Google Autocomplete Suggestions API

Get current Google Search autocomplete predictions for a bounded list of seed queries. This Actor is for SEO teams, content researchers, and AI agents that need clean, locale-aware suggestion snapshots for keyword discovery alongside search-volume metrics and Google search-result research.

Best fit for this Actor

Use it when you need the words Google currently predicts after a phrase, including optional prefix or suffix character expansion. Each completed lookup becomes one small JSON record with the exact query, ordered suggestions, selected language and country, source URL, and collection time. It works as a focused suggestions workflow, while keyword volumes, rankings, paid ads metrics, and SERP results each benefit from their own dedicated data source.

Google describes autocomplete as predictions derived from common searches and web patterns, which also vary with language, location, and live interest. Treat each record as a timestamped research signal rather than a factual statement or a popularity score.

From seed list to content brief

A content strategist starts with project management software, runs the seed in US English, and exports the returned long-tail suggestions. With suffix expansion enabled, the strategist gets a wider but explicitly capped set of live query variants. The team clusters the suggestions array, keeps sourceUrl and collectedAt for provenance, then chooses topics only after combining this signal with its own audience and volume research.

Quick start input

{
  "queries": ["project management software", "best running shoes"],
  "language": "en",
  "country": "US",
  "expansionMode": "none",
  "maxLookups": 100
}

For a deliberately bounded long-tail pass, add "expansionMode": "suffix". The Actor will fetch the original seed plus suffix variants, but will stop exactly at maxLookups rather than creating an open-ended run.

Input reference

Field	Type	What it controls
queries	string[]	1–50 seed phrases, each up to 120 characters. Exact repeated phrases are deduplicated.
language	string	BCP 47-style suggestion-language hint such as en, es, or pt-BR; defaults to en.
country	string	Two-letter country hint such as US, GB, or PK; defaults to US.
expansionMode	enum	none, prefix, suffix, or both; expansion requests additional query variants.
alphabet	string	Unique letters/digits used for expansion; defaults to abcdefghijklmnopqrstuvwxyz.
maxLookups	integer	A 1–500 hard cap over all original and expanded lookups; defaults to 100.

What data you receive

One dataset item is one completed Google autocomplete lookup, even if the live response contains an empty suggestion array. The stable array format avoids sparse suggestion_01 / suggestion_02 fields and is easier to pass into clustering, spreadsheets, and AI workflows.

Field	Meaning
seedQuery	Original seed from your input.
query and expansion	Exact lookup phrase and whether it is a seed, prefix, or suffix variant.
suggestions and suggestionCount	Ordered, de-duplicated current predictions and their count.
language and country	Context hints sent with the lookup; not a geographic guarantee.
sourceUrl and collectedAt	Replayable public endpoint URL and UTC collection timestamp.

{
  "seedQuery": "best running shoes",
  "query": "best running shoes",
  "expansion": "seed",
  "language": "en",
  "country": "US",
  "suggestions": [
    "best running shoes for men",
    "best running shoes for women"
  ],
  "suggestionCount": 2,
  "sourceUrl": "https://suggestqueries.google.com/complete/search?client=firefox&hl=en&gl=US&q=best+running+shoes",
  "collectedAt": "2026-08-24T10:00:00.000Z",
  "provider": "Google Search autocomplete"
}

Terminal outcomes

Read OUTPUT or RUN_SUMMARY before using the dataset downstream. COMPLETE means every planned lookup persisted; PARTIAL preserves completed records and identifies collection issues or a cost-cap stop; VALID_EMPTY means valid lookups returned no predictions; INVALID_INPUT explains which bounded input to correct; and UPSTREAM_FAILED clearly reports an unavailable lookup path. The terminal record gives downstream workflows a visible outcome instead of an ambiguous empty result.

Use through the API

curl "https://api.apify.com/v2/acts/khadinakbar~google-autocomplete-api/runs?token=$APIFY_TOKEN" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{"queries":["project management software"],"language":"en","country":"US","maxLookups":1}'

After the run, retrieve the default dataset and read OUTPUT plus RUN_SUMMARY. Preserve sourceUrl and collectedAt when a later workflow needs to distinguish an older suggestion snapshot from a current one.

Use with AI agents through Apify MCP

Find current English Google autocomplete predictions for project management software in the United States. Use no A-Z expansion and one lookup. Return the suggestions with their source URL and collection time, label them as autocomplete predictions rather than search volume, and read the terminal outcome before answering.

This Actor is a bounded, read-only query-suggestion tool. Use maxLookups to budget calls before an agent expands a seed list. It returns query snapshots with a clear complementary role next to semantic analysis, keyword difficulty, ranking reports, and Google-result collection.

Pricing

Event	Price	When charged
Actor start	$0.00005	Once when a run starts.
Autocomplete lookup	$0.005	After one lookup snapshot is successfully persisted, including a valid empty suggestion list.

This Actor uses Pay per event plus Apify platform usage. A default one-lookup run has up to $0.00505 in events, plus platform usage. maxLookups caps the lookup events at maxLookups × $0.005; an expanded run stays within that actor-level boundary. Check the live Pricing tab for the current event details and platform-usage configuration.

Best results

• Submit clear phrase prefixes rather than full questions when you want the widest completion set.
• Select the language and country that match the audience you are researching, then retain them with each exported record.
• Use A-Z expansion only when the added coverage is worth the extra lookup count; start with a small alphabet and maxLookups for exploration.
• Compare snapshots collected on different dates carefully: Google says predictions can change with trends, language, and location.

Builder's note

I designed the output as query-level and array-based after inspecting the live response shape. A fixed suggestion_01 layout looks tidy in a spreadsheet yet loses ordering context whenever Google returns fewer predictions. This contract records exactly one response per billable lookup and pairs it with a source URL and timestamp so agent consumers receive a stable, compact entity.

Focused standalone workflow

This Actor is designed as a focused standalone workflow for collecting current autocomplete snapshots. It works well before a content team clusters phrases, plans a brief, or passes selected terms into its preferred volume, rank, and SERP research tools.

Responsible use

Use publicly available autocomplete data for lawful, authorized research and comply with Google’s terms and applicable laws. Treat predictions as contextual query-completion signals, and combine them with suitable evidence before making high-impact decisions.