Keyword Suggestions: Google, YouTube, Bing, Amazon Autocomplete
Pricing
from $0.55 / 1,000 suggestion returneds
Keyword Suggestions: Google, YouTube, Bing, Amazon Autocomplete
Keyword suggestions from Google, YouTube, Bing, Amazon and DuckDuckGo autocomplete. A-Z, question and preposition expansion, merged and deduped, per country and language. No proxy, no key.
Pricing
from $0.55 / 1,000 suggestion returneds
Rating
0.0
(0)
Developer
Detroit Web Agent - Open Data
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
3 days ago
Last modified
Features
Share
Keyword suggestions for SEO, content and product research: this Actor returns the autocomplete suggestions that Google, YouTube, Bing, Amazon and DuckDuckGo show as people type, expanded with A-Z, 0-9, question words and prepositions, merged across engines and deduped. One tiny JSON call per query, no proxy, no browser, no API key.
Type a seed such as "coffee maker" and the Actor asks each engine what people search next: "coffee maker with grinder", "coffee maker walmart", "how to clean a coffee maker". It repeats that for every expansion you pick, so one seed becomes hundreds of real long-tail keywords with the engine, the rank each engine gave it, the country and the language. Use it as a Google autocomplete scraper, a YouTube keyword tool, an Amazon search suggestion finder or a Bing keyword suggestion tool, all in one run.
Run in 30 seconds
- Open the Input tab. The prefill researches "coffee maker" on Google, YouTube and Bing with plain and A-Z expansion, capped at 300 rows.
- Click Start. It finishes in a few seconds.
- Open the Suggestions table in the Output tab, or export JSON, CSV or Excel.
- Replace the seed with your own topic, add engines and expansions, raise Max suggestions and run again.
What does this Actor do?
Autocomplete is the cheapest, freshest keyword signal there is: it shows what real people type right now, per country and per platform. This Actor calls each engine's public suggestion endpoint, which returns a small JSON list, so it needs no proxy and no browser. A run with five engines and 40 expansions is about 200 tiny calls and finishes in well under a minute.
| Engine | What it tells you | Country and language |
|---|---|---|
| General web search demand, with Google's relevance score | Region and interface language | |
| YouTube | Video topics and how-to demand | Region and interface language |
| Bing | Bing and Copilot search demand | Market, for example en-US or de-DE |
| Amazon | Product search terms shoppers use | Marketplace of the country (US, UK, DE, FR, IT, ES, CA, JP, IN, AU, MX, BR and more) |
| DuckDuckGo | Privacy search demand | Region and language |
| Google Shopping | Product phrases in Google Shopping | Region and language |
| Google Images | Visual search phrases | Region and language |
Expansion modes
| Mode | Queries sent per seed | Example |
|---|---|---|
| Plain | 1 | coffee maker |
| A-Z | 26 | coffee maker a, coffee maker b ... |
| 0-9 | 10 | coffee maker 1 ... |
| Question words | 14 | how coffee maker, what coffee maker, why coffee maker, can coffee maker ... |
| Prepositions | 13 | coffee maker for, coffee maker with, coffee maker vs, coffee maker near ... |
You can combine modes. Each query is sent to each engine you picked, so the number of calls is seeds x queries x engines.
Output fields
Every row has the same core fields, so it drops straight into a sheet or a database.
| Field | Meaning |
|---|---|
seed | The keyword you gave |
query | The exact text sent to the engine, for example coffee maker b |
expansion | Which mode produced the query: plain, alphabet, digits, questions, prepositions |
suggestion | The suggested keyword |
searchUrl | Link to that suggestion's search results on the engine, ready to open |
engine | Engine that returned it (separate rows mode) |
rank | Position in that engine's list, 1 is first |
relevance | Google's own relevance score when Google or YouTube provides one |
suggestionType | The engine's label for the suggestion, when it gives one |
engines, engineCount, bestRank | In merged mode: every engine that returned it, how many, and the best rank |
timesSeen | How many queries returned the same suggestion |
country, language | The region and language used |
wordCount, charCount | Length of the suggestion |
isQuestion | True for question style keywords such as "how to descale a coffee maker" |
Nullable fields (relevance, suggestionType) are null when the engine does not send them.
Merged mode and duplicates
- Remove duplicates is on by default: each suggestion is returned once per engine, with its best rank.
- Merge engines turns the dataset into one row per unique suggestion with an
engineslist. Keywords that two or more engines agree on are strong demand signals, andengineCountlets you sort by that agreement. - Turn duplicates off if you want to see every hit from every query.
Input
| Field | What it does |
|---|---|
seeds | One or more words or phrases, one per line. |
engines | google, youtube, bing, amazon, duckduckgo, googleShopping, googleImages. Default google, youtube, bing. |
expansions | plain, alphabet, digits, questions, prepositions. Default plain, alphabet, questions. |
country | Two-letter code such as us, gb, de, fr, jp. |
language | Language code such as en, de, fr, es, ja. |
mergeEngines | One row per unique suggestion with an engines list. |
removeDuplicates | Return each suggestion once. Default on. |
maxResults | Hard cap on rows for the whole run. Default 5000. |
concurrency | Parallel calls, 1 to 15. Default 6. |
Example input
{"seeds": ["coffee maker", "espresso machine"],"engines": ["google", "youtube", "bing", "amazon"],"expansions": ["plain", "alphabet", "questions", "prepositions"],"country": "us","language": "en","mergeEngines": true,"maxResults": 3000}
Output sample
{"seed": "coffee maker","query": "coffee maker","expansion": "plain","suggestion": "coffee maker with grinder","searchUrl": "https://www.google.com/search?q=coffee%20maker%20with%20grinder&hl=en&gl=us","country": "us","language": "en","wordCount": 4,"charCount": 25,"isQuestion": false,"timesSeen": 12,"engines": ["google", "youtube", "bing", "amazon", "duckduckgo"],"bestRank": 1,"engineCount": 5}
The values above are an illustration of merged mode. Your own run returns the real, current suggestions.
Use cases
- SEO keyword research: build a long-tail keyword list for a topic in minutes, with question keywords for FAQ and blog briefs.
- YouTube SEO: find the titles and phrases people type into YouTube, then compare them with Google.
- Amazon and ecommerce: collect the product terms shoppers use on each marketplace for listings, ads and backend keywords.
- Content planning: the question words mode gives "how", "what", "why" and "can" topics ready to turn into articles.
- Local and international SEO: repeat a seed across countries and languages to see how demand differs.
- Market research and ads: see what is trending around a brand, competitor or product name, and build negative and positive keyword lists.
- Data for AI agents: feed a seed into an agent or n8n flow and get a clean list of related queries back.
How it compares with a SERP scraper
This is not a search results scraper. It does not load result pages, so it never needs a proxy, a browser or a captcha solver, and it does not return URLs or ads. It returns only the suggestion lists, which is exactly what keyword research needs and is why a row costs a fraction of a cent.
Limits
- Autocomplete lists change over time and by region. Two runs on different days can differ. That is the point of the data.
- Each engine returns about 4 to 15 suggestions per query. Use A-Z, questions and prepositions to go deeper.
- Search volume and keyword difficulty are not included. Autocomplete shows what people type, not how many people type it.
- An engine can occasionally slow down or refuse a request. The Actor retries, and if an engine does not answer at all it is skipped and listed in the run
OUTPUTand warnings instead of failing the run. - Amazon suggestions use the marketplace matching your country code. Countries without an Amazon marketplace use the US one.
- Very large runs stop cleanly before the time limit and report in
OUTPUThow many queries were not asked.
Pricing and cost control
Pay per suggestion: $0.0009 per suggestion on the free plan, $0.000554 on the Gold plan (Bronze $0.000839, Silver $0.000715), plus a $0.00005 start fee per run. That is $0.90 per 1,000 suggestions on the free plan.
| Plan | Per suggestion | Per 1,000 |
|---|---|---|
| Free | $0.0009 | $0.90 |
| Bronze | $0.000839 | $0.84 |
| Silver | $0.000715 | $0.72 |
| Gold and above | $0.000554 | $0.55 |
You only pay for rows you receive. Use Max suggestions or Apify's Maximum cost per run to cap spend; the run stops cleanly at the limit. Duplicates are removed by default, so you are not charged for the same keyword twice.
Integrations and scheduling
Run it on a schedule to track how autocomplete for a brand or topic changes week by week, then send results to Google Sheets, Slack, email, Make, Zapier or n8n with the Apify integrations, or call it through the REST API and the Python and JavaScript clients.
Use with the API, n8n, Make, Zapier or AI agents
API (one call, returns the rows as JSON):
$curl -X POST "https://api.apify.com/v2/acts/transparent_meteorite~keyword-suggestions-autocomplete/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" -H "Content-Type: application/json" -d '{"seeds":["coffee maker"],"engines":["google","bing"],"expansions":["plain","alphabet"],"maxResults":200}'
Python
from apify_client import ApifyClientclient = ApifyClient("YOUR_APIFY_TOKEN")run = client.actor("transparent_meteorite/keyword-suggestions-autocomplete").call(run_input={"seeds": ["coffee maker"], "engines": ["google", "youtube"], "expansions": ["plain", "questions"]})for row in client.dataset(run["defaultDatasetId"]).iterate_items():print(row["suggestion"], row.get("engine"), row.get("rank"))
- n8n, Make, Zapier: use the Apify app's "Run Actor" step with actor
transparent_meteorite/keyword-suggestions-autocomplete, paste the input JSON, then map the dataset items to your sheet. - AI agents (MCP): connect the Apify MCP server at
https://mcp.apify.com?tools=transparent_meteorite/keyword-suggestions-autocompleteto Claude, Cursor or any MCP client, and your agent can run this Actor as a tool.
Fair use and legal
The Actor calls the public suggestion endpoints that search boxes use. It does not log in, scrape result pages or collect personal data. Autocomplete suggestions are short phrases generated from aggregated queries. Follow each engine's terms for how you use the data. This is an independent tool, not affiliated with Google, YouTube, Microsoft, Amazon or DuckDuckGo.
FAQ
Do I need an API key or a proxy? No. There is no login, key or proxy.
Which engine should I start with? Google for general keywords, YouTube for video topics, Amazon for product terms. Bing and DuckDuckGo add a second opinion, and merged mode shows where they agree.
How many suggestions will I get? Roughly 8 to 15 per query per engine. A seed with A-Z and question expansion on three engines usually returns 700 to 1,300 unique rows.
Does it return search volume? No. It returns the suggestions and where each ranks. Pair it with a trends or volume source if you need numbers.
Can I use a different country or language? Yes. Set country (for example de) and language (for example de). Amazon switches to that marketplace.
Why do I get fewer rows than expected? Duplicates are removed by default, an engine returned fewer suggestions for a rare seed, or you hit Max suggestions.
Why was an engine skipped? If an engine does not answer six times in a row with no successes it is skipped for the run, and the warning is saved in OUTPUT.
Support
Seeing odd results for a seed, country or engine? Open an issue on the Issues tab with the run URL and the seed.
Related actors
- YouTube Video Scraper to pull the videos behind the YouTube keywords you found.
- Technical SEO Audit to check the pages you plan to rank for these keywords.
- Google SERP Real URLs Scraper to see who ranks for a keyword today.
- Amazon Product Sales Signals to test the Amazon terms against real product demand.
Changelog
- 0.1: First private build. Five engines plus Google Shopping and Images, A-Z, 0-9, question and preposition expansion, merged and deduped modes.