Google Autocomplete Keyword Scraper
Hundreds of long-tail keywords per seed from Google, YouTube and Shopping autocomplete. From $0.0003 / keyword, pay per result, no browser, no login.
▶ Run it on Apify · Python example · All scrapers
Turn one keyword into hundreds of real search suggestions — exactly what people type in Google, YouTube and Google Shopping. The Actor expands each seed with A–Z, 0–9, question words and prepositions (in the right language), deduplicates everything and returns clean JSON, CSV or Excel. Free keyword research at scale, no browser, no API key.
✅ Google Search, YouTube and Google Shopping autocomplete ✅ A–Z and 0–9 expansion — the classic “alphabet soup” long-tail method ✅ Question keywords (how, what, why… / como, porque… / comment, pourquoi…) in 6 languages — perfect for FAQ and blog content ✅ Prepositions (for, with, vs, near, best, cheap…) for buying-intent keywords ✅ Any country and language, several countries per run ✅ Deduplicated, with position, word count and the pattern that found each suggestion ✅ ~250 unique suggestions per seed in about 5 seconds
🚀 Why this autocomplete scraper is different
| This Actor | Typical autocomplete scrapers | |
|---|---|---|
| Sources | Google + YouTube + Shopping | Google only |
| Localized question words | Yes (en, pt, es, fr, de, it) | English only |
| Deduplicated across expansions | Yes | Duplicates to clean yourself |
| Price | $0.0003 per unique suggestion | Often per request, duplicates included |
| Blocking handling | Retries, new IPs, residential fallback | Fails at scale |
💡 Use cases
- SEO keyword research: long-tail keywords and questions for articles, FAQs and landing pages.
- YouTube SEO: video ideas and titles from what viewers actually search.
- E-commerce & Amazon-style listings: buying-intent keywords from Google Shopping (“… for”, “best …”, “cheap …”).
- PPC: discover negative and long-tail keywords for Google Ads.
- Content planning & AI writing: feed real questions to ChatGPT/Claude to write content people search for.
- Market research: what people associate with a brand or product, country by country.
🛠️ How to use it (3 steps)
- Enter seed keywords (one per line).
- Pick sources (Google, YouTube, Shopping), expansions and countries/language.
- Click Start and export the suggestions as CSV, Excel or JSON.
📥 Input example
{
"keywords": ["air fryer", "pokemon cards"],
"sources": ["google", "youtube"],
"expansions": ["alphabet", "questions", "prepositions"],
"countries": ["us", "gb"],
"language": "en",
"maxSuggestionsPerKeyword": 300
}
📤 Output example (real data)
{
"keyword": "cartas pokemon",
"suggestion": "como jogar cartas pokemon",
"source": "google",
"country": "pt",
"language": "pt",
"pattern": "questions",
"query": "como cartas pokemon",
"position": 1,
"words": 4,
"isQuestion": true
}
⚙️ Input parameters
| Field | Type | Default | Description |
|---|---|---|---|
keywords |
array | ["pokemon cards"] |
One keyword per line. Each one is expanded into hundreds of real search suggestions. |
sources |
array | ["google"] |
Where suggestions come from. |
expansions |
array | ["alphabet"] |
Letters/words added after the keyword to discover long-tail suggestions. Questions are added before it. |
includePrefix |
boolean | false |
Doubles the queries; finds suggestions like ‘best pokemon cards’. |
countries |
array | ["us"] |
2-letter codes: us, gb, de, fr, pt, br, in… Each keyword runs in every country. |
language |
string | "en" |
Language code, e.g. en, pt, es, fr, de, it. Question and preposition words follow it. |
maxSuggestionsPerKeyword |
integer | 300 |
Per keyword, source and country. Controls the cost. |
proxyConfiguration, maxRetries |
object, integer | Apify Proxy, auto | Proxy and retries; the defaults work for most runs. |
📋 Output fields
| Field | Type | Description |
|---|---|---|
keyword |
string | Seed keyword |
suggestion |
string | Autocomplete suggestion |
source |
string | google, youtube or shopping |
country |
string | Country of the suggestions |
language |
string | Language of the suggestions |
pattern |
string | seed, alphabet, numbers, questions, prepositions (-before) |
query |
string | Exact text sent to the autocomplete |
position |
integer | Position in the dropdown (1 = first) |
words |
integer | Number of words (long-tail) |
isQuestion |
boolean | Question keyword |
Export as JSON, CSV, Excel, XML or HTML, or read it through the API.
💰 Pricing — pay per unique suggestion
| What you run | Suggestions | Cost (free plan) |
|---|---|---|
| 1 seed, A–Z, Google | ~250 | ~$0.08 |
| 10 seeds, A–Z + questions, Google | ~3,000 | ~$0.90 |
| Cap at 50 per seed | 50 per seed | $0.015 per seed |
Higher Apify plans pay less: $0.00025 (Bronze), $0.0002 (Silver), $0.00015 (Gold and above). Platform usage and proxies are included. Use maxSuggestionsPerKeyword and a maximum cost per run to control spending.
🔌 API & integrations
Python
from apify_client import ApifyClient # pip install "apify-client>=3"
client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("victorrodi/google-autocomplete-keyword-scraper").call(
run_input={"keywords": ["air fryer"], "expansions": ["questions"]}
)
for row in client.dataset(run.default_dataset_id).iterate_items():
print(row["suggestion"])
cURL
curl -X POST "https://api.apify.com/v2/acts/victorrodi~google-autocomplete-keyword-scraper/run-sync-get-dataset-items" \
-H "Authorization: Bearer $APIFY_TOKEN" -H "Content-Type: application/json" \
-d '{"keywords": ["air fryer"], "sources": ["youtube"], "maxSuggestionsPerKeyword": 50}'
Works with Google Sheets, Make, Zapier, n8n and MCP / AI agents: “Find 50 question keywords about air fryers and group them into blog topics.”
🤖 Use it from AI agents (Claude, ChatGPT, Cursor…)
This Actor is available to AI assistants through the Apify MCP server. Add the server to Claude Desktop, Claude Code, Cursor or any MCP client, then simply ask:
“Give me 200 long-tail keyword ideas for ‘solar panels’ from Google autocomplete, grouped by intent.”
“Which questions do people ask on YouTube about ‘video editing’? Use victorrodi/google-autocomplete-keyword-scraper.”
The clean, flat output is designed to be read directly by LLMs.
🛡️ Reliability
- Requests are paced and retried with a new IP and exponential backoff.
- If the site keeps blocking, the Actor switches to residential IPs automatically.
- One failing item never stops the run, and you are never charged for it.
- Every run writes a
RUN_SUMMARYrecord (results, failures, retries, blocks). - Measured on a real run: 1,597 suggestions for 6/6 keywords with 0 blocks.
More examples and docs for all our Actors: Vicshock/trendpulse-scrapers.
📌 Tips
- Start with a 2-3 word seed (“running shoes”, not “shoes”): Google returns more specific long-tail ideas.
- Use the questions expansion to build FAQ sections and blog outlines that match what people really type.
- Run the same seed in several countries and languages to find local keywords your competitors miss.
- YouTube suggestions are great for video titles and tags; Shopping suggestions reveal product attributes buyers search for.
- Suggestions are deduplicated across expansions, so you only pay once per unique keyword.
- Schedule a monthly run to spot new searches as they appear.
❓ FAQ
Is this the same as Google Keyword Planner? No search volumes — these are the real queries Google suggests, ordered by Google. Combine with a volume tool or Google Trends.
Why do some letters return nothing? Google only suggests queries with enough searches; rare combinations return nothing and are not charged.
Which languages get localized questions? English, Portuguese, Spanish, French, German and Italian; other languages fall back to English question words.
Is it legal? The Actor reads public autocomplete suggestions — no personal data. Respect the terms and laws that apply to you.
💬 Support
Missing a feature? Open an issue in the Issues tab — it is answered fast.