Google Trends Scraper - Interest, Regions & Rising Queries

Extract Google Trends data for any keyword without an API key. This scraper returns interest over time, interest by region, and top and rising related queries as clean, flat rows you can drop straight into a spreadsheet, a database, or a BI tool.

Google does not offer a public Trends API. This Actor gives you one.

What this Actor does

• Extract Google Trends interest over time for any search term, from the past hour to 2004
• Get interest by region with country, state, or metro-level breakdowns
• Find rising search queries before they peak - the single most useful signal in Trends
• Track related topics to map out an entire subject area
• Compare keyword popularity across countries by running the same term with different regions
• Export Google Trends data to CSV, JSON, or Excel in one click
• Monitor keyword seasonality by scheduling daily or weekly runs

Why this Actor is reliable

Most Google Trends tools break, and they all break the same way.

Google's Trends endpoints return 429 Too Many Requests for any call made without a warmed session cookie. The widely used pytrends library was archived in April 2025 and no longer performs that step, so anything built on it fails on the very first request. On top of that, Google enforces its limit per IP address, not per session: a single address survives roughly 10-15 requests before it is cut off no matter what cookies it presents.

This Actor handles both problems directly. It warms a real session before every batch of calls, and it rotates to a fresh residential IP for each keyword, retrying on a new address when Google pushes back. That is why it keeps working where simpler scrapers stall.

Input

Field	Type	Default	Description
keywords	array	["bitcoin"]	Search terms to analyze. Up to 500 per run.
geo	string	"" (worldwide)	Two-letter country code, e.g. US, GB, DE, KZ.
timeframe	string	today 12-m	Period to analyze, from now 1-H to all.
dataTypes	array	["timeline","geo","relatedQueries"]	Which datasets to return.
category	integer	0	Google Trends category ID. 0 means all categories.
language	string	en-US	Interface language for returned labels.
proxyConfiguration	object	Residential	Residential proxies are strongly recommended.

The Actor runs with sensible defaults. Press Start without changing anything and you will get 12 months of worldwide data for bitcoin.

Example input

{
  "keywords": ["electric car", "hybrid car", "hydrogen car"],
  "geo": "US",
  "timeframe": "today 12-m",
  "dataTypes": ["timeline", "geo", "relatedQueries"]
}

Output

Every row is flat and self-describing, so you never have to unpack nested JSON. The type field tells you which dataset the row belongs to.

Interest over time

{
  "keyword": "electric car",
  "geo": "US",
  "timeframe": "today 12-m",
  "type": "timeline",
  "date": "Aug 4 - Aug 10, 2025",
  "timestamp": 1754265600,
  "value": 73,
  "isPartial": false
}

Interest by region

{
  "keyword": "electric car",
  "geo": "US",
  "timeframe": "today 12-m",
  "type": "geo",
  "region": "California",
  "regionCode": "US-CA",
  "value": 100
}

Related queries

{
  "keyword": "electric car",
  "geo": "US",
  "timeframe": "today 12-m",
  "type": "related_query",
  "rank": "rising",
  "query": "cheapest electric car 2026",
  "value": null,
  "formattedValue": "+250%"
}

Rows with "rank": "top" are the most searched related terms overall. Rows with "rank": "rising" are the fastest growing ones, which is usually where the useful signal is.

Use cases

SEO keyword research. Pull rising related queries for a seed term and you get a list of searches gaining traction right now. Feed those into your content calendar before the competition notices them.

Product and market research. Compare interest for competing products across regions to decide where to launch, or confirm whether a category is genuinely growing before committing budget.

Seasonality planning. Run a five-year timeframe to see when demand for a product peaks, then schedule inventory and ad spend against the real curve instead of guesswork.

Content and media planning. Track which topics are accelerating in a specific country and time publication to match.

Feeding AI agents and RAG pipelines. Trends data is one of the few genuinely fresh public signals about human attention. This Actor returns it in a shape that is trivial to embed or pass to a model as context.

Pricing

This Actor uses pay per result. You are charged per row returned, so a run that fails costs you nothing. Platform costs such as proxy traffic are included in the per-result price - there are no separate compute or bandwidth charges to reason about.

A typical keyword with all three data types returns roughly 60-90 rows: about 52 weekly timeline points for a 12-month range, plus regional rows and up to 50 related queries.

Tips for best results

• Keep the keyword list focused. Google returns relative interest, scaled 0-100 within each request. Values are only comparable within a single keyword's own results.
• Use residential proxies. They are the default for a reason. Datacenter IPs are recognised and throttled by Google almost immediately.
• Short timeframes give finer granularity. now 7-d returns hourly points; today 12-m returns weekly ones; all returns monthly.
• Rare keywords legitimately return nothing. If Google has too little search volume for a term it reports no data. That is a real answer, not an error.
• Rising queries are co-searches, not synonyms. This surprises people, so it is worth stating plainly: "rank": "top" returns terms about your keyword (for yoga mat: black yoga mat, yoga mat bag, travel yoga mat), while "rank": "rising" returns whatever is growing fastest among people who also searched it. During a sale season that can include unrelated products such as robot vacuum. This is genuine Google behaviour, not a data error - top is what you want for keyword research, rising is what you want for spotting momentum. A formattedValue of Breakout means growth above 5000%.
• Narrow with category when a word is ambiguous. "Java" as a programming language and "Java" as an island produce very different curves.

Frequently asked questions

Do I need a Google account or API key? No. The Actor requires no credentials of any kind.

How many keywords can I process in one run? Up to 500. Each keyword is fetched independently on its own rotated IP, so a failure on one does not affect the others.

Why do some keywords fail? Google rate-limits aggressively. The Actor retries each keyword several times on fresh IP addresses, but under heavy load a term can still fall through. Failed keywords are listed in the run log and in the RUN_SUMMARY record, and you are not charged for rows that were never returned.

Can I schedule this to run automatically? Yes. Use Apify Schedules to run it hourly, daily, or weekly and append to the same dataset to build a time series of your own.

Is the data the same as the Google Trends website? Yes. It comes from the same endpoints the website itself calls, so the numbers match what you see in the browser.

Can an AI agent call this Actor directly? Yes. It supports pay-per-event billing and runs with limited permissions, so it is callable through the Apify MCP server and payable via x402 without an Apify account.

Integrations

Connect the output to Google Sheets, Airtable, Slack, Zapier, Make, or any HTTP webhook using Apify's built-in integrations, or pull results straight from the Apify API in your own code.