Google Autocomplete & Keyword Suggestions: 10 Search Engines

This Google autocomplete scraper and keyword suggestions tool collects autocomplete keyword ideas for SEO, content and e-commerce from ten search boxes in one run: Google, YouTube, Google Shopping, Bing, DuckDuckGo, Amazon, eBay, Apple App Store, Yahoo and Yandex. These are the same public suggestion endpoints the search boxes call, so there are no logins and no API keys.

Give it seed keywords, pick engines and countries, and switch on the expansions you want. You get one row per suggestion, plus a cross-engine agreement score that shows which ideas several engines agree on.

What makes it different

• Many engines, one dataset. Compare what people type into Google, YouTube, Amazon, eBay and the App Store side by side.
• Expansion modes. Alphabet soup (seed a to seed z, seed 0 to seed 9), question words (how, what, why, when, where, can, is, does, best ...), prepositions and comparisons (for, with, without, near me, vs, or ...), your own {seed} templates, and recursive depth (suggestions of suggestions, capped).
• Agreement score. Every row carries foundOnEngines (how many of the engines you queried returned this suggestion) and independentSources, which counts Bing, DuckDuckGo and Yahoo as one source because their suggestions overlap heavily.
• Per-seed summary (unique suggestions, top words, which engines agree and overlap) in the SUMMARY record.
• Country and language per run (US-en, DE-de, GB-en, FR-fr ...). Question and preposition words are built in for en, de, fr, es, it, pt and nl.
• Reliable. Retries with backoff, a fresh proxy session when an engine answers with a block page, and an engine that fails every request is stopped and reported rather than hammered. Failures are listed in the NOTES record, never silent.
• Cost caps. maxSuggestions and maxRequests stop the run, and the planned request count is logged at the start.

Input example

{
  "seedKeywords": ["coffee maker", "standing desk"],
  "engines": ["google", "youtube", "bing", "duckduckgo", "amazon"],
  "locales": ["US-en"],
  "alphabetSoup": true,
  "questionModifiers": true,
  "prepositionModifiers": false,
  "customTemplates": ["best {seed} for"],
  "recursiveDepth": 1,
  "maxSuggestions": 2000
}

Output example (one row per engine and suggestion)

{
  "engine": "google",
  "seed": "coffee maker",
  "query": "coffee maker",
  "suggestion": "coffee maker with grinder",
  "position": 2,
  "country": "US",
  "language": "en",
  "modifierType": "base",
  "depth": 0,
  "foundOnEngines": 8,
  "enginesQueried": 10,
  "independentSources": 6,
  "engines": ["google", "youtube", "google-shopping", "bing", "duckduckgo", "amazon", "ebay", "yahoo"],
  "timesSeen": 1
}

• query is what was typed into the search box, position is the rank in that engine's answer.
• modifierType is base, alphabet, number, question, preposition, custom or recursive; depth is 0 for the seed and its modifiers, 1+ for recursive levels.
• Duplicates are removed within an engine (case and spacing ignored). The row kept is the earliest and highest-ranked one; timesSeen says how many queries returned it. A suggestion that only repeats the query you typed is dropped.
• Rows are ordered so the suggestions found on the most engines come first.

SUMMARY (key-value store) has one object per seed and country: uniqueSuggestions, suggestionsPerEngine, foundOnNEngines, topAgreed, topWords, engineOverlap. NOTES lists skipped engines, failures and any limit that cut the run short.

Use cases

• SEO and content: long-tail ideas, questions people ask, comparison keywords (x vs y).
• E-commerce: what shoppers type on Amazon and eBay, per country.
• YouTube and app research: video topic ideas and App Store search hints.
• Find the ideas that show up on several engines and are therefore safer bets.

Engines and limits (please read)

• Autocomplete only. There is no search volume, CPC or competition data, because the engines do not publish it.
• Suggestions are personalised a little by location. Google and YouTube accept any country and language. Amazon (18 marketplaces), eBay (13 sites), Yahoo (12 regions) and the App Store (18 storefronts) are chosen by country only, and an unsupported country is skipped and listed in NOTES. Amazon India is not available from the suggestion API.
• DuckDuckGo sometimes adds local-business style suggestions based on the server's location.
• Question and preposition words exist only for en, de, fr, es, it, pt and nl. For other languages use Custom templates (they work in any language) and, for non-Latin scripts, Custom characters for the alphabet. Default letters are a-z and 0-9 (Cyrillic for ru and uk).
• Each engine returns about 4 to 14 suggestions per query, so a big expansion means many requests: with alphabet soup, questions and prepositions, one seed on one engine is about 65 requests. recursiveDepth multiplies that. Use maxRequests and maxSuggestions.
• Google, YouTube and Google Shopping use the residential Apify proxy by default because Google blocks datacenter and home IPs quickly. Other engines are queried directly; there is an option to use the datacenter proxy if one starts blocking.
• Autocomplete endpoints are unofficial. If an engine changes or blocks them the run reports it in NOTES instead of returning a silent empty result.

Amazon autocomplete keywords

Add amazon to Engines and enter a product idea as the seed. You get what shoppers type into the Amazon search box, which is a quick way to find product keywords and long-tail phrases.

YouTube keyword ideas

Add youtube to Engines. The suggestions come from the YouTube search box, so they show what people search for on YouTube: video topics, titles and tags.

Google autocomplete questions

Turn on Question modifiers. The Actor adds "how", "what", "why", "can" and similar words to each seed, so you get the questions people ask Google about your keyword.

Google autocomplete API

Call it from the Apify API, Python, JavaScript, Make, Zapier or n8n. Example with the Apify Python client:

from apify_client import ApifyClient

run = ApifyClient("<APIFY_TOKEN>").actor("data-done-right/keyword-suggestions-tool").call(run_input={
    "seedKeywords": ["standing desk"], "engines": ["google", "amazon"], "questionModifiers": True})

Then read the suggestions from the run's default dataset.

Use it in Claude, ChatGPT and Cursor

Connect Apify's MCP server (mcp.apify.com) and your AI assistant can collect keyword ideas itself. For example: "give me question keywords for 'standing desk' from Google, and the ideas Amazon and Bing also suggest." The only required input is seedKeywords; engines picks the search boxes.

How much does it cost

Pay per event: $0.00055 per suggestion row delivered (about $0.55 per 1,000 rows), with no per-run start fee. Paid Apify plans get 10% (Silver) to 20% (Gold and up) off. Failed requests and skipped engines cost nothing, and the caps above bound the bill. Platform usage (proxy traffic is tiny: about 5 KB per request) is small for these runs.

Tips

• Start with 1 seed and 2 or 3 engines to see volume, then scale up.
• Compare foundOnEngines >= 3 rows across Google, Amazon and YouTube to find ideas with broad demand.
• Schedule weekly runs with Apify Schedules to watch how suggestions change.
• Pair with our Google Trends Actor to check interest over time for the best ideas.

FAQ

How do I get Google keyword suggestions in bulk?

Enter your seed keywords, choose Google (and any other engines), and switch on alphabet soup and question modifiers. You get one row per suggestion, with the modifier that produced it.

Where do the suggestions come from?

From the public autocomplete endpoints behind each search box: Google, YouTube, Google Shopping, Bing, DuckDuckGo, Amazon, eBay, the Apple App Store, Yahoo and Yandex. No login or API key is needed.

Does it show search volume or keyword difficulty?

No. Autocomplete only. Engines do not publish volume, CPC or competition data, so this tool does not invent it. The foundOnEngines agreement score shows which ideas several engines agree on.

Can I find question keywords, like "how" and "can" searches?

Yes. Question modifiers (how, what, why, when, where, can, is, does, best and more) are built in for English, German, French, Spanish, Italian, Portuguese and Dutch. Use custom templates for other languages.

Can I get keyword suggestions for YouTube or Amazon?

Yes. Pick YouTube, Amazon, eBay or the App Store, and set the country. Amazon, eBay, Yahoo and the App Store are chosen by country only.

How much do keyword suggestions cost?

It is pay per event: $0.00055 per suggestion delivered with no start fee, and failed requests are free. maxSuggestions and maxRequests cap the bill, so start with one seed and a few engines.

Related tools

• Google Trends Scraper checks interest over time and rising queries for your best ideas.
• AI Visibility Tracker shows whether ChatGPT, Claude and Gemini mention your brand for those topics.