Glossary
Long-tail query
A long, specific search that few people type in that exact form: a full question or a description of a situation rather than a product name.
A long-tail query is a search specific enough that its exact wording is rare: “I have eight foot ceilings in my bedroom and a regular fan feels too low, what flush mount fans actually work well?” rather than “ceiling fan”. The name comes from a chart of searches by frequency, where a short run of popular head terms is followed by a long tail of searches made rarely or once. The tail is big. In 2019 Google said that 15 percent of the searches it sees each day are ones it had not seen before; that figure is now several years old. In May 2025 it said people were bringing “more complex, longer and multimodal questions” as AI Overviews spread.
For advertisers the problem is matching. A keyword list rarely holds a 20-word scenario, so whether an ad can appear depends largely on broad match or keywordless matching such as AI Max for Search, and the advertiser cannot see the exact searches in advance.
We ran 47 long scenarios, with a median length of 21 words (Oct 10, 2026, US searches). An AI Overview appeared on 45 of the 47 searches. They ran only in our answer capture, a second set of searches that recorded the AI answer and the sources it linked, but picked up ads on few of them (29 of 379), so how often such questions carry ads was not measured. The one long scenario with ads recorded was a question about Kubernetes secrets kept in Git by a team of 12 DevOps engineers: 4 text ads, all below the AI Overview, from Wiz and CrowdStrike, each offering a guide or cheat sheet rather than a product. The long questions guide sets out what this data can and cannot show.
Talk with us
We look at the questions your buyers ask around AI Overviews, who buys ads or gets cited there, and where your brand is missing. 30 minutes, free, no slides.
Get my free AI search check →Free
We re-run the searches and publish new ads. One short email when it lands.