Keyword Research Tool – Google, YouTube & Amazon Autocomplete Scraper

Turn any seed keyword into hundreds of long-tail keyword ideas and questions using the public search-suggestion (autocomplete) boxes of Google, YouTube, Amazon, Bing, DuckDuckGo, Yahoo and eBay. No login, no API key, no browser, no proxy needed. Pay only per keyword returned: $0.60 per 1,000 keywords.

Autocomplete is where real people's searches show up first, which makes it the cheapest source of keyword ideas for SEO, YouTube SEO, Amazon/eBay listing keywords, content calendars and PPC.

What it does

For every seed keyword the Actor:

1. Queries the autocomplete endpoint of each source you select (Google, YouTube, Amazon, Bing, DuckDuckGo, Yahoo, eBay).
2. Optionally expands the seed – seed a … seed z, question words (how seed, what seed, can seed…), prepositions/comparisons (seed for, seed vs, seed near…) and digits – so you get far more than the ~10 suggestions a single query returns.
3. Merges and de-duplicates the results across sources and returns one row per unique keyword with the sources it appeared in, its rank in each source, a word count and a simple rule-based intent label.

Keywords that show up in several suggestion boxes (sourceCount 3+) are the ones multiple engines consider popular, so they are usually the best candidates to target first.

Use cases

• SEO keyword research – long-tail ideas and a free alternative to paid keyword-suggestion tools (this Actor does not provide search volume, see Limitations).
• YouTube SEO – find what people type into YouTube search for your topic.
• Amazon / eBay listing keywords – see what shoppers search for in each marketplace and country.
• Content planning and FAQ pages – turn on question expansion to collect "how / what / why / can" questions.
• PPC / ad-group building – bulk long-tail phrases grouped by seed.
• Programmatic use – call it via API, schedule it, or plug it into Make, Zapier, n8n or Google Sheets.

Input

Field	Description
seedKeywords	List of seed keywords or phrases.
sources	google, youtube, amazon, bing, duckduckgo, yahoo, ebay (default: google + youtube).
country	us, gb, ca, au, de, fr, es, it, br, mx, jp (default us).
language	Optional language override for Google/YouTube (e.g. en, de).
expandAlphabet	A–Z expansion (default on).
expandQuestions	Question-word expansion (default off).
expandPrepositions	Preposition / comparison expansion (default off).
expandNumbers	0–9 expansion (default off).
maxSuggestionsPerSeed	Cap on unique keywords (and cost) per seed, default 200.
maxConcurrency	Parallel requests, default 5.
proxyConfiguration	Optional, normally not needed.

Example input:

{
  "seedKeywords": ["keyword research"],
  "sources": ["google", "youtube"],
  "country": "us",
  "expandAlphabet": true,
  "expandQuestions": true,
  "maxSuggestionsPerSeed": 100
}

Output

One dataset item per unique keyword (real example from a test run):

{
  "seedKeyword": "keyword research",
  "keyword": "keyword research free",
  "sources": ["google", "youtube", "bing", "duckduckgo"],
  "sourceCount": 4,
  "ranks": {"google": 1, "youtube": 7, "bing": 1, "duckduckgo": 1},
  "bestRank": 1,
  "foundVia": "keyword research",
  "expansionType": "seed",
  "wordCount": 3,
  "charCount": 21,
  "intent": "informational",
  "startsWithSeed": true,
  "country": "us",
  "language": "en",
  "scrapedAt": "2026-10-25T14:53:49.850277+00:00"
}

A question keyword found through question expansion:

{
  "seedKeyword": "keyword research",
  "keyword": "what is keyword research in google ads",
  "sources": ["google", "youtube"],
  "sourceCount": 2,
  "ranks": {"google": 7, "youtube": 2},
  "foundVia": "what keyword research",
  "expansionType": "question:what",
  "intent": "question"
}

• ranks – position (1 = first suggestion) in each source's suggestion list.
• foundVia / expansionType – the first query that produced the keyword (seed, alphabet:x, question:how, preposition:for, number:3).
• intent – simple rule-based label from the words in the keyword: question, comparison, transactional, commercial or informational. It is a quick filter, not a machine-learning classification, and it only understands English words.
• startsWithSeed – whether the keyword begins with your seed (handy to drop loosely related suggestions).
• Rows are sorted by sourceCount (high to low), then bestRank. If a seed returns nothing, you get one type: "error" item for it (not charged).

Download as JSON, CSV, Excel, HTML or XML, or read it through the Apify API.

Pricing

Pay-per-event, no subscription:

• $0.0006 per keyword returned (= $0.60 per 1,000 keywords).
• Actor start: $0.00005 per run (Apify's standard start event).

Examples: 1 seed × A–Z expansion × Google + YouTube with a 200-keyword cap ≈ $0.12. 10 seeds × 200 keywords ≈ $1.20. Use maxSuggestionsPerSeed and the run's "maximum cost per run" setting to cap spend. Platform usage for a typical run is a fraction of a cent.

FAQ

Does it give search volume, CPC or keyword difficulty? No. Autocomplete endpoints do not expose volume. This Actor finds which keywords people search; use it to build the list, then check volume in a tool of your choice.

Where does the data come from? The public suggestion endpoints that power the search boxes of each site. No login, cookies, or accounts are used and no personal data is collected – suggestions are anonymous, aggregated phrases.

Why do I get fewer keywords than expected? Each query returns about 5–10 suggestions per source. A–Z expansion over a niche seed typically yields 100–400 unique keywords per seed; very narrow seeds simply have fewer suggestions. Turn on question/preposition/number expansion for more.

Are results personalised? Suggestions are requested without cookies or a signed-in account from a data-centre IP, so they reflect general (country-level) suggestions, but each engine may still vary results slightly by time and location of the request.

Which countries and languages are supported? The 11 countries listed above. Yahoo suggestions are US-only and eBay is available for US, UK, CA, AU, DE, FR, IT, ES (unsupported source/country pairs are skipped with a log warning).

Can I run it from the API / schedule it? Yes – it's a normal Apify Actor (API, schedules, webhooks, integrations).

Limitations

• No search volume / CPC / difficulty data (see above).
• Suggestion lists are short, so large results come from expansion, which multiplies the number of requests (27 requests per source per seed with A–Z, 41 with A–Z + questions).
• The endpoints are unofficial public endpoints of third-party sites and may change or rate-limit heavy use; the Actor retries with back-off, and failed queries are counted in the log. For very large batches you can enable the optional Apify Proxy.
• Intent labels are keyword-rule based and English only.
• Amazon marketplaces: US, UK, CA, AU, DE, FR, ES, IT, BR, MX, JP. India is not supported.

Legal

The Actor reads only publicly available search-suggestion data and stores no personal data. You are responsible for using the data in line with the terms of the respective sites and applicable law.