Google Trends Scraper: Interest Over Time, Regions & Related

Get Google Trends data as clean, flat rows: interest over time, interest by region (country, state, metro or city), related queries (top and rising), comparisons of up to 5 terms on one scale, and the searches trending now in any country. No browser, no login, no API key.

Built for reliability. Google Trends rate-limits scrapers hard (HTTP 429), and that is why Google Trends runs so often time out or fail. This Actor is designed around it: it completes Google's cookie handshake, paces its requests, retries every throttled request on a new residential IP, keeps what it already has when one data type fails, and never fails the whole run because of one term. Every problem is explained per term in the run summary.

Price: $0.30 per 1,000 rows ($0.0003 per data point), Apify platform and proxy costs included. You pay only for rows you get.

What you get

type	One row is	Main fields
interest_over_time	One time point of one term (minute, hour, day, week or month)	date, value (0-100), isPartial
interest_by_region	One country, state, metro area or city for one term	regionCode, regionName, value
related_query	One related search, top or rising	query, rankType, value, formattedValue ("Breakout", "+250%")
related_topic	One related topic, top or rising. Optional, off by default: Google rarely sends topics to automated requests (see limits)	query (topic title), topicType, topicId
trending_search	One search trending now in a country	query, search volume, growth %, start time, categories, related searches

Every row also has the term, all comparedTerms of the request, geo, timeRange, category, property and a trendsUrl that opens the same view on Google Trends, so the numbers can be checked and cited.

• Any location: worldwide, any country (US, GB, DE, IN, ...), regions (US-CA) and US metro areas.
• Any time range: past hour, 4 hours, day, 7 days, 30 days, 90 days, 12 months, 5 years, 2004 - present, or a custom date range.
• Categories and search properties: Web, News, Images, YouTube and Google Shopping search, any Google Trends category.
• Comparisons: up to 5 terms on one shared 0-100 scale, exactly like the Compare button.
• Google Trends URLs work as input: paste an explore URL and its settings are used.
• Topic IDs (e.g. /m/02vqfm) work as search terms.

Why this Actor is reliable

• New IP on every 429. Each worker (up to 3 run in parallel) has its own proxy session, i.e. its own residential IP. When Google throttles a request, it is retried on a fresh IP within seconds, up to 6 times. The only longer wait (at most about a minute) happens when Google first has to accept a brand-new cookie.
• Cookie handling that Google accepts. Google answers the first request of a new visitor with HTTP 429 and a cookie; the Actor completes that handshake automatically. On IPs that Google watches, a brand-new cookie is refused for about a minute, so the Actor keeps cookies when it changes IPs and reuses a cookie Google already accepted instead of burning retries.
• Polite pacing. One request at a time per IP with a minimum gap, and a new IP after every 30 requests, so no single IP sends a burst.
• Partial results instead of failures. If, say, the region map of a term fails after all retries, you still get its interest over time and related queries; the summary says what is missing and why.
• One bad term never stops the run. Invalid input lines are reported and skipped. A run is marked failed only when nothing could be fetched because of a problem on our side (for example Google blocking every request); input mistakes end as succeeded, with the reason in the status message and the summary.
• Expired tokens are refreshed. Google's per-request tokens are renewed automatically if they expire during retries.
• Survives platform migrations. Progress is saved after every query. If Apify moves the run to another server, it resumes where it stopped: finished queries are neither fetched nor charged again.
• Clean stop at your spending limit. With a maximum cost per run (or maxItems), the Actor stops exactly at the limit: it sends no further requests (not even retries) and skips the remaining queries instead of fetching data you won't pay for. A query cut by the limit is marked partial, one with no rows saved skipped.
• Fast health checks. A typical query takes 4 requests and a few seconds; run time is capped so a run never runs into its timeout while waiting for Google.

How to use

1. Add Search terms (one per line). Each term gets its own 0-100 scale.
2. To compare terms on one scale, add Comparison groups, e.g. coffee, tea, matcha (up to 5 terms per line).
3. Set Location (empty = worldwide), Time range, and optionally Category and Search property.
4. Choose the data: interest over time, interest by region (and its resolution), related queries (related topics are optional and usually empty), and whether you want top, rising or both.
5. Optional: turn on Trending now for a country.
6. Run it and download the dataset as JSON, CSV or Excel, or read it via API. The SUMMARY record in the run's key-value store lists every query with its status, rows per type, errors and notes.

Input example

The input of a real test run:

{
  "searchTerms": ["coffee", "pickleball", "https://trends.google.com/trends/explore?q=bitcoin&geo=GB&date=now%207-d"],
  "comparisonGroups": ["chatgpt, gemini, claude"],
  "geo": "US",
  "timeRange": "today 12-m",
  "includeTrendingNow": true,
  "trendingGeo": "US"
}

It returned 1,010 rows: 50 trending searches; 154 rows each for coffee and pickleball (53 weeks, 51 states, 50 related queries); 207 rows for bitcoin in the UK over 7 days (hourly, from the URL's own settings); and 445 rows for the three-way comparison.

Other options: "timeRange": "custom", "customTimeRange": "2024-01-01 2024-12-31", "property": "youtube", "category": 71, "regionResolution": "CITY", "includeLowVolumeRegions": true, "relatedRanking": "rising", "includeRelatedTopics": true, "trendingTimeRange": "4", "trendingMaxItems": 100, "maxItems": 500, "language": "de".

Output examples

Real rows from the run above. The first one is complete; in the others, fields that are null for that type are left out to keep this page short (the dataset always has every field).

Interest over time:

{
  "type": "interest_over_time",
  "term": "coffee",
  "comparedTerms": ["coffee"],
  "geo": "US",
  "timeRange": "today 12-m",
  "category": 0,
  "property": "web",
  "value": 60,
  "formattedValue": "60",
  "hasData": true,
  "date": "2025-10-12",
  "formattedDate": "Oct 12 – 18, 2025",
  "timestamp": 1760227200,
  "resolution": "WEEK",
  "isPartial": false,
  "regionCode": null,
  "regionName": null,
  "coordinates": null,
  "query": null,
  "topicType": null,
  "topicId": null,
  "rankType": null,
  "rank": null,
  "increasePercent": null,
  "endedAt": null,
  "isActive": null,
  "categories": null,
  "relatedQueries": null,
  "trendsUrl": "https://trends.google.com/trends/explore?date=today%2012-m&geo=US&q=coffee&hl=en-US",
  "scrapedAt": "2026-10-17T13:30:48Z"
}

Interest by region in a comparison (each term's share of the region, the three terms add up to 100%):

{
  "type": "interest_by_region",
  "term": "chatgpt",
  "comparedTerms": ["chatgpt", "gemini", "claude"],
  "geo": "US",
  "timeRange": "today 12-m",
  "value": 58,
  "formattedValue": "58%",
  "hasData": true,
  "resolution": "REGION",
  "regionCode": "US-CA",
  "regionName": "California",
  "trendsUrl": "https://trends.google.com/trends/explore?date=today%2012-m&geo=US&q=chatgpt,gemini,claude&hl=en-US"
}

Related query, rising:

{
  "type": "related_query",
  "term": "chatgpt",
  "comparedTerms": ["chatgpt", "gemini", "claude"],
  "geo": "US",
  "timeRange": "today 12-m",
  "query": "chatgpt caricature",
  "rankType": "rising",
  "rank": 1,
  "value": 3200,
  "formattedValue": "+3,200%"
}

Trending now (from a cloud run on 18 Oct 2026, related searches shortened):

{
  "type": "trending_search",
  "geo": "US",
  "timeRange": "past 24 hours",
  "query": "texans vs jaguars",
  "rank": 1,
  "value": 500000,
  "formattedValue": "500K+",
  "increasePercent": 1000,
  "date": "2026-10-18T01:50:00Z",
  "isActive": true,
  "endedAt": null,
  "categories": ["Sports"],
  "relatedQueries": ["texans vs jaguars", "houston texans", "where to watch texans vs jacksonville jaguars", "wembley stadium", "nfl london games 2026"],
  "trendsUrl": "https://trends.google.com/trending?geo=US&hours=24&hl=en-US"
}

Related topic rows (when switched on and Google sends them) have query (topic title), topicType (e.g. "Beverage") and topicId (e.g. /m/02vqfm, usable as a search term). City rows have regionName and coordinates ({"lat": ..., "lng": ...}) instead of a regionCode.

Output fields

Field	Description
type	interest_over_time, interest_by_region, related_query, related_topic or trending_search
term, comparedTerms	The term of this row, and all terms of the same request (their values share one scale)
geo, timeRange, category, property	Request settings (geo "" = worldwide; property web, news, images, youtube or shopping)
value, formattedValue	The number and how Google shows it (see "How to read the values")
hasData	false when Google has not enough data for that time point (value 0). Region rows always have data: regions without data are left out
date, formattedDate, timestamp	Start of the period in UTC (YYYY-MM-DD, or date-time for hourly / minute data), Google's label, Unix time. Trending: start of the trend
resolution	Time resolution (MINUTE, EIGHT_MINUTE, HOUR, DAY, WEEK, MONTH) or region resolution (COUNTRY, REGION, DMA, CITY)
isPartial	true for the current, still incomplete period
regionCode, regionName, coordinates	Region of a region row (cities have coordinates instead of a code)
query, topicType, topicId	Related query, related topic (title, type, ID) or trending search
rankType, rank	top or rising, and the position in the list (trending: position in Google's list)
increasePercent, endedAt, isActive, categories, relatedQueries	Trending searches: growth, end time (null while active), Google's categories, searches that belong to the trend
trendsUrl	Opens the same view on Google Trends
scrapedAt	When the run started (UTC)

The dataset has the views All rows, Trending now and All fields. To work with one data type, filter by type.

How to read the values

• Google Trends never shows search counts. Interest over time is relative: 100 is the peak for that request, 50 is half as popular, 0 means not enough data.
• Values of one request share a scale. Compare terms inside one comparison group; two separate runs (or two separate terms) are each scaled to their own peak. To compare more than 5 terms, use several groups that share one anchor term.
• Interest by region: single term, 100 is the region where the term has the largest share of all searches. In comparison groups, values are each term's share of the region (they add up to 100%).
• Related queries / topics: top lists are 0-100 relative to the most popular one; rising lists show the growth in percent, and "Breakout" means growth above 5,000%.
• Trending now: value is Google's approximate search volume (e.g. 200,000 shown as "200K+"), increasePercent the growth.
• The last point of a time series is usually the current period (isPartial: true); its value rises until the period ends.

Pricing

Pay per event, no subscription and no extra usage fees:

Event	Price
result: one row in the dataset (one data point of any type)	$0.0003 ($0.30 per 1,000 rows)
Actor start (standard Apify event, per run)	$0.00005

Cost examples (row counts from our test runs):

Request	Rows	Cost
1 term, past 12 months, interest over time only	53-54 (weekly points incl. the current week)	$0.016
1 term, past 12 months, US, all default data types	about 155 (54 weeks, 51 states, 50 related queries)	$0.047
1 term, past 30 days, interest over time only	31 (daily)	$0.009
3 terms compared, 12 months, US, all data types	about 445	$0.13
Trending now, one country, 50 trends	50	$0.015
10 terms, past 5 years, interest over time only	2,610 (261 weeks each)	$0.78

To pay less, switch off the data types you don't need, choose top or rising only, or set Max rows.

What is never charged: regions where Google has no data (also with "include low search volume regions" on), a term whose whole time series has no data (a query with no data at all ends as no_data and costs nothing), empty related lists, failed or skipped queries and the SUMMARY record. If Google has data for a term at even one time point, its whole series is returned and charged, including points with value 0 and hasData: false; the summary notes how many points have data. Set Maximum cost per run in the run options (or maxTotalChargeUsd via API); the Actor stops cleanly at that amount and keeps everything fetched so far.

Use it from the API, an app or an AI agent

Run synchronously and get the rows in one call (keep your own API token in the APIFY_TOKEN environment variable):

curl -X POST "https://api.apify.com/v2/acts/fairfetch~google-trends-scraper/run-sync-get-dataset-items?maxTotalChargeUsd=0.50" \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"comparisonGroups": ["chatgpt, gemini, claude"], "geo": "US", "timeRange": "today 3-m", "includeInterestByRegion": false, "includeRelatedQueries": false}'

The Actor uses pay-per-event pricing without usage fees and limited permissions, so AI agents can call it through the Apify MCP server and pay per run. Tips for agents: put terms that must be compared in one comparison group, filter rows by type, check SUMMARY.perQuery[].status (ok, partial, no_data, failed, skipped) and rowsPushed, and cite trendsUrl.

Performance and limits

• About 4 requests per term (5 with related topics, 1 for Trending now) and 5-15 KB of traffic. With the default proxy a query needs roughly 5-10 seconds (requests about one second apart); up to 3 queries run in parallel, each on its own IP.
• Google Trends compares at most 5 terms per request; bigger groups are reported as invalid input.
• Region resolution: countries for worldwide; regions, cities and (US only) metro areas inside a country. Not every combination has data, especially for YouTube or News search.
• Related topics are off by default. Google sends them to visitors it treats as regular users, but tags automated requests as scrapers and then returns an empty related-topics list, almost always (it was empty in every one of our cloud runs). Comparisons have no related-topics widget at all. Related queries are not affected. You can still switch topics on: empty lists cost nothing and are noted in the summary.
• Very rare terms, tiny regions and very short time ranges often have no data. When Google has no data at all for a query, it ends as no_data and nothing is charged. A rare term with data at only a few time points still returns (and charges) its whole series.
• At most 1,000 queries (search terms, URLs and comparison groups) per run; any beyond that are listed in the summary and not run.
• Without a proxy (e.g. local runs) all requests share one IP, so the Actor slows down to about 10 requests per minute and waits out throttling.

FAQ

Is there an official Google Trends API? Google has announced an official Trends API, but access has been limited to a small group of testers. This Actor needs no access approval: it reads the same data the Google Trends website shows.

Why do I get 429 errors with other tools? Google Trends throttles quickly, especially cloud servers. This Actor uses residential IPs, completes Google's cookie handshake, paces requests and retries throttled requests on a new IP. You can see how often that happened in SUMMARY (throttledResponses, ipRotations).

Why are the values different from yesterday's run? Google computes Trends from a sample of searches, so values can change slightly between requests, and every request is scaled to its own peak. Compare terms inside one comparison group.

Can I get absolute search volumes? No. Google Trends only publishes relative interest (0-100). Trending now has Google's approximate search volume per trend.

Can I compare more than 5 terms? Not in one request. Use several comparison groups that all contain the same anchor term and rescale them against it.

What do "Breakout" and "+250%" mean? They are rising related queries or topics: "+250%" means searches grew by 250% compared with the previous period; "Breakout" means more than 5,000%.

Which time zone are dates in? UTC. Daily, weekly and monthly points are dated by the start of the period.

What happens if a term fails? The other terms still run. The failed term, the reason (for example "HTTP 429 ... gave up after 7 attempts") and anything that was still fetched are in the SUMMARY record and the status message. Failed terms are not charged.

Do I need a proxy? It's on by default and included in the price (Apify residential proxy). Turning it off is only advised for local test runs.

Legal

This Actor reads aggregated, anonymous statistics that Google Trends shows publicly to every visitor. It does not log in, does not bypass any login or paywall and collects no data about Google users. Search terms that Google lists (related queries, trending searches) are free text and can contain names of people in the news; handle them with care. Google and Google Trends are trademarks of Google LLC; this Actor is independent and not affiliated with or endorsed by Google. You are responsible for how you use the data and for complying with Google's terms and the laws that apply to you.