Google Trends Scraper - Interest Over Time, Region & Queries

Get Google Trends data for your keywords in seconds: interest over time, interest by region, and top and rising related queries, one clean row per search term. Compare up to 5 terms on the same 0-100 scale, exactly like the Google Trends comparison view. Built for SEO and content teams, market researchers, dashboards and AI agents that call Google Trends from code or on a schedule.

• Fast. No browser. A one-term run took 3.8-6.7 s of run time in production (256 MB); 30 terms took 54 s.
• Terms Google cannot answer are free, with the reason. A term with too little search volume, a request Google rejects (for example an unknown country code) or a request Google refuses comes back as a free row that says why. If Google refused every term or none could be read (blocked, unreadable or timed out), the run is not charged and ends with an error so your schedule sees it. A run whose terms were only rejected (HTTP 400) or had no data ends normally with those free rows and is not charged either.
• Retries on Google's rate limit. Requests go out directly first; when Google answers HTTP 429 or 403, the request is sent again from another IP address through Apify's US residential proxy (up to 6 attempts per request). Check pages and CAPTCHAs are never solved.
• Compare terms. Turn on compareTerms and the terms share one scale (100 = the highest point of any term in the group), so the values of different terms can be compared directly instead of each term peaking at its own 100.
• Watch mode for schedules. With onlyChanges, later runs return a term only when Google Trends has a new complete data point since the last run; everything else is one free no-change row.
• Extra columns Google does not give you directly: latest value and whether it is a partial (still counting) period, last complete value, change from the previous complete point, average, peak value and date, top region.

Unofficial. Uses the same public, logged-out data the Google Trends website loads. No login and no Google account needed.

What you get

One row per search term. Example (production run 0x7mZVKqy5460YfyQ, 2026-09-24, apify in the US, past 12 months, lists shortened to their first entries):

{
  "status": "ok",
  "changeType": null,
  "searchTerm": "apify",
  "comparedWith": [],
  "geo": "US",
  "timeRange": "today 12-m",
  "timeRangeResolved": "2025-09-24 2026-09-24",
  "resolution": "WEEK",
  "trendsUrl": "https://trends.google.com/trends/explore?date=today%2012-m&geo=US&q=apify&hl=en-US",
  "latestValue": 32,
  "latestDate": "2026-09-20T00:00:00.000Z",
  "latestIsPartial": true,
  "lastCompleteValue": 34,
  "lastCompleteDate": "2026-09-13T00:00:00.000Z",
  "previousCompleteValue": 35,
  "changeFromPrevious": -1,
  "averageValue": 55.9,
  "peakValue": 100,
  "peakDate": "2026-06-07T00:00:00.000Z",
  "topRegion": "Wyoming",
  "topRegionCode": "US-WY",
  "topRegionValue": 100,
  "interestOverTime": [
    { "date": "2025-09-21T00:00:00.000Z", "label": "Sep 21 – 27, 2025", "value": 35, "isPartial": false }
  ],
  "interestByRegion": [
    { "geoCode": "US-WY", "geoName": "Wyoming", "value": 100 },
    { "geoCode": "US-DC", "geoName": "District of Columbia", "value": 35 }
  ],
  "relatedQueriesTop": [
    { "query": "apify scraper", "value": 100 },
    { "query": "api", "value": 59 }
  ],
  "relatedQueriesRising": [
    { "query": "tavily", "growth": "Breakout", "value": 24350 },
    { "query": "composio", "growth": "Breakout", "value": 8750 }
  ],
  "checkedAt": "2026-09-24T09:31:11.257Z"
}

Column	Meaning
status	ok for a result row. Other values are free rows (below)
changeType	Watch mode only: first-check or changed. Null otherwise
searchTerm, comparedWith	The term, and the other terms of its comparison group (empty when not comparing)
geo, timeRange, timeRangeResolved, resolution	What was asked, the exact dates Google used, and the point size (MINUTE, EIGHT_MINUTE, HOUR, DAY, WEEK, MONTH)
category, searchType	Category ID and search type (web, images, news, youtube, shopping)
trendsUrl	The same query on the Google Trends website
latestValue, latestDate, latestIsPartial	The newest point. Google marks the current, still-counting period as partial
lastCompleteValue, lastCompleteDate	The newest point that is not partial
previousCompleteValue, changeFromPrevious	The complete point before it, and the difference
averageValue, peakValue, peakDate	Over all points with data in the range
pointsCount, pointsWithData	Number of points, and how many have data (Google leaves a point empty when volume is too low)
interestOverTime	Every point: date (ISO, UTC), label (Google's label), value (0-100, null = no data), isPartial
interestByRegion, topRegion, topRegionCode, topRegionValue, regionResolution, regionsWithData	Regions with data, highest first: countries for worldwide, states or provinces for a country, metro areas for a US state
relatedQueriesTop, topQueriesCount	Google's top related queries (value 0-100)
relatedQueriesRising, risingQueriesCount	Google's rising related queries with growth as Google shows it (+250%, Breakout)
changes, isFirstCheck, previousCheckAt	Watch mode: newCompletePoints since the last run, plus newRisingQueries and newTopQueries not seen in earlier runs of this watch
note	Free rows only: why nothing (or not everything) was returned
checkedAt	Time of the check

Values are Google Trends' relative interest (0-100), not search volumes. Related topics are not included: on 2026-09-24 Google's endpoint for them returned an empty list for every term we tried outside the website.

Free rows (not charged)

status	When
no-data	Google Trends returned no data points for this term with these settings (too little search volume)
no-change	Watch mode: no new complete data point since the last run
partial	The interest over time was read, but interest by region or related queries could not be read. The row is returned with those columns null, free of charge, and watch mode does not remember it
rejected	Google answered HTTP 400: usually an unknown geo code, a date range Google does not accept, or a category number that does not exist
blocked	Google refused every attempt, also from other IP addresses
unreadable	Google's answer could not be read, or Google showed a check page (never solved, not retried)
bad-input	The input could not be used; nothing was requested
timed-out	The run was about to reach its timeout, so it stopped asking Google Trends and returned the rows it already had. Run fewer terms per run or raise the timeout
budget-reached	The run hit the maximum total charge you set. Terms not returned are not charged, and in watch mode not remembered, so the next run returns them

Input

Field	Type	Default	Description
searchTerms	array of strings	coffee	Keywords, up to 100 per run. Duplicates (ignoring case) are looked up once
compareTerms	boolean	false	Compare the terms together in groups of up to 5 on one scale
timeRange	string	today 12-m	now 1-H, now 4-H, now 1-d, now 7-d, today 1-m, today 3-m, today 12-m, today 5-y, all
customTimeRange	string	(none)	2024-01-01 2024-12-31 style dates; overrides timeRange
geo	string	worldwide	Country (US, GB, JP) or region (US-CA, GB-ENG) code
category	integer	0	Google Trends category ID (the cat= number in a Google Trends URL); 0 = all
searchType	string	web	web, images, news, youtube, shopping
includeInterestByRegion	boolean	true	Add interest by region
includeRelatedQueries	boolean	true	Add top and rising related queries
onlyChanges	boolean	false	Watch mode (below)
watchName	string	(none)	Separate memories for two schedules that watch the same terms. A watch remembers each term together with its comparison group, geo, time range, category and search type: changing any of them starts a new first check
resetMonitoringState	boolean	false	Forget this watch's memory at the start of the run

Example:

{ "searchTerms": ["iphone", "samsung galaxy", "google pixel"], "compareTerms": true, "geo": "US", "timeRange": "today 12-m" }

How close is it to the Google Trends website?

We opened the same 4 queries (bitcoin, iphone, taylor swift, python; US; past 12 months) in a browser on 2026-09-24 and compared them with a production run of this Actor:

• Interest over time: the peak week was the same week in all 4, the shape correlated at 0.990-0.998, and single weeks differed by 0.8-2.1 points on average (at most 7). Google serves requests that do not come from its website from a separate sample, so a few points of difference are expected; the same request repeated from different IP addresses returned identical numbers (53 of 53 weeks).
• Top region: the region this Actor reports first appeared on the page in all 4.
• Rising related queries: Google samples them, so they change between requests (0-3 of the first 5 were also on the page). Use them as ideas, not as a stable ranking.

Watch mode

1. Run once with onlyChanges: true and your terms: every term comes back with changeType: "first-check".
2. Create a Schedule with the same input (daily for today 12-m or today 1-m, hourly for now 7-d).
3. Later runs return a term only when Google Trends has a new complete data point since the last run (a new week for the 12-month range, a new day for 30 days, a new hour for 7 days). The row lists the new points in changes.newCompletePoints, and related queries this watch has not seen before in changes.newRisingQueries and changes.newTopQueries.

Related queries and regions alone never count as a change: Google samples them, and in production two requests 11 seconds apart returned the same interest over time but four different rising queries. A term without a new point comes back as one free no-change row.

Pricing

Pay per event:

• Run start - charged once per run, only when at least one term was answered (a result row or a watch-mode no-change). Not charged when every term is no-data, partial, rejected, blocked, unreadable or timed-out, or when the run's maximum total charge has no room for the start fee plus one result row (then nothing is requested at all).
• Search term result - one per ok row. Free rows are never charged.

The exact prices are shown on the Pricing tab.

Measured in production (2026-09-24, build 0.1, 256 MB)

Input	Result	Run time
{} (coffee, worldwide, 12 months)	1 row, 53 weekly points	6.7 s
apify, US	1 row	4.0 s
iphone, samsung galaxy, google pixel, compared, US	3 rows on one scale	3.8 s
taylor swift, travis kelce, now 7-d, watch mode	2 first-check rows	6.4 s
30 popular terms, one run	30 rows, 121 requests, no retry needed	54 s
coffee with geo ZZ	1 free rejected row	2.0 s

From inside Apify, direct requests, datacenter proxy and US residential proxy each returned the full data 6 of 6 times (2026-09-24).

Support

Found a term that returns something different from the Google Trends website, or need another field? Open an issue on the Issues tab with the input you used.