Keyword Autocomplete Scraper
Keyword Autocomplete Scraper queries the public autocomplete endpoints of Google, YouTube, Bing, Amazon and DuckDuckGo for each seed keyword and returns one JSON row per suggestion: seed, query, engine, suggestion, position, language, country and scrapedAt. Alphabet or question expansion pulls the long-tail suggestions a bare seed never returns. It costs $0.0003 per suggestion ($0.30 per 1,000), with no API key, and empty or failed queries are free.
from $0.0003 per autocomplete suggestion scraped, $0.3 per 1,000, down to $0.0002 on higher Apify plans ยท Updated Nov 2026, version 0.1.3
Key facts
- Price
- from $0.0003 per autocomplete suggestion scraped, $0.3 per 1,000, down to $0.0002 on higher Apify plans
- Pricing model
- Pay per event
- Output
- JSON, CSV, Excel, XML, RSS via Apify datasets
- Access
- Apify Console, REST API, Python and JS clients, MCP tool keyword_suggestions
- Version
- Updated Nov 2026, version 0.1.3
- Actor
subimpact/keyword-autocomplete-suggestions
Who it is for
- SEO teams building keyword lists before a content plan
- Ecommerce sellers mining Amazon search terms
- YouTube creators looking for topics with real demand
- Content strategists gathering question keywords for briefs
Use cases
Expand a seed with the alphabet
Set expand to alphabet and each keyword is sent alone and then with a to z and 0 to 9 appended, 37 queries per engine per keyword. Every row records the expanded query in the query field, so coffee advent calendar can be traced to coffee a. Position tells you where it sat in that engine's list rather than leaving you to guess the strength.
Mine question keywords
expand set to questions sends the seed plus who, what, where, when, why, how, can, is, are and does, then vs, for and near. That is how how coffee is made or coffee vs tea surface as separate suggestions with their own rank. The result maps neatly onto FAQ sections, comparison pages and brief headings.
Compare engines for one seed
engines takes google, youtube, bing, amazon and duckduckgo, so the same keyword list can be run across all five. Each row carries its engine, which makes a side-by-side view of what shoppers type on Amazon versus what viewers search on YouTube. The seed and query fields keep the comparison anchored to the same input.
Localise suggestions
language and country set the locale, sent as hl and gl to Google and YouTube, setlang to Bing and accept-language to Amazon, with country also selecting the Amazon marketplace where one exists. Rows carry both values, so a my run and a us run of the same seeds produce separate datasets that are easy to keep apart.
Track how suggestions shift
Suggestions change with seasons and news, so schedule the same keyword list monthly and keep position and scrapedAt. A suggestion moving up the list is a demand signal before any search volume tool reflects it. dedupe keeps one row per keyword and engine, so the row counts across months stay comparable.
Output fields
| Field | Type | Description |
|---|---|---|
seed | string | The keyword from your input that this suggestion was collected for, as you wrote it. |
query | string | The expanded query actually sent to the engine for this row: the seed itself, or the seed plus an expansion term ("coffee a", "how coffee", "coffee vs"). |
engine | string | Which autocomplete engine returned the suggestion: google, youtube, bing, amazon or duckduckgo. |
suggestion | string/null | The autocomplete suggestion text, exactly as the engine returned it. null on empty/error status rows. |
position | integer/null | 1-based rank of the suggestion in that engine's response for this query (1 = the engine's top suggestion). null on empty/error status rows. |
language | string | Interface language used for the request (hl / setlang / accept-language), e.g. en, de, pt-BR. |
country | string | Two-letter country code used for the request (gl / Amazon marketplace), e.g. us, gb, jp. |
scrapedAt | string | ISO-8601 timestamp of the request that produced this row. |
status | string | ok = a suggestion row (charged). empty = the engine returned no suggestions for that query (free). error = the request failed (free). Empty and error rows carry a human-readable reason in `error`. |
error | string/null | Why an empty or error row was delivered (up to 400 characters). null on ok rows. |
Sample output
From one of our own test runs on 20 Nov 2026, trimmed. Personal data is redacted.
2 rows, first 3 shown
[
{
"seed": "coffee",
"query": "coffee",
"engine": "amazon",
"suggestion": "coffee",
"position": 1,
"language": "en",
"country": "us",
"scrapedAt": "2026-11-20T02:53:19.147Z",
"status": "ok",
"error": null
},
{
"seed": "coffee",
"query": "coffee",
"engine": "amazon",
"suggestion": "coffee maker",
"position": 2,
"language": "en",
"country": "us",
"scrapedAt": "2026-11-20T02:53:19.147Z",
"status": "ok",
"error": null
}
]Input
| Field | Type | Required | Description | Default |
|---|---|---|---|---|
keywords | array | Yes | Seed keywords to collect autocomplete suggestions for, e.g. ["coffee", "running shoes"]. Duplicates and blank entries are dropped. Up to 10,000 per run. | - |
engines | array | No | Which autocomplete engines to query: google (Google Search), youtube (YouTube), bing (Bing), amazon (Amazon product search) or duckduckgo (DuckDuckGo). Default ["google"]. Pick several to compare what each engine suggests for the same ke... | ["google"] |
language | string | No | Interface language for the suggestions, e.g. en, de, es, ja, pt-BR. Sent as hl to Google/YouTube, setlang to Bing and accept-language to Amazon. DuckDuckGo's autocomplete ignores it (the value is still written to every row). | en |
country | string | No | Two-letter country code for the suggestions, e.g. us, gb, de, jp, my. Sent as gl to Google/YouTube and used to pick the Amazon marketplace (US, GB, DE, JP and IN have their own; other countries fall back to the US marketplace). DuckDuckG... | us |
expand | string | No | none queries only the seed. alphabet also queries the seed plus a-z and 0-9 (37 queries per keyword and engine). questions also queries the seed plus who/what/where/when/why/how/can/is/are/does and vs/for/near (14 queries per keyword and... | none |
maxSuggestionsPerKeyword | integer | No | How many suggestions to keep for each keyword on each engine (1-1,000, default 200). Expansion stops for that keyword/engine once the limit is reached. | 200 |
maxItems | integer | No | Maximum number of suggestion rows delivered in one run (1-100,000, default 1,000). Counts delivered suggestions only: free empty/error status rows do not use up the cap. Default 1,000 costs $0.30. | 1000 |
dedupe | boolean | No | When true (default), a suggestion that an engine already returned for the same keyword is delivered once, even when a later expanded query returns it again (e.g. "coffee" from both "coffee" and "coffee shop"). Duplicates are counted in t... | true |
concurrency | integer | No | How many keyword/engine tasks run in parallel (1-20, default 5). Each task sends its expanded queries one after another (up to 37 requests with alphabet expansion), so 5 is already gentle on the engines' rate limits. | 5 |
proxyConfiguration | object | No | Optional. Plain HTTP works and is the default; switch on Apify Proxy if an engine starts rate-limiting your runs or if you need suggestions as seen from another country. | - |
Example input
{
"keywords": [
"coffee",
"running shoes"
],
"proxyConfiguration": {
"useApifyProxy": false
}
}Pricing
| Event | What it means | Price | Lowest tier |
|---|---|---|---|
suggestion-scraped | One autocomplete suggestion delivered with its engine, rank and the expanded query that produced it. Platform usage included. Empty and failed queries are free status rows. | $0.0003 | $0.0002 |
$0.3 at the listed price, or $0.2 on higher Apify plans, for 1,000 autocomplete suggestion scraped units.
Prices from the Apify Store on 20 November 2026. Apify shows the final price before each run.
How it compares with other Apify actors
Price per 1,000 suggestions at Apify Store list prices. Every row is a live listing on the Store, read on 20 Nov 2026, so a buyer can check each number at the source.
| Actor | Price per 1,000 suggestions (Free plan) | Higher plans (Gold) | Start fee per run | Platform usage | Users | Rating | Last build |
|---|---|---|---|---|---|---|---|
| Keyword Autocomplete Scraper: Google, YouTube, Amazon, Bing This API | $0.3 | $0.2 | None | Included | 1 | No reviews yet | 18 Nov 2026 |
| Google Autocomplete API Scraper | $1.00 | $0.88 | None | Included | 1,131 | No reviews yet | 7 Nov 2026 |
| Google Autocomplete Scraper: Keyword Suggestions | Charges per keyword expanded rather than per suggestion returned. | None | Included | 966 | 4.8 (3 reviews) | 12 Nov 2026 | |
| Google Autocomplete & Keyword Suggest Scraper | $0.5 | $0.5 | $0.005 | Included | 243 | 4.5 (2 reviews) | 15 Nov 2026 |
| ๐ All-in-One Autocomplete Keywords Tool | $2.99 | $2.99 | $0.09 | Paid separately | 181 | No reviews yet | 25 Nov 2025 |
Where Google Autocomplete API Scraper is stronger
Where Google Autocomplete API Scraper is stronger
- Larger user base and a longer review history on the Store
- Many seeds across languages and countries in one run
- Adds a relevance score and a suggestion subtype per row
Where this API is stronger
- Five engines rather than Google alone
- Alphabet and question expansion in the same input set
- Queries that return no suggestions cost nothing
Where Google Autocomplete Scraper: Keyword Suggestions is stronger
Where Google Autocomplete Scraper: Keyword Suggestions is stronger
- Google relevance score attached to each suggestion
- Recursive depth with question and number expansion
- Very active listing with frequent recent runs
Where this API is stronger
- Charged per suggestion delivered rather than per keyword expanded
- Google, YouTube, Bing, Amazon and DuckDuckGo in one run
- Deduplication across seeds as an input flag
Where Google Autocomplete & Keyword Suggest Scraper is stronger
Where Google Autocomplete & Keyword Suggest Scraper is stronger
- Alphabet and number expansion with a recursion depth setting
- Deduplication across seeds plus a total item cap
- Optional metadata block on suggestion rows
Where this API is stronger
- Several search engines rather than Google alone
- Rank and the expanded query recorded for every suggestion
- No start fee before the first suggestion row
Where ๐ All-in-One Autocomplete Keywords Tool is stronger
Where ๐ All-in-One Autocomplete Keywords Tool is stronger
- Covers Google, Pinterest, Instagram, TikTok and Amazon
- Very simple single keyword input
- Established publisher listing on the Store
Where this API is stronger
- Engine list input so one seed can run across five engines
- Expanded query and rank recorded per suggestion
- Platform usage included in the suggestion price
Autocomplete actors are mostly Google only, and they differ in whether they charge per suggestion or per keyword expanded. The devilscrapes and memo23 listings charge per row with deep expansion, and automation-lab charges per keyword. Ours spans five engines and charges per suggestion, so broad expansion does not multiply the bill by empty results.
Prices are Apify Store list prices, checked 20 Nov 2026. Platform usage 'Paid separately' means Apify bills compute on top of the event price. Check each listing before you buy.
How to run it
- Add your seeds to keywords, for example ["coffee", "running shoes"]; up to 10,000 per run, with duplicates removed.
- Pick engines, for example ["google", "youtube", "bing"], and set language and country for the locale you want.
- Choose expand: none for the plain seed, alphabet for a to z plus digits, or questions for who, what, how and similar.
- Set maxSuggestionsPerKeyword, 200 by default, and maxItems, 1,000 by default, as the row cap for the run.
- Start the run in the Apify Console; one row per suggestion arrives with its engine, position and the query that produced it.
- Export as JSON, CSV or Excel, and filter on status to separate real suggestions from free empty rows.
cURL
curl -X POST "https://api.apify.com/v2/acts/subimpact~keyword-autocomplete-suggestions/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
-H "Content-Type: application/json" \
-d '{"keywords":["coffee","running shoes"],"proxyConfiguration":{"useApifyProxy":false}}'Python
from apify_client import ApifyClient
client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("subimpact/keyword-autocomplete-suggestions").call(run_input={"keywords":["coffee","running shoes"],"proxyConfiguration":{"useApifyProxy":false}})
for item in client.dataset(run["default_dataset_id"]).iterate_items():
print(item)JavaScript
import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: '<YOUR_APIFY_TOKEN>' });
const run = await client.actor('subimpact/keyword-autocomplete-suggestions').call({"keywords":["coffee","running shoes"],"proxyConfiguration":{"useApifyProxy":false}});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items);MCP
claude mcp add subimpact-data -e APIFY_TOKEN=<YOUR_APIFY_TOKEN> -- npx -y @subimpact/data-mcp
# Then ask your agent to use the keyword_suggestions tool.n8n
Use the Apify node (Run an Actor).
Actor ID: subimpact/keyword-autocomplete-suggestions
Input JSON: paste the example input above.
MCP users: @subimpact/data-mcp exposes these actors as tools, so an agent can call keyword_suggestions directly.
Limits
- Suggestions carry no search volume, only the engine's own ranking.
- DuckDuckGo's autocomplete ignores the language setting.
- The India Amazon marketplace is mapped but was not verified end to end; a failing marketplace returns free error rows.
- With alphabet expansion a single keyword sends 37 requests per engine, so lower concurrency if an engine starts refusing.
Questions people ask
What does a suggestion cost?
$0.0003 per delivered suggestion, which is $0.30 per 1,000 suggestions. There is no start fee, and empty results and errors are free rows, so a run against an engine that refuses requests costs nothing. The requests themselves are not charged, only the rows.
Do I need an API key?
No. The Actor calls the engines' public autocomplete endpoints over plain HTTP, with no key, no browser and no account. The suggestions are what each engine would show a user, so they include that engine's own locale handling and ranking.
Are these search volumes?
No. Autocomplete shows what people type, in the engine's own order, but it carries no volume numbers. position records the rank within that engine's response for that query, which is the only ordering signal in the data.
What does expansion actually do?
alphabet sends 37 queries per keyword and engine: the seed alone, then the seed with each letter and digit appended. questions sends 14: the seed plus who, what, where, when, why, how, can, is, are and does, then vs, for and near. Each row keeps the query that produced it.
Why is DuckDuckGo ignoring my language?
DuckDuckGo's autocomplete endpoint has no language parameter, so it returns its global list. The language value is still written to every row for consistency, but it does not affect that engine's suggestions.
Which Amazon marketplace is used?
The country input selects it: US, GB, DE, JP and IN have their own marketplaces, and any other country falls back to the US one. Amazon completions are marketplace specific, so the same seed returns different suggestions per country.
How many suggestions can one run return?
Up to 100,000 rows, with maxItems defaulting to 1,000 and maxSuggestionsPerKeyword defaulting to 200 per keyword and engine. Free empty and error rows do not consume the cap, so the cap is spent on real suggestions.
Need a data source that is not here?
We build and run scrapers for sites that have no public API. Tell us the site, the fields and how often you need them.