Google Autocomplete & Keyword Suggest Scraper

Scrape Google Autocomplete suggestions at scale. Give it seed keywords and get back ranked long-tail phrases — expanded with A–Z and 0–9 suffixes, question prefixes (how/what/why/…), multiple countries and languages, and Google's own relevance score attached to every phrase. One clean row per suggestion, JSON or CSV out.

Why use this scraper?

• Google's relevance score included — each suggestion carries the score Google itself uses to rank it (client=chrome data), not just an ordered list.
• Real long-tail expansion — each seed fans out through 26 letters, 10 digits, and 13 question prefixes, then optionally recurses: suggestions become new seeds (expansionDepth).
• Multi-country, multi-language — sweep gl/hl combinations in one run (e.g. US + GB + DE) and compare what Google suggests where.
• A volume proxy nobody else exposes — volumeProxy counts how many distinct query variants surfaced each phrase. A completion that recurs across suffixes, countries and depth levels is more broadly searched.
• No API key, no login — the endpoint is public; the actor handles pacing and budgets.

Use cases

• SEO keyword research — mine long-tail and question keywords for content plans, briefs and clusters.
• PPC discovery — find query variants worth bidding on (or adding as negatives).
• Content ideation — expandQuestions surfaces FAQ-style queries ("can coffee grinder be used for spices") ready for articles and schema markup.
• Market comparison — same seed, different countries: see how demand phrasing differs by market.
• Brand monitoring — watch what Google autocompletes after your brand name.

How it works

Each seed keyword is queried against Google's own autocomplete endpoint, then fanned out through the expansions you enable (A–Z, 0–9, question prefixes) and, optionally, recursed — suggestions become new seeds. Every phrase comes back with Google's relevance score, deduplicated across the run.

Input

Field	Type	Default	Notes
keywords	array	— (required)	Seed terms, e.g. ["coffee grinder"]
expandAlphabet	boolean	false	Append a–z to each seed
expandNumbers	boolean	false	Append 0–9 to each seed
expandQuestions	boolean	false	Prefix how/what/why/where/when/which/who/can/is/vs/for/without/near
expansionDepth	integer	0	Feed discovered suggestions back as new seeds (keyword tree)
regions	array	["US"]	ISO country codes → Google gl param
languages	array	["en"]	Language codes → Google hl param
maxSuggestionsPerKeyword	integer	0	Per-seed cap after ranking (0 = no cap)
maxItems	integer	1000	Hard cap on total rows
deduplicateSuggestions	boolean	false	One row per unique phrase across seeds
includeMetadata	boolean	true	Include Google's suggestion type per row
maxConcurrency	integer	10	Parallel requests
proxy	object	—	Optional; the endpoint works without one

Example input

{
    "keywords": ["coffee grinder"],
    "expandAlphabet": true,
    "expandQuestions": true,
    "regions": ["US", "GB"],
    "languages": ["en"],
    "maxSuggestionsPerKeyword": 100
}

Output

One row per suggestion:

{
    "suggestion": "coffee grinder electric",
    "seedKeyword": "coffee grinder",
    "query": "coffee grinder e",
    "country": "US",
    "language": "en",
    "rank": 1,
    "volumeProxy": 3,
    "googlePosition": 0,
    "relevance": 1252,
    "depth": 0,
    "surfacedByQueries": ["coffee grinder e", "coffee grinder"],
    "suggestType": "QUERY",
    "scrapedAt": "2026-07-29T12:00:00.000Z"
}

Key output fields

Field	Meaning
relevance	Google's own score for the suggestion — higher = Google ranks it stronger
volumeProxy	How many distinct query variants surfaced this phrase (relative popularity)
googlePosition	Best 0-based slot Google served it at
rank	Final rank within its seed after sorting by relevance → volume → position
suggestType	Google's tag: QUERY, NAVIGATION, ENTITY, …
depth	0 = found directly from the seed; 1+ = found via recursion

FAQ

Is this search volume? No — Google Autocomplete has no official volume numbers. relevance is Google's internal ranking score and volumeProxy is a cross-variant recurrence count; together they order keywords by how strongly Google surfaces them.

Do I need a proxy? Usually not; the endpoint tolerates normal request volumes. Heavy multi-country runs route through per-country exits automatically.

How many requests does a run make? Seeds × suffix variants × countries × languages, plus recursion levels. A hard budget (5,000 fetches) and a per-level seed cap keep runs bounded; the log tells you if a budget was hit.

Can I get question keywords only? Enable expandQuestions and filter rows whose suggestion starts with a question word.

Support

• Bug or feature request? Open a ticket on the Issues tab — the fastest way to reach me; I actively maintain this scraper.
• Email: muhamed.didovic@gmail.com
• Website: muhamed-didovic.github.io

Explore More Scrapers

• TikTok Keyword Suggestions Scraper — the same engine for TikTok autocomplete
• Ubersuggest Scraper — keyword metrics with volume estimates
• Google Search Scraper — SERP, AI Overview, ads and operators
• Ahrefs Scraper — DR, backlinks, keywords and traffic

⚠️ Disclaimer

This actor collects publicly available autocomplete data from Google for lawful purposes such as keyword research and market analysis. You are responsible for how you use the output, including compliance with Google's Terms of Service and applicable laws in your jurisdiction.

SEO Keywords

Google autocomplete scraper, Google suggest scraper, keyword suggestion scraper, long-tail keyword research tool, Google keyword ideas, autocomplete API alternative, question keywords scraper, SEO keyword discovery, PPC keyword expansion, Google search suggestions export.