YouTube Keyword Tool
Discover what people are searching on YouTube. Get 20+ keyword suggestions from any seed term.
Discover what people are searching on YouTube. Get 20+ keyword suggestions from any seed term.
Start broad — 'drone review', 'react tutorial', 'sourdough'. The tool returns YouTube's own autocomplete suggestions for that seed.
US, UK, India, Brazil, Germany, Japan — autocomplete suggestions vary noticeably by region. Pick the market your channel targets.
These are the highest-confidence variants YouTube has on your seed.
We run 26 parallel queries (seed + each letter A-Z) and de-duplicate. This surfaces the long-tail keywords most creators miss.
We bucket suggestions by intent: Questions (what/why/how), Comparisons (vs / or), Best & top, Tutorials, and Other. Each group is a different angle for your content plan.
Field-tested principles for getting more out of this part of the workflow.
Read the related guide
Where keyword research sits in the bigger picture — and why YouTube autocomplete beats most paid keyword tools for early-stage research.
12 min read
YouTube's autocomplete is one of the most under-used SEO signals on the platform. Every suggestion you see when typing is backed by real search data — that's what makes it valuable. Most creators only see the first 10 hints YouTube shows them. This tool surfaces those, and lets you expand to 100+ long-tail variants in one click.
Suggestions are fetched server-side from the same public endpoint YouTube's own search bar uses, with a region/language pair you choose. Nothing is stored beyond a per-seed cache so repeat lookups don't re-hit YouTube.
The same autocomplete YouTube shows when you start typing in its search bar. We hit the public suggest endpoint server-side, so you don't have to type letter-by-letter — and you can pivot the region without leaving the tab.
It runs 26 parallel queries — your seed + ' a', ' b', ' c', through ' z' — to surface the long-tail variants YouTube has data on. So 'drone review' becomes 'drone review at night', 'drone review beginners', 'drone review camera', and so on. Most seeds produce 80-150 unique variants after de-duplication.
YouTube doesn't expose volume in its public suggest API. For volume estimates you need a paid keyword tool that scrapes SERPs or uses clickstream data (Ahrefs, Semrush, Keywordtool.io). This tool is for discovery — what are people actually typing — not for volume sizing.
A lot. US suggestions, UK suggestions, and Indian suggestions for the same seed are often noticeably different — different spellings, different popular brands, different angles. If your channel targets a specific country, pick that region.
Pattern matching on the suggestion text. Anything containing 'what / why / how' goes to Questions, 'vs / or / alternative' to Comparisons, 'best / top / free' to Best & top, 'tutorial / guide' to Tutorials. Everything else lands in Other variants. Use Flat view if you want them in YouTube's original order.
No. The autocomplete suggest endpoint is a public, undocumented-but-stable interface that every browser and dozens of SEO tools use. We hit it the same way your browser would.
Suggest requests are cheap (a single request per seed for base mode, or 27 parallel for expand). 50 lookups/day is more than any single creator needs in a session. We rely on light rate limiting + 6-hour cache to keep YouTube happy.
Pair this with the rest of the SEO toolkit.