Google & YouTube Autocomplete Keyword Miner
Pricing
Pay per usage
Google & YouTube Autocomplete Keyword Miner
Expands seed topics into the real searches people type on Google and YouTube, with question detection, relevance scores and any language/country. Built for content and SEO planning.
Pricing
Pay per usage
Rating
0.0
(0)
Developer
Ricardo Faya
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
a day ago
Last modified
Share
Turn a handful of seed topics into the searches people actually type on Google and YouTube — in any language and country — with question detection built in.
Why this exists
We ran a content site with hundreds of pages and almost no non-brand search traffic. Search Console had nothing to teach: you can't learn demand from a site nobody finds yet. The autocomplete of Google and YouTube can — it's public, free, and ordered by what people type. One run of this approach produced 8,276 real searches and 732 page ideas for us.
What you get
For every seed, the Actor queries:
- the seed itself,
- question words in front of it (
como,o que é,vale a pena… /how to,what is… /cómo,qué es…), - a–z after it, for the long tail,
on Google Search and/or YouTube, and returns one deduplicated row per keyword:
{"keyword": "como usar inteligência artificial no trabalho","source": "youtube","seed": "inteligência artificial","query": "como inteligência artificial","expansion": "question: como","position": 2,"relevance": null,"isQuestion": true,"wordCount": 6,"language": "pt-BR","country": "BR"}
relevanceis Google's own suggestion weight (Google source only). It is an order, not a monthly search volume — use it to compare keywords, not to promise traffic.- Turn on Only return questions to get a list of ready-made video and article titles.
Tips
- YouTube-only questions are gold for video creators: filter
source = youtubeandisQuestion = true. - For more than a few hundred queries, enable Apify Proxy. Google blocks datacenter IPs quietly (HTTP 403 with an empty body).
- The run protects itself: after 12 blocked queries in a row it stops and keeps everything collected so far, instead of spending hours retrying. A blocked query is never recorded as "no suggestions".
Output
- Dataset: one row per unique keyword per source.
- Key-value store
OUTPUT: run summary (queries run, keywords, questions, blocked responses, whether it stopped early).
Limits
Autocomplete shows what is typed, not how often. Pair it with a volume tool if you need absolute numbers.