Google Trends API: Interest, Regions, Related Queries

Google Trends API: interest over time, by region, related queries and topics

Why agents choose this

Tested Oct 25, 2026: 39 of 39 real keywords returned data (100%), vs 0 of 40 for apify/google-trends-scraper; values match Google's own numbers within 3 points.

Quick call (one field is enough):

{ "keywords": ["chatgpt", "claude"] }

Optional: geo (e.g. "US"), timeRange ("12m" default). Everything else is advanced.

Test (Oct 25, 2026, 40 keywords incl. 1 nonsense term)	This Actor	apify/google-trends-scraper	data_xplorer/google-trends-fast-scraper
Keywords with a 12-month series	39/39	0/40 (navigation timeouts, aborted after 8.5 min)	0/10 (returned "trending searches", not the keyword)
Nonsense keyword	status: "no_data", not charged	–	–
Time for 20 keywords (all sections)	33 s	–	–

• Accuracy: 249 of 270 weekly points identical to a direct pytrends call for the same 5 keywords (92%); the rest differ by at most 3 points, which is Google's own sampling between calls.
• Every row has status, errorCode, hint, charged, and a normalization note: values are 0–100 relative to the peak in this window, not search volume. Pass several keywords with compare: true to put them on one scale.
• Never guesses: low-volume terms come back as status: "no_data" (NO_DATA) and are not charged, instead of a row of zeros.
• Cost: $0.0025 per keyword with data + $0.001 per run. Failed and no-data keywords are never charged.

Test conditions: our own run on Apify cloud, 40 keywords, worldwide, past 12 months, one afternoon. Competitors may do better on another day. Not an independent audit.

Why agents pick this

• $0.0025 per keyword, with all sections included. Comparing 3 keywords costs $0.0085. Rate-limited or failed keywords are never charged.
• The summary is already computed: latest, peak, peak date, average, % change and a rising, falling or flat call. Your agent doesn't have to do math over 52 data points.
• Honest errors. Google rate-limits Trends hard. When retries across IPs run out, you get RATE_LIMITED on that keyword, not an empty array that looks like zero interest.
• Every row links its sourceUrl on Google Trends, so the agent can cite it.

What it does: for up to 25 keywords, returns Google Trends data as flat JSON, one row per keyword: interestOverTime points, a ready-made summary (latest, peak, peak date, average, % change, rising/falling/flat), interestByRegion, and relatedQueries / relatedTopics (top and rising). Every row has status, sourceUrl (the matching Google Trends page) and, on failure, errorCode + errorMessage.

Use it when an agent needs to: check whether interest in a topic is rising, compare brands or products, find where a term is popular, or discover rising related searches for content and SEO ideas.

Don't use it for: absolute search volumes (Trends gives relative 0-100 values only), keyword CPC/volume data, or real-time "trending now" lists.

Cost per call

Event	Price	When
keyword	$0.0025	per keyword returned with data (all requested sections included)
actor-start	$0.001	once per batch run (not charged in Standby/HTTP mode)

Comparing 3 keywords = $0.0085. Rate-limited or failed keywords are never charged.

Rate limits (honest)

Google Trends rate-limits automated traffic hard (HTTP 429). This Actor retries with backoff and rotates IPs (direct, then Apify datacenter, then residential proxy) up to 10 times per request. Most calls finish in 5-20 s. If Google still refuses, you get status: "error", errorCode: "RATE_LIMITED" and nothing is charged; retry after a few minutes. requestAttempts and ipTiersUsed in OUTPUT show what happened.

Input (only keywords is required)

{ "keywords": ["chatgpt", "claude", "gemini ai"], "geo": "US", "timeRange": "90d" }

Field	Default	Notes
keywords	–	Up to 25.
geo	worldwide	US, GB, US-IA, ...
timeRange	12m	1h, 4h, 1d, 7d, 30d, 90d, 12m, 5y, all, or "YYYY-MM-DD YYYY-MM-DD".
sections	all three	timeline, regions, related.
outputMode	full	summary drops the timeline points (saves tokens).
compare	true	Shared 0-100 scale per group of 5 keywords.
property	web	youtube, news, images, shopping.
category	0	Trends category ID.
maxRegions / maxRelated	15 / 10	List lengths.

Output

{
  "status": "ok", "keyword": "claude", "geo": "US", "timeRange": "90d", "property": "web",
  "comparedWith": ["chatgpt", "gemini ai"],
  "summary": { "points": 93, "latest": 14, "peak": 25, "peakDate": "2026-10-14", "average": 17.6, "changePct": 17.6, "trend": "rising" },
  "interestOverTime": [{ "date": "2026-07-26", "value": 15 }],
  "interestByRegion": [{ "region": "District of Columbia", "code": "US-DC", "value": 32 }],
  "relatedQueries": { "top": [{ "query": "claude code", "value": 51, "formattedValue": "51" }], "rising": [{ "query": "claude opus", "value": 3000, "formattedValue": "+3,000%" }] },
  "relatedTopics": { "top": [{ "title": "Claude", "type": "Chatbot", "mid": "/g/...", "value": 100, "formattedValue": "100" }], "rising": [] },
  "isPartial": true, "sourceUrl": "https://trends.google.com/trends/explore?q=chatgpt%2Cclaude%2Cgemini+ai&geo=US&date=today+3-m&hl=en-US",
  "fetchedAt": "2026-10-25T05:00:00.000Z", "errorCode": null, "errorMessage": null
}

changePct compares the average of the last quarter of the window with the first quarter. Rising related values are % growth; "Breakout" means more than +5000%.

Fast calls: Standby (HTTP) mode

$curl "https://catalyst-labs--google-trends-api.apify.actor/?keywords=chatgpt,claude&geo=US&outputMode=summary&token=$APIFY_TOKEN"

Use from MCP

https://mcp.apify.com/?tools=catalyst-labs/google-trends-api → tool catalyst-labs--google-trends-api.

Data comes from Google Trends' public pages; no login, no personal data. Not affiliated with Google.

Made by Catalyst Labs. Also see: Agent Web Reader, Document to Markdown, YouTube Transcript API.