Keyword Autocomplete Scraper: Google, YouTube, Amazon, Bing
Pricing
from $0.20 / 1,000 autocomplete suggestion scrapeds
Keyword Autocomplete Scraper: Google, YouTube, Amazon, Bing
Every autocomplete suggestion for your keywords from Google, YouTube, Bing, Amazon and DuckDuckGo as JSON: suggestion, engine, rank and the expanded query that produced it. Alphabet and question expansion, no API key. From $0.30 per 1,000 suggestions.
Pricing
from $0.20 / 1,000 autocomplete suggestion scrapeds
Rating
0.0
(0)
Developer
subimpact network
Maintained by CommunityActor stats
0
Bookmarked
1
Total users
1
Monthly active users
2 days ago
Last modified
Categories
Features
Share
Keyword Autocomplete Suggestions API (Google, YouTube, Bing, Amazon, DuckDuckGo)
Every autocomplete suggestion for your keywords, as JSON. Give the actor a list of seed keywords and it queries the public autocomplete endpoints of up to five engines — Google Search, YouTube, Bing, Amazon and DuckDuckGo — and returns one clean row per suggestion, with the engine, its rank and the expanded query that produced it.
- No start fee. $0.30 per 1,000 suggestions ($0.0003 per suggestion delivered).
- No API key, no browser. Plain HTTP against the engines' public suggestion endpoints.
- 5 engines. Google by default; add YouTube, Bing, Amazon and DuckDuckGo with one input field.
- Expansion. Query the seed alone, or also
seed + a–z + 0–9(alphabet), or also the seed pluswho/what/where/when/why/how/can/is/are/doesandvs/for/near(questions) to pull suggestions the plain seed never returns. - Up to 100,000 suggestions per run, 5 keyword/engine tasks in parallel by default.
What you get
One dataset item per suggestion. Useful for keyword research, store listing and ad-group ideas, YouTube topic discovery, Amazon product-search terms, question mining for content briefs, and tracking how suggestions change over time.
| Field | Meaning |
|---|---|
seed | The keyword from your input |
query | The expanded query actually sent (the seed, or coffee a, how coffee, coffee vs, …) |
engine | google, youtube, bing, amazon or duckduckgo |
suggestion | The suggestion text, exactly as the engine returned it |
position | 1-based rank in that engine's response for that query (1 = top suggestion) |
language, country | The locale used for the request |
scrapedAt | ISO-8601 timestamp of the request that produced the row |
status | ok for a suggestion row; empty / error for the free diagnostic rows described below |
error | Why an empty/error row was delivered; null on ok rows |
Input
{"keywords": ["coffee", "running shoes"],"engines": ["google", "youtube", "bing"],"language": "en","country": "us","expand": "alphabet","maxSuggestionsPerKeyword": 200,"maxItems": 1000,"dedupe": true,"concurrency": 5}
| Input | Type | Default | Notes |
|---|---|---|---|
keywords | array | — (required) | Seed keywords. Trimmed, de-duplicated, up to 10,000 per run. |
engines | array | ["google"] | Any of google, youtube, bing, amazon, duckduckgo. |
language | string | en | hl for Google/YouTube, setlang for Bing, accept-language for Amazon. Accepts de, ja, pt-BR, … DuckDuckGo's autocomplete has no language parameter, so it returns its global list (the value is still written to every row). |
country | string | us | Two-letter code. gl for Google/YouTube and the Amazon marketplace selector. |
expand | string | none | none, alphabet (37 queries per keyword/engine) or questions (14 queries per keyword/engine). |
maxSuggestionsPerKeyword | integer | 200 | Cap per keyword and engine, 1–1,000. |
maxItems | integer | 1000 | Row cap for the whole run, 1–100,000. Counts delivered suggestions; free empty/error rows do not use it up. |
dedupe | boolean | true | Deliver a suggestion once per keyword and engine even if a later expanded query returns it again. |
concurrency | integer | 5 | Keyword/engine tasks in parallel, 1–20. Each task sends its expanded queries one after another. |
proxyConfiguration | object | off | Optional Apify Proxy. Plain HTTP works; turn it on only if an engine starts rate-limiting your runs. |
A default run (the two pre-filled keywords on Google, no expansion) returns roughly 20 suggestions and costs well under a cent.
Output
One dataset item per suggestion. This is a real row from a live run (Google, country: "us"):
{"seed": "coffee","query": "coffee","engine": "google","suggestion": "coffee near me","position": 1,"language": "en","country": "us","scrapedAt": "2026-11-15T09:08:50.640Z","status": "ok","error": null}
A query that the engine answers with nothing, or a request that fails, is delivered as a free
diagnostic row — suggestion and position are null and nothing is charged for it (shape
example):
{"seed": "coffee","query": "coffee qq","engine": "google","suggestion": null,"position": null,"language": "en","country": "us","scrapedAt": "2026-11-15T09:08:11.420Z","status": "empty","error": "google returned no autocomplete suggestions for \"coffee qq\"."}
Engines and expansion
| Engine | Endpoint used | Notes |
|---|---|---|
google | suggestqueries.google.com/complete/search?client=firefox | Google Search suggestions; honours hl / gl. |
youtube | same endpoint with ds=yt | YouTube search suggestions. |
bing | api.bing.com/osjson.aspx | Bing suggestions; honours setlang. |
amazon | completion.amazon.<tld>/api/2017/suggestions | Amazon product-search suggestions. The marketplace and its id must match, so country picks the pair: US (ATVPDKIKX0DER), GB (A1F83G8C2ARO7P), DE (A1PA6795UKMFR9), JP (A1VC38T7YXB528), IN (A21TJRUUN4KGV). Any other country uses the US marketplace. |
duckduckgo | duckduckgo.com/ac/?type=list | DuckDuckGo autocomplete; no locale parameter. |
expand: "alphabet" sends the seed and then coffee a … coffee z, coffee 0 … coffee 9
(coffee a returns coffee advent calendar, coffee and peppers, …).
expand: "questions" sends the seed and then who coffee … does coffee, plus coffee vs,
coffee for, coffee near (how coffee returns how coffee is made, how coffee was discovered, …).
Every row records which of those queries produced the suggestion, so you can tell a seed suggestion
from a long-tail one.
Billing
- $0.30 per 1,000 suggestions — $0.0003 for each suggestion row delivered.
- No start fee. You are not charged for the requests themselves, for empty results, or for
errors. Every row with
status: "empty"orstatus: "error"is free, which means a run against a blocked engine costs nothing. maxItems(and the run's max charge) stops the run cleanly; the status message reports how many rows were actually delivered.
Notes and limits
- These are the engines' public suggestion endpoints, so the suggestions are exactly what each engine would show a user, including that engine's own locale and ranking behaviour. They are not search volumes; no search-volume data is included.
- Bing's endpoint answers
HTTP 204with an empty body to desktop-browser user agents from some networks; the actor sends the mobile user agent Bing accepts (verified live 2026-11-15). - Amazon's completions are marketplace-specific; the India marketplace (
completion.amazon.in) answeredHTTP 502during our live checks on 2026-11-15, so India is mapped but not verified end to end. A failing marketplace produces freeerrorrows, never charges. - With
concurrency: 5andexpand: alphabeta single keyword sends 37 requests per engine. Each keyword/engine task sends its queries one after another, so a run only ever has aboutconcurrencyrequests in flight however many expansion queries it has. If an engine ever starts refusing requests, lowerconcurrency, or enable Apify Proxy inproxyConfiguration.
Run monitoring and privacy
If a run fails or times out, this Actor tells its developer the run ID and status, so we can fix problems fast. It never sends your input, your results or your account data.