Keyword Suggestions Scraper

Keyword Suggestions Scraper collects keyword suggestions from the autocomplete of Google, YouTube, Amazon, Bing, eBay and TikTok — plus DuckDuckGo, Yandex, Baidu and Naver — for any seed keyword. It expands each seed with the alphabet soup method (seed + a…z, seed + 0…9, question words, prepositions, your own prefixes and suffixes) and can go a second level deep for long-tail keywords.

Every row is one unique keyword suggestion with the seed it came from, the source and its position in the dropdown, the expansion that produced it, every other source that suggested the same phrase (sources), plus computed fields: wordCount, characterCount, isQuestion and a heuristic search intent. $0.11 per 1,000 keywords on the Starter plan ($0.125 on the Free plan, $0.085 on Business and higher). No fee per run, no login, no API key.

Tested head-to-head — 13 October 2026

Same request for this Actor and six popular keyword-suggestion Actors in the Apify Store: seed running shoes, Google, alphabet expansion, 20 keywords wanted, US / English.

	This Actor	Other Actors tested
Rows for "20 keywords"	20, all unique	19–405; three of six ignored the limit (270, 360 and 405 rows), one returned 16 duplicates
Cost of this run (Free plan)	$0.0025, no start fee	$0.015–$0.10; five of six list a start fee ($0.00005–$0.20)
Run time	2 s	3–78 s
Autocomplete sources	10 (Google, YouTube, Bing, Amazon, eBay, TikTok, DuckDuckGo, Yandex, Baidu, Naver)	1–10; none of them reads TikTok, Yandex, Baidu or Naver
Columns per row	17, incl. sources merged, position, intent, question flag	8–31

What is Keyword Suggestions Scraper?

It is a Google Autocomplete scraper and keyword research tool in one Actor. The search boxes of Google, YouTube, Amazon, Bing, eBay and TikTok suggest what real people type. Those suggestions are the raw material of keyword research: blog topics, YouTube titles, Amazon listing keywords, ad keywords and FAQ questions. Collecting them by hand means typing the seed, then the seed plus "a", plus "b"… into every search box. This Actor does exactly that through the public autocomplete endpoints the sites' own search boxes call, merges and de-duplicates the answers, and gives you one clean table.

What it is not: it does not return search volume, CPC or keyword difficulty. These are keyword suggestions, not volume metrics. Pair them with Google Trends Scraper to see whether interest in a keyword is rising or falling.

What data can Keyword Suggestions Scraper extract?

Field	Example	Meaning
keyword	is bali visa free for indians	The suggestion, as the source returned it
seed	bali visa	Your seed keyword
source	google	First source that suggested it
sources	["google", "youtube"]	Every source that suggested the same phrase for this seed
position	1	Rank in the source's dropdown (1 = top)
query	is bali visa	Exact text typed into autocomplete
expansion	is	What was added to the seed: letter, digit, question word, preposition, custom word, or the parent keyword at depth 2
expansionType	questions	seed, alphabet, numbers, questions, prepositions, custom or depth2
depth	1	1 = seed and its expansions, 2 = found by re-querying a level-1 keyword
language, country	en, US	Locale used
wordCount, characterCount	6, 29	Length of the keyword
isQuestion	true	Starts with a question word or ends with "?"
intent	question	Heuristic search intent (see below)
scrapedAt	2026-10-01T08:04:21.518Z	When the row was made

Sources and how they are localised

Source	source value	Localised by
Google	google	language + country
YouTube	youtube	language + country
Bing	bing	language + country (market)
Amazon	amazon	country → marketplace: US, GB, DE, FR, IT, ES, CA, JP, AU, MX, BR, NL, SE, PL, AE, SG, TR, SA, BE
eBay	ebay	country → site: US, CA, GB, AU, AT, BE, FR, DE, IT, NL, ES, CH, IE, PL
TikTok	tiktok	not localised — TikTok answers for the location of the server
DuckDuckGo	duckduckgo	language + country
Yandex	yandex	language
Baidu	baidu	not localised (Chinese search)
Naver	naver	not localised (Korean search)

A country that has no Amazon marketplace or eBay site in the lists above gives one free error row for that source; the other sources still run.

What the intent field means — honestly

intent is a heuristic, not a model and not Google data. It looks for English words in the keyword, in this order:

1. navigational — login, sign in, official website, customer service, phone number, .com …
2. transactional — buy, price, cost, cheap, coupon, deal, for sale, near me, book, rent, download …
3. commercial — best, top, vs, review, compare, alternative, brands, worth it …
4. question — the keyword starts with how / what / why / when / where / who / which / can / is / are / does / do / will / should … or ends with "?"
5. informational — everything else.

It is useful for sorting thousands of rows quickly. For non-English keywords most rows will read informational, because the word lists are English.

How much does it cost to scrape keyword suggestions?

Event	Free plan	Starter (Bronze)	Scale (Silver)	Business (Gold)
💰 Keyword suggestion (one unique row)	$0.000125	$0.00011	$0.0001	$0.000085
Run start	$0 — no start fee	$0	$0	$0
Error rows, filtered-out suggestions, duplicates	$0	$0	$0	$0

On the Starter plan that is $0.11 per 1,000 keywords, about 9,000 keywords per $1. On the Apify Free plan it is $0.125 per 1,000 keywords, so the $5 of monthly free credit covers about 40,000 keywords. Business, Platinum and Diamond pay $0.085 per 1,000.

Many keyword Actors charge a fee every time a run starts. This one does not: a scheduled run that finds 10 keywords costs $0.0011 on Starter ($0.00125 on the Free plan), not that plus a start fee. Duplicates are merged before billing — a phrase suggested by Google, YouTube and Bing is one row and one charge, with all three in sources. The cap Max keywords per seed is a hard ceiling on what you pay per seed.

The Actor uses plain HTTP requests, no browser and no proxy, so it is fast: the default run (1 seed, Google, alphabet soup, 50 keywords) finishes in about 5 seconds; 2 seeds × 4 sources with every expansion and depth 2 gave 4,000 keywords in 42 seconds in our test.

How to scrape keyword suggestions with Keyword Suggestions Scraper

1. Open the Actor and type one or more seed keywords, one per line.
2. Pick the sources — Google is selected by default; add YouTube, Amazon, Bing, eBay, TikTok or others.
3. Set language and country (two-letter codes like en / US, de / DE).
4. Choose the alphabet soup expansion: alphabet, numbers, questions, prepositions — or leave it empty for the plain suggestions of the seed.
5. Set Max keywords per seed and, for deeper long-tail research, Depth = 2.
6. Press Start. Download the table as CSV, Excel or JSON, or read it through the API.

⬇️ Input

The prefilled input — one seed, Google, alphabet soup, 50 keywords:

{
  "keywords": ["running shoes"],
  "sources": ["google"],
  "language": "en",
  "country": "US",
  "expansions": ["alphabet"],
  "maxKeywordsPerSeed": 50
}

Alphabet soup: questions and prepositions for a blog plan

{
  "keywords": ["bali visa", "digital nomad visa"],
  "sources": ["google", "youtube"],
  "expansions": ["questions", "prepositions"],
  "maxKeywordsPerSeed": 300
}

In our test this returned 418 unique keywords in 10 seconds, 174 of them questions.

Amazon keyword tool: product keywords for a marketplace

{
  "keywords": ["iphone"],
  "sources": ["amazon", "ebay", "bing"],
  "language": "de",
  "country": "DE",
  "expansions": ["numbers"],
  "requireSeedWords": true,
  "excludeWords": ["hülle"],
  "maxKeywordsPerSeed": 100
}

requireSeedWords keeps only suggestions that contain every word of the seed; excludeWords drops anything containing those words. Dropped suggestions are free.

Long-tail keywords: depth 2

{
  "keywords": ["coffee"],
  "sources": ["google"],
  "expansions": [],
  "depth": 2,
  "maxKeywordsPerSeed": 60
}

Level 1 asks for "coffee"; level 2 types every suggestion back in ("coffee near me" → "coffee near me open now", "coffee near me drive thru" …). The expansion of a depth-2 row is its parent keyword.

Other languages: your own prefixes and suffixes

{
  "keywords": ["виза в индонезию"],
  "sources": ["google", "yandex"],
  "language": "ru",
  "country": "RU",
  "expansions": [],
  "customPrefixes": ["как"],
  "customSuffixes": ["2026", "цена"],
  "maxKeywordsPerSeed": 80
}

The built-in question words and prepositions are English. For any other language, put your own words in Custom prefixes and Custom suffixes.

Field	Default	What it does
keywords	["running shoes"]	Seed keywords
sources	["google"]	google, youtube, amazon, bing, ebay, tiktok, duckduckgo, yandex, baidu, naver
language	en	Two-letter language code
country	US	Two-letter country code (UK is read as GB)
expansions	["alphabet"]	alphabet, numbers, questions, prepositions
maxKeywordsPerSeed	50	Unique keywords kept per seed (1–100,000)
depth	1	2 = re-query every level-1 suggestion
customPrefixes / customSuffixes	—	Your own words before / after the seed
requireSeedWords	false	Keep only keywords containing every seed word
excludeWords	—	Drop keywords containing any of these words
maxConcurrency	4	Parallel requests per source (1–10)

⬆️ Output

One row per unique keyword suggestion. Real rows from a test run:

[
  {
    "type": "keyword",
    "keyword": "is bali visa free for indians",
    "seed": "bali visa",
    "source": "google",
    "sources": ["google", "youtube"],
    "position": 1,
    "query": "is bali visa",
    "expansion": "is",
    "expansionType": "questions",
    "depth": 1,
    "language": "en",
    "country": "US",
    "isQuestion": true,
    "intent": "question",
    "wordCount": 6,
    "characterCount": 29,
    "scrapedAt": "2026-10-01T08:04:21.518Z"
  },
  {
    "type": "keyword",
    "keyword": "bali visa for u.s. citizens on arrival",
    "seed": "bali visa",
    "source": "google",
    "sources": ["google"],
    "position": 4,
    "query": "bali visa for",
    "expansion": "for",
    "expansionType": "prepositions",
    "depth": 1,
    "language": "en",
    "country": "US",
    "isQuestion": false,
    "intent": "informational",
    "wordCount": 7,
    "characterCount": 38,
    "scrapedAt": "2026-10-01T08:04:21.983Z"
  }
]

The Keyword suggestions tab in the Console shows keyword, seed, source, sources, position, expansion, intent, question flag, word count, country and language. The All fields tab adds the exact query, expansion type, depth and character count.

Error rows

If a source does not answer for a seed after retries, or the input cannot be used (a seed longer than 150 characters, a seed with no letters or digits, a country without an Amazon marketplace or eBay site), the Actor adds a free row:

{ "type": "error", "seed": null, "source": "ebay", "error": "eBay has no site for country \"JP\" here. Use one of: US, CA, GB, AU, AT, BE, FR, DE, IT, NL, ES, CH, IE, PL. Not charged.", "scrapedAt": "2026-10-01T08:06:02.000Z" }

An empty answer is not an error: a nonsense seed or a seed nobody searches simply has no suggestions. The run finishes successfully and adds a free row with kind: "no-results" that says so, instead of leaving the table blank:

{ "type": "error", "kind": "no-results", "seed": "xqzvbnmkjhgfdswerty", "error": "Google autocomplete has no suggestions for \"xqzvbnmkjhgfdswerty\" — people do not search for it in US/en. Try a shorter or more common phrase. Not charged." }

The same kind of free row tells you when your filters removed every suggestion. When a language or country code is not recognised, en / US is used instead and a free notice row at the end of the table says so. A seed made only of symbols (@@@ !!!) is not sent at all — Bing answers such queries with unrelated trending searches — and gets a free error row. A run summary (requests, empty answers, retries per source) is saved in the key-value store as SUMMARY.

Use cases for keyword suggestions

SEO and content planning

Turn one topic into hundreds of real search phrases. Question rows (isQuestion = true) become FAQ sections and "People also ask" style headings; wordCount ≥ 4 gives you long-tail keywords that are easier to rank for.

YouTube keyword research

YouTube autocomplete shows what people search on YouTube, not on Google — "running shoes for beginners", "running shoes review". Use it for titles, tags and video ideas.

Amazon and eBay listing keywords

Amazon and eBay suggestions show how shoppers name products: sizes, brands, materials, "for men". Use them in titles, bullet points and backend keywords, per marketplace.

Paid search

intent = transactional or commercial rows are candidates for ad groups; excludeWords keeps competitor brands out.

Social and short video

TikTok suggestions show trending phrasing for hooks, captions and hashtags.

AI agents and LLM pipelines

An agent can call the Actor with one seed and get structured, de-duplicated keywords back in seconds — for content briefs, topic clusters or competitor research.

Integrations

Every run is available through the Apify API, and the Actor works with n8n, Make, Zapier, Google Sheets, Airbyte, LangChain and the Apify MCP server.

Python:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_API_TOKEN>")
run = client.actor("lergassy/keyword-suggestions-scraper").call(run_input={
    "keywords": ["running shoes"],
    "sources": ["google", "youtube", "amazon"],
    "expansions": ["alphabet", "questions"],
    "maxKeywordsPerSeed": 500,
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(row["keyword"], row["sources"], row["intent"])

JavaScript:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: '<YOUR_API_TOKEN>' });
const run = await client.actor('lergassy/keyword-suggestions-scraper').call({
    keywords: ['running shoes'],
    sources: ['google', 'bing'],
    expansions: ['alphabet'],
    maxKeywordsPerSeed: 200,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items.length);

Schedule the Actor daily or weekly in the Console to watch how suggestions change, and send the dataset to Google Sheets or a webhook.

🤖 For AI Agents & LLM Apps

Compact reference for agents calling this Actor through the Apify MCP server or the Apify API (lergassy/keyword-suggestions-scraper).

Purpose: one row = one unique autocomplete suggestion for a seed keyword from Google, YouTube, Amazon, Bing, eBay, TikTok, DuckDuckGo, Yandex, Baidu or Naver. Answers "what do people type after X?", "which questions do people ask about X?", "which product keywords does Amazon suggest for X?".

Minimal input:

{ "keywords": ["running shoes"] }

Output: type, keyword, seed, source, sources, position, query, expansion, expansionType, depth, language, country, wordCount, characterCount, isQuestion, intent, scrapedAt. Error rows: type = "error", seed, source, error, scrapedAt. A seed with no suggestions gives an error row with kind = "no-results" (free, not a fault). Notice rows (an input value was replaced): type = "notice", notice. Error and notice rows come after the keyword rows.

Behaviors an agent should know:

• maxKeywordsPerSeed (default 50) caps rows and cost per seed; raise it together with expansions or depth: 2.
• No search volume, CPC or difficulty — suggestions only; intent is an English keyword heuristic.
• Use requireSeedWords and excludeWords instead of post-filtering; dropped rows are not billed.
• Duplicates across sources are merged into one row (sources lists them all); a keyword already delivered for an earlier seed in the same run is not repeated.
• Billed per keyword row: $0.000125 on the Free plan, $0.00011 on Starter, $0.0001 on Scale, $0.000085 on Business, Platinum and Diamond. Error rows, notice rows and run start are free.
• "Max cost per run" is respected: the run writes only the rows it can charge and finishes green with a status message.
• sources is the only array field; everything else is flat.

❓ FAQ

Is it legal to scrape Google Autocomplete and other keyword suggestions?

The Actor reads public autocomplete suggestions — the same short phrases anyone sees while typing in a search box. They are aggregated search phrases, not personal data. Check the terms of each site and your local law for your use case; when in doubt, consult a lawyer.

Does it give search volume?

No. These are keyword suggestions, not volume metrics. Suggestions show what people search; for how much interest over time, use Google Trends Scraper.

Why did I get fewer keywords than Max keywords per seed?

Each autocomplete query returns up to about 10 suggestions (Bing up to about 20), and many repeat. Add more expansions, more sources or depth: 2 to reach a higher number. Some sources return nothing for some queries — in our largest test Amazon answered empty for 228 of 310 queries built from questions, prepositions and depth 2. That is normal and free.

How much does a typical run cost?

On the Starter plan the default run returns 50 keywords for about $0.0055, 1,000 keywords cost $0.11 and 10,000 keywords $1.10. On the Free plan the same runs cost $0.00625, $0.125 and $1.25. There is no start fee and no proxy fee.

Can I use Keyword Suggestions Scraper with the Apify API?

Yes. Start runs, read datasets and schedule jobs with the Apify API, or the Python and JavaScript clients shown above.

Can I use it through an MCP server?

Yes. Add the Apify MCP server (https://mcp.apify.com) to Claude, Cursor or any MCP client and call lergassy/keyword-suggestions-scraper as a tool.

Can I integrate it with other apps?

Yes — n8n, Make, Zapier, Google Sheets, Airbyte, Slack and webhooks through Apify integrations.

Is each keyword one row?

Yes. One row per unique keyword per run. A phrase suggested by several sources is one row with all of them in sources, and you pay for it once.

Which question words and prepositions are used?

Questions: how, what, why, when, where, who, which, can, is, are, does, do, will, should. Prepositions: for, with, without, near, vs, to, like, and, in, on, from, under. Alphabet: a–z. Numbers: 0–9.

Is TikTok localised by country?

No. TikTok's suggestion endpoint answers for the location of the server that asks. Google, YouTube, Bing and DuckDuckGo follow your language and country; Amazon and eBay follow the country.

Your feedback

Found a bug, need another source or a new expansion list? Open an issue on the Issues tab — we read every one. If the Actor saved you time, a review on the Store page helps other people find it.

You might also like

Actor	What it does
Google Trends Scraper	Interest over time, by region, related queries and topics — trends to go with your suggestions
Google News Scraper	News articles for any keyword, topic or site
Google Ads Transparency Scraper	Ads any advertiser runs on Google
YouTube Search Scraper	YouTube search results for your keywords
Wappalyzer Scraper	Technologies behind any website

Also known as: Google Autocomplete scraper, Google Suggest scraper, keyword research tool, keyword generator, YouTube autocomplete scraper, YouTube keyword tool, Amazon keyword tool, Amazon autocomplete scraper, eBay keyword tool, Bing autosuggest, TikTok search suggestions, long-tail keyword generator, alphabet soup keyword tool, AnswerThePublic alternative, keyword suggestion API.