GSC Regex for Question and Informational Keywords

If you just need the regex, here it is. Paste this into the GSC query filter (set to Custom (regex), match type: Matches regex):

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

That isolates every query in your GSC data that begins with a question word — the cleanest signal of informational, question-based search intent. These are the queries that map to "People Also Ask" boxes, featured snippets, and AI Overview citations. They are also the highest-leverage source of new blog post ideas and content gap fills.

How to Apply It in Google Search Console

1. Open GSC → Performance → Search Results
2. Click "+ New" in the filter bar at the top
3. Select "Query..."
4. Change the dropdown from "Queries containing" to "Custom (regex)"
5. Make sure "Matches regex" is selected (not "Doesn't match regex")
6. Paste the regex above
7. Click Apply

The filter applies across whatever date range you have selected. For meaningful analysis, set the date range to at least 90 days — short windows produce too much noise on question queries because individual long-tail searches happen infrequently.

What the Regex Actually Does

The pattern ^(who|what|where|when|why|how|should|can|is|are|will|does|do|which)\b reads as:

• ^ — start of the query string
• (who|what|...) — match any of the listed question words
• \b — word boundary, ensuring "what" matches "what is" but not "whatsapp"

The order of words doesn't matter for matching, but I've put the most common question starters first for readability. Add or remove words to suit your content — for instance, if you sell B2B software, "how" and "what" will be your highest-volume question starters; "where" and "when" may be near-zero.

Useful Variations

Questions That Don't Start With a Question Word

Plenty of question queries don't begin with the canonical question words. "Best way to track keywords" is informational intent without a leading "how." To catch these:

^(who|what|where|when|why|how|should|can|is|are|will|does|do|which|best way|tips for|guide to)

Question Words Anywhere in the Query

Removing the ^ anchor catches questions where the question word appears further in. This produces more matches but also more noise:

\b(who|what|where|when|why|how|should|can)\b

Comparison-Style Questions Only

For evaluation-stage queries specifically (more commercial than pure informational):

\b(vs|versus|or|compared to|difference between|better than)\b

What to Do With the Filtered List

Once the regex is applied, sort the resulting queries by impressions descending. The top of that list is your content opportunity map:

• High impressions, no current ranking page: You're being shown for the query because Google considers your domain relevant, but you don't have a dedicated piece of content. These are the cleanest content gaps in the list.
• High impressions, page ranking 11+: You have a page but it's not competitive. Optimize for featured snippet capture — direct answer in the first 40-60 words, structured list or paragraph format.
• High impressions, page ranking 1-3 with low CTR: AI Overview suppression is likely. The query is informational, so AI is answering it; this is structural and worth knowing about even if not directly fixable.

The GSC Limitation: It Doesn't Save Your Filter

This is where the manual approach starts to hurt. Every time you log out of GSC and come back, that regex needs to be re-entered. There is no way to save a query filter as a preset in GSC's native interface. If you're tracking question keywords as an ongoing cohort — month after month — you're pasting the same regex into the filter every time you want to check progress.

It also doesn't persist across the 16-month data retention limit. The question keywords that were sending you impressions in 2023 are gone now. The trend that matters most for informational content strategy — "is question-intent traffic growing or shrinking on this site over multiple years?" — can't be answered from native GSC alone.

The Done-For-You Version

The underlying mechanic: your GSC data lives in BigQuery (via the official export), Keyword History queries it, and your tags are stored against that queryable archive rather than against the GSC session. The regex above can be the basis of a permanent cohort that you check on monthly rather than rebuild monthly.

One More Pattern Worth Saving

If you're optimizing for featured snippets specifically, the questions Google most often promotes to snippet position start with very specific phrases:

^(how to|what is|why does|how does|how do you|how can I)\b

That subset — about 6-8 patterns — accounts for the majority of featured snippet opportunities I've seen produce results on the sites I've worked on. Worth tagging as its own cohort if featured snippet capture is a strategic priority.

Question-intent queries are a content compass. The regex finds them; the question is what you do with the list. Most sites have 50-200 high-impression question queries they haven't built content around — and finding those takes minutes once the filter is applied.