Google Trends Scraper & Keyword Ideas

Extract Google Trends data at scale: one rich row per keyword with interest over time, average, peak and its date, interest by country or region, US metro areas (DMA), top and rising related queries, plus the full Trending Now list of any country and autocomplete keyword ideas. Compare more than 5 terms on one scale. Keywords without data are delivered free — you only pay for rows with data.

No login, no browser, no cookies of yours. Works for any country, time range, category and Google property (web, images, news, YouTube, Shopping).

What you get

Output type	One row per	Main fields
keyword-trend	keyword	timeline (0-100 per date), average, peak, peakDate, latestValue, regions, dmas, cities, relatedTop, relatedRising, hasData
comparison-set	comparison (up to 25 terms)	averages and timeline.values on one shared scale, regionShares, relatedTopByTerm, relatedRisingByTerm, anchor, normalizedBy
trending-item	trending search	title, volume and volumeLabel ("2M+"), growthPct, startedAt, endedAt, isActive, relatedQueries, categoryIds, articleCount
keyword-idea	unique autocomplete idea	idea, seed, query, modifier, modifierType, position, containsSeed, language, country
ideas-then-trends	idea of each seed	a full keyword-trend row for each autocomplete idea, with seed and idea

One row type per run, so every export (JSON, CSV, Excel) has clean, consistent columns. Every field is always present (null when empty).

Why this Actor

Competitors: public stats, pricing and output schema read through the Apify API on 13 Oct 2026 (no third-party Actor was run, and none publishes a sample of its data, so for them we show the most their own output schema allows). This Actor: 50 private test runs on Apify, 13–14 Oct 2026.

	This Actor	Most-used Google Trends Actor	Typical alternatives
Runs that succeed	50 of 50 test runs (5 output types, 14 countries and worldwide, 10 time ranges, 5 search types, invalid inputs)	56.6 % of 22,991 runs in 30 days (4,965 timed out)	92.8–99.5 %
Data per keyword	1 row: series + average + peak + regions + metros + top & rising queries (median 53 points, 51 regions, 25 top and 15 rising queries; n = 518 rows)	1 item, no output schema	~150 thin rows per keyword (one per point, region or query)
Filled fields per row	21.9 on average (518 keyword rows); 16.8 per trending row (4,357); 12.0 per idea row (2,630)	not declared	at most 12 per trend row and 13 per trending row by their own schema; 9 per autocomplete row
Charged = rows with data	7,518 of 7,518 in 50 runs, 0 duplicates	every row, with or without data	every row
Price for a keyword with data	$0.0028	$0.003	$0.005 per thin row, or $0.002 + $0.025 per run
Price for a keyword without data	$0	$0.003	charged
Start fee	$0.00001 (platform minimum)	none	up to $0.025–$0.09 per run
Keywords per run	up to 500	many	often 1
Compare more than 5 terms (Google's limit is 5)	yes, merged through a shared anchor	not in its input	not in their input
Full Trending Now	yes, every item (487 in US / 24 h; 3,172 in US / 7 days)	no	some
Autocomplete ideas + trends in one run	yes	no	no
Output schema (Output tab)	yes	no	rarely

Measured in testing (13 Oct 2026): 320 keywords across 10 locations, 6 time ranges, 4 search types and 3 categories → 98.4 % complete rows; the rest had their time series and regions but not their related queries, and were delivered without charge. From inside Apify with 8 keywords in parallel: 33–58 complete keywords per minute, and after an automatic retry when Google throttled, 100 % of keywords recovered on the primary connection and 91.7 % on the backup one (24 keywords each).

How much data to expect

Google Trends is relative (0-100) and some combinations simply have little data. Measured fill rates on 315 complete rows (13 Oct 2026):

Rows (n)	Has data	Regions	Top queries	Rising queries
Past 12 months, any search type (53)	98.1 %	100 %	98.1 %	96.2 %
Web search, any range (159)	97.5 %	98.7 %	96.2 %	92.5 %
YouTube search (53)	98.1 %	100 %	94.3 %	94.3 %
Image search (52)	88.5 %	82.7 %	71.2 %	67.3 %
News search (51)	76.5 %	62.7 %	35.3 %	29.4 %
Past day, now 1-d (53)	83.0 %	73.6 %	52.8 %	45.3 %

Rows with hasData: false are still delivered (so you know the term was checked) and are never charged. US metro areas: 209 of 210 with data for a popular term. Cities: Google returns city data for few terms (27 of 216 rows had at least one city), so includeCities is off by default.

Input examples

Trend of several keywords in the US (this is the prefilled example):

{
  "outputType": "keyword-trend",
  "keywords": ["seo", "web scraping", "chatgpt"],
  "geo": "US",
  "timeRange": "today 12-m"
}

Compare 7 AI assistants on one scale (first term is the anchor):

{
  "outputType": "comparison-set",
  "keywords": ["chatgpt", "gemini", "perplexity", "copilot", "deepseek", "grok", "meta ai"],
  "geo": "US",
  "timeRange": "today 3-m"
}

Everything trending in Mexico and Spain in the past 24 hours:

{ "outputType": "trending-item", "trendingGeos": ["MX", "ES"], "trendingHours": "24" }

Keyword ideas, then the trend of the 10 best ideas of each seed:

{ "outputType": "ideas-then-trends", "seeds": ["crm software"], "maxIdeasPerSeed": 10, "geo": "US" }

An empty input returns the worldwide trend of "seo" for the past 12 months in a few seconds.

Output example (keyword-trend, shortened)

{
  "outputType": "keyword-trend",
  "keyword": "seo",
  "geo": "US",
  "timeRange": "today 12-m",
  "hasData": true,
  "points": 53,
  "average": 57.9,
  "peak": 100,
  "peakDate": "2026-07-05",
  "latestValue": 32,
  "timeline": [{ "date": "2025-10-12", "time": 1760227200, "value": 38, "isPartial": false }],
  "regionResolution": "REGION",
  "regions": [{ "geoCode": "US-WY", "geoName": "Wyoming", "value": 100, "hasData": true }],
  "dmas": [{ "geoCode": "759", "geoName": "Cheyenne WY-Scottsbluff NE", "value": 100 }],
  "relatedTop": [{ "query": "seo company", "value": 100, "formattedValue": "100" }],
  "relatedRising": [{ "query": "regalseo", "value": 36700, "formattedValue": "Breakout", "isBreakout": true }],
  "sampled": true,
  "error": null
}

Pricing

Pay per event — you pay for rows with data, never for empty, partial or failed rows.

Event	Price (free plan)	Gold plan
Keyword trend row (keyword-trend, also each idea in ideas-then-trends)	$0.0028 ($2.80 per 1,000)	$0.0008
Comparison batch (comparison-set, per batch of up to 5 terms)	$0.004	$0.002
Trending search row (trending-item)	$0.0005	$0.0002
Keyword idea row (keyword-idea)	$0.0003	$0.0001
Actor start	$0.00001	$0.00001

maxItems is a hard cap on charged rows, and the run stops cleanly when your maximum cost per run is reached. Duplicates are removed within a run and never charged twice.

Good to know

• Google Trends samples searches. A value can differ by a few points between two runs (measured: up to 3 points out of 100). Values within one row are consistent.
• Related topics are not included. Google returns them empty to every automated client; this Actor does not pretend otherwise.
• Trending articles: you get how many news articles Google links to each trend (articleCount), not their titles or URLs.
• No absolute search volumes. Trends is relative; the trending list gives volume buckets ("200K+"), exactly as Google shows them.
• When Google throttles, the Actor switches connection, retries later and, if a row still cannot be fetched, delivers an error row that explains why — never charged. All error rows are also listed in the ERRORS record of the run.
• Invalid input (an unknown country code, a bad date range) ends the run in green with an error row that says what to fix, and nothing is charged.

FAQ

Can I use topics instead of search terms? Yes: put a Knowledge Graph id such as /m/019qb_ in keywords.

Which time ranges are supported? Past hour, 4 hours, day, 7 days, 30 days, 90 days, 12 months, 5 years, since 2004, or custom dates (customTimeRange: "2025-01-01 2025-06-30").

How do I filter trending searches by category? Put Google Trends category ids in trendingCategories. Google's own trending list ignores the category, so the Actor filters it for you, and filtered-out items are not charged.

Why are some keywords marked partial? Their time series arrived but an optional block (regions or related queries) did not after every retry. Partial rows are delivered free.