Google Trends Scraper (fast, no browser)

Get Google Trends interest over time, interest by region and top related queries as tidy rows, for $0.001 per search term.

The Actor calls the same JSON endpoints the Google Trends website uses, so there's no browser to start. One run holds up to 500 comparisons of up to 5 terms each, in any geo, timeframe, category and Google property (web, news, images, YouTube, Shopping). On Apify, the prefilled input (coffee, tea in the US, past 12 months) saved 258 rows in 7 s in one run and in 24 s in another.

What you get

Every row goes to the default dataset. The type field tells you what the row is, and there are no nested tables, so the data drops straight into CSV, Excel, Google Sheets, a BI tool or an AI agent.

type	One row per	Main fields
interest_over_time	term and time point	date, timestamp, resolution, formattedTime, value (0-100), hasData, isPartial
interest_by_region	term and region	regionCode, regionName, regionResolution, value (0-100), comparisonShare, latitude, longitude
related_query	related query	query, rankType (top or rising), rank, value, formattedValue, isBreakout, mentionsTerm, link
related_topic	related topic	topicId, topicTitle, topicType, rankType, rank, value, formattedValue, isBreakout, mentionsTerm, link

Every row also carries these fields.

Field	Type	Meaning
type	string	Row type, see the table above
comparisonId	string	c1, c2, ... in input order
comparisonTerms	array	All terms of the comparison this row belongs to
term, termIndex	string, number	The term of this row and its position in the comparison
geo	string	US, US-CA, worldwide, ...
timeframe	string	today 12-m or a custom range such as 2024-01-01 2024-12-31
category, categoryName	number, string	Google Trends category id and name (0 = All categories)
property	string	web, news, images, youtube or froogle (Google Shopping)
scrapedAt	string	ISO time the comparison was fetched

A few fields need a word of explanation.

• value runs from 0 to 100. Over time it's relative to the highest point among all terms of the comparison. By region it's relative to the top region.
• comparisonShare has a value only in comparisons of 2 or more terms, and is null for a single term. It's the term's share, in percent, of the compared terms' searches in that region.
• resolution is the time step Google picked (HOUR, DAY, WEEK, MONTH and so on). isPartial is true for the last, still incomplete point.
• formattedValue on rising rows shows Google's label, such as +250% or Breakout. isBreakout is true for Breakout, which means growth above 5,000%.
• mentionsTerm is true when the related query contains every word of the term that has 3 or more letters (a plural "s" or "es" is fine). It is null for topic ids. The FAQ has the full rule. Use it to drop unrelated entries (see the FAQ).

The dataset has column views called Overview, Interest over time columns, Interest by region columns and Related queries & topics columns. Apify views pick columns, but they cannot filter rows, so filter on the type column to see one row type.

The key-value store also gets an OUTPUT record, which is the run report. It lists each comparison with its status (ok, partial, no-data, failed, stopped or skipped), row counts by type, termResultsCharged, notes, errors and request statistics.

Example output

These rows come from a real run of the prefilled input on Apify (2026-11-05, coffee, tea, US, past 12 months, Apify Proxy). That run saved 106 interest-over-time rows, 102 region rows and 50 related queries.

[
  {
    "type": "interest_over_time",
    "comparisonId": "c1",
    "comparisonTerms": ["coffee", "tea"],
    "term": "coffee",
    "termIndex": 0,
    "geo": "US",
    "timeframe": "today 12-m",
    "category": 0,
    "categoryName": "All categories",
    "property": "web",
    "scrapedAt": "2026-11-05T13:38:17.167Z",
    "date": "2025-11-02",
    "timestamp": "2025-11-02T00:00:00.000Z",
    "resolution": "WEEK",
    "formattedTime": "Nov 2 – 8, 2025",
    "value": 68,
    "hasData": true,
    "isPartial": false
  },
  {
    "type": "interest_by_region",
    "comparisonId": "c1",
    "comparisonTerms": ["coffee", "tea"],
    "term": "coffee",
    "termIndex": 0,
    "geo": "US",
    "timeframe": "today 12-m",
    "category": 0,
    "categoryName": "All categories",
    "property": "web",
    "scrapedAt": "2026-11-05T13:38:17.167Z",
    "regionCode": "US-WY",
    "regionName": "Wyoming",
    "regionResolution": "region",
    "value": 100,
    "comparisonShare": 70,
    "latitude": null,
    "longitude": null
  },
  {
    "type": "related_query",
    "comparisonId": "c1",
    "comparisonTerms": ["coffee", "tea"],
    "term": "tea",
    "termIndex": 1,
    "geo": "US",
    "timeframe": "today 12-m",
    "category": 0,
    "categoryName": "All categories",
    "property": "web",
    "scrapedAt": "2026-11-05T13:38:17.167Z",
    "query": "green tea",
    "rankType": "top",
    "rank": 1,
    "value": 100,
    "formattedValue": "100",
    "isBreakout": false,
    "link": "https://trends.google.com/trends/explore?q=green+tea&date=today+12-m&geo=US",
    "mentionsTerm": true
  }
]

Pricing

You pay per search term with data, not per row. There's one event.

Event	Price	Charged when
Term result	$0.001	Once per search term per comparison that returns at least one row. It covers all outputs of that term: interest over time, by region and related queries.

Rows are free. Terms with no data, comparisons that fail, URLs that can't be read and comparisons skipped after an early stop cost nothing. Platform usage (compute and proxy traffic) is included in the event price, so it isn't billed on top. The Actor needs no browser. Its build's default memory is 256 MB, which is enough. Runs on the Apify Store currently start at 512 MB, and you can lower that in Run options.

Some worked examples.

Input	Term results	Cost
The prefilled input, coffee, tea in the US (258 rows in our run)	2	$0.002
100 single keywords, one per line, in 1 geo	100	$0.10
20 lines of 5 terms each, in 10 geos (200 comparisons)	1,000	$1.00
coffee, tea and bitcoin in US and GB	6	$0.006

The price is the same whether you turn outputs on or off. To cap spend, set the maximum cost per run in Run options (or maxTotalChargeUsd in the API). The Actor then saves only the terms it could charge for and stops at the limit. The Apify SDK can charge one term over the cap ($0.001), and that term is saved.

Quick start in Apify Console

1. Open the Input tab. It's already filled with coffee, tea and geo US.
2. Click Start. The run costs $0.002.
3. Open the Output tab, or export the dataset as CSV, JSON or Excel.

To run your own terms, put one comparison per line. Terms on the same line are compared on one 0-100 scale, like on the Google Trends website.

Use it from the API

This call runs the Actor and returns the dataset rows in one request.

curl -X POST "https://api.apify.com/v2/acts/zovo-one~google-trends-scraper-fast/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "searchTerms": ["coffee, tea", "bitcoin"],
    "geos": ["US", "GB"],
    "timeframe": "today 5-y",
    "relatedQueriesTop": true
  }'

run-sync-get-dataset-items waits up to 300 seconds. For bigger inputs, start the run with POST /v2/acts/zovo-one~google-trends-scraper-fast/runs and read the dataset when it finishes.

Use it from Python and Node.js

Python, with pip install apify-client.

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("zovo-one/google-trends-scraper-fast").call(run_input={
    "searchTerms": ["coffee, tea"],
    "geos": ["US"],
    "timeframe": "today 12-m",
})

for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    if row["type"] == "interest_over_time":
        print(row["date"], row["term"], row["value"])

Node.js, with npm install apify-client.

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('zovo-one/google-trends-scraper-fast').call({
    searchTerms: ['coffee, tea'],
    geos: ['US'],
    timeframe: 'today 12-m',
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items.filter((row) => row.type === 'interest_by_region'));

Use it from AI agents with the Apify MCP server

The Apify MCP server at mcp.apify.com exposes Store Actors as tools for Claude, Cursor, VS Code and other MCP clients. Each client adds servers in its own way. VS Code and Claude Desktop, for example, don't take the JSON below as written. For your client, open the MCP configurator, pick this Actor and copy the setup it gives you. A client that reads an mcpServers block with a url key, such as Cursor, can use this config.

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com?tools=zovo-one/google-trends-scraper-fast"
    }
  }
}

On first connect your browser opens so you can sign in to Apify. You can also send your token in an Authorization: Bearer <APIFY_TOKEN> header instead. An agent that uses the default server tools can find this Actor with search-actors, run it with call-actor and read the rows with get-dataset-items.

Then ask in plain words, for example "Compare Google Trends interest for coffee and tea in the US over the past 5 years and tell me which months peak." The tidy rows and the type field make the result easy for a model to read.

Input reference

Field	What it does
searchTerms	One comparison per line, up to 5 comma-separated terms. Google topic ids such as /m/02vqfm work too.
trendsUrls	Google Trends explore URLs. Each URL is one comparison and keeps its own terms, geo, date, category and property. The fields below don't change URLs.
geos	US, US-CA, GB-ENG, US-CA-807 (US metro) or worldwide. Countries use ISO codes (GB, not UK). Each geo repeats every line, so 3 lines in 2 geos make 6 comparisons.
timeframe	Past hour, 4 hours, day, 7 days, 30 days, 90 days, 12 months, 5 years, or 2004 to now.
customTimeRange	YYYY-MM-DD YYYY-MM-DD, from 2004-01-01 to today. Overrides timeframe.
category, customCategoryId	A top-level category, or any category id from the cat= parameter of a Trends URL.
property	web, news, images, youtube or froogle (Google Shopping).
interestOverTime, interestByRegion, relatedQueriesTop	On by default.
relatedQueriesRising, relatedTopics	Off by default (see the FAQ).
includeRawResponses	Saves Google's raw JSON per comparison in the key-value store (RAW-c1, ...). Free.
language	Interface language (hl), such as en-US, de, es-419. Changes region names, not numbers.
proxyConfiguration	Apify Proxy, on by default (datacenter).
maxConcurrency	1 to 3 comparisons in parallel. Default 1.
maxRetries	1 to 10 retries per request. Default 5.

Some URL formats aren't supported. URLs with different date ranges per term (date=2024-01-01 2024-06-30,2025-01-01 2025-06-30) come from another Google endpoint, so paste one URL per date range. Hourly custom ranges (date=2024-01-01T00 2024-01-07T00) and date=all_2008 don't work either. Such a URL is reported as failed in OUTPUT with the reason, it isn't charged, and the rest of the run goes on.

Limits

These are hard caps in the code. Each run allows 5 terms per comparison, 50 different geos and 500 comparisons (lines times geos, plus URLs). A term can have 100 characters, a search-term line 1,000 and a URL 8,192. Bad input is rejected before any request, for example 6 terms on one line, UK instead of GB or an end date in the future.

Google rate-limits Trends traffic per IP. It often answers a new visitor with HTTP 429 and a cookie. The Actor retries once with that cookie, then moves to a new proxy IP with backoff waits of about 1–2 s, 2–4 s, 4–8 s, 8–16 s, then 15–30 s. In one Apify run Google sent 6 of 13 requests back with HTTP 429, and the Actor still finished in 24 s.

When a request fails on every retry (HTTP 429, 403, 5xx, a network error or a timeout), that comparison gives up after one retry ladder (about 50 s of waiting, or about 3 minutes when every attempt times out after 20 s). Its other outputs are skipped. After 3 such comparisons in a row the run stops with a clear message, so it doesn't burn your timeout. In 3 of 5 local runs of the default input (one direct connection, no proxy), Google answered HTTP 429 on every retry and the run ended after about 1 minute with no data and no charge. If you see that on Apify, select the RESIDENTIAL proxy group and keep maxConcurrency at 1.

Run time grows with the input. Plan for roughly 10 to 25 seconds per comparison at concurrency 1, and more when Google rate-limits. That's about 2 to 5 minutes for 10 comparisons, 15 to 45 minutes for 100, and 1.4 to 3.5 hours for 500 comparisons. Set the run timeout to fit. About 30 seconds before the timeout the Actor stops starting requests, saves what it has and lists the rest as skipped in OUTPUT.

Terms that contain a comma aren't supported, because the comma separates terms. Trending now and daily trending searches aren't in this version. Region data comes at the level Google shows by default (countries for worldwide, states or regions for a country).

FAQ

Data compared with the Google Trends website

Google labels the requests from this Actor USER_TYPE_SCRAPER (you can see it in OUTPUT as googleUserType). In our tests on 2026-11-05, interest over time and by region looked right, with the peaks of "christmas" and "super bowl" in the right weeks over 5 years. We haven't compared the numbers with the website side by side.

Rising queries and related topics are off by default

Most rising queries were unrelated filler for automated requests, 38 of 50 in one coffee, tea run ("top songs this week", "sports scores today"). Related topics came back empty in 7 of 7 tests. You can turn both on, and they add no charge, because you pay per term.

Filtering unrelated related queries

Top related queries are mostly on topic, but some are generic. In the Apify run above, 8 of 50 had mentionsTerm: false, such as "travel" and "jobs" for coffee. mentionsTerm is true when the query contains every word of the term with 3 or more letters, each as a whole word or with a plural "s" or "es". So "teas" matches "tea", but "teams" doesn't. Short terms like AI or c++ must appear as a whole phrase. It's a simple text check, so it also drops some related entries such as "chocolate" for coffee.

Terms or regions with no rows

Google hides regions with too little search volume, and so does this Actor. A term with no data at all gets no rows and isn't charged. The OUTPUT notes tell you which term got nothing. In a comparison of several terms, a series of zeros is kept, because there 0 means "below 1% of the top term's peak".

Weekly ranges can start a few days early

For ranges Google shows in weeks, each point starts on a Sunday, so the first week can begin before your start date. date is the start of each point, in UTC.

Data from runs that stop early

Rows that were already fetched are charged and saved when the run stops for the circuit breaker or the time limit. They aren't saved after the maximum cost per run is reached (they couldn't be charged) or when you abort the run.

Legal

This Actor reads public, aggregated Google Trends data. It doesn't collect personal data. Make sure your use complies with Google's terms and the laws that apply to you.