CSV, and adjust the patterns until the split is right. To go further with the same export, try the Striking Distance Keyword Finder or the Traffic Drop Analyzer.

RE2 mistakes the builder catches

Before you copy a Google Search Console regex, the builder checks it for the mistakes that make Search Console reject it or quietly return the wrong rows:

• Unsupported syntax: lookaheads like (?!brand), lookbehinds, backreferences, possessive repeats, and \Z.
• Empty alternatives: a stray | at the start or end, or ||, matches every row, so the filter quietly keeps everything.
• Spaces next to |: brand | other only matches with the spaces.
• Wildcards from other tools: blog* means “blo” followed by any number of g’s. Use .* for “anything”.
• Unescaped dots: in example.com the dot matches any character. Write example\.com.
• Unbalanced brackets and repeat counts over 1,000, which RE2 rejects.

A pattern with an error is held back from the output, so what you copy works.

Use the same filter in Looker Studio, the API, and BigQuery

In Looker Studio, the builder writes a calculated field with REGEXP_CONTAINS. Looker Studio regex is case-sensitive, so the formula adds (?i), and an R"…" raw string keeps backslashes as they are. The field labels each row as your group or Other, ready to compare in a chart.

The Search Console API request uses the searchanalytics.query method with includingRegex and excludingRegex filters in one group, so both apply together. Filter values can be up to 4,096 characters, and a request returns up to 25,000 rows. For BigQuery, the SQL reads the bulk data export tables, leaves out anonymized queries, and calculates average position with Google’s formula. For GA4, use the GA4 AI Traffic Regex Generator.

Privacy

The Search Console regex builder runs in your browser. Your exports are read on your device and never uploaded or saved.