Google Trends Scraper (Reliable)

Get Google Trends data for many keywords in one run: interest over time, interest by region, related queries (top and rising), plus related topics when Google provides them. You can compare up to 5 terms in a single chart. It works with any country or region and any time range, from the past hour to 2004–present, across Web, YouTube, News, Images and Shopping search.

Built for reliability. Each query gets a fresh IP session. When Google rate limits it, the query is retried with backoff and steps up to residential IPs if needed, keeping any sections that already succeeded. The actor also slows down on its own if Google starts blocking. A failed query is reported in the dataset and never charged.

Why this actor

• Batch-friendly pricing. You pay per query that returns data, plus a tiny start fee, so 1,000 keywords in one run cost about $2. A single-keyword call costs about 2 cents.
• Fast. It calls Google Trends' own JSON endpoints instead of loading a browser. Most queries finish in a few seconds.
• Complete. Timeline, subregions, related queries (top and rising) and related topics.
• Honest billing. Errors, queries where Google returns no data, and empty extra sections aren't charged. Runs stop cleanly at your max charge limit.

Input

Field	Example	Notes
searchTerms	["air fryer", "iphone, samsung galaxy"]	One query per line. Separate up to 5 terms with commas to compare them.
geo	US, GB, US-CA	Leave empty for worldwide. Case doesn't matter.
timeRange	today 12-m	Presets, or a custom "2024-01-01 2024-06-30" via the API (real dates from 2004-01-01).
sections	["relatedQueries"]	Extra sections: interestByRegion, relatedQueries, relatedTopics. Interest over time is always included.
property	web, youtube, news, images, shopping	Which search vertical to use.
category	0	Google Trends category ID.
maxConcurrency	5	Parallel queries (1–20). Lowered automatically under rate limiting.
maxRetries / queryTimeoutSecs	6 / 300	Retry budget and time limit per query.

Output (one item per query)

{
  "searchTerm": "iphone, samsung galaxy",
  "keywords": ["iphone", "samsung galaxy"],
  "geo": "US",
  "timeRange": "today 12-m",
  "interestOverTime": [{ "date": "2025-11-02T00:00:00.000Z", "formattedTime": "Nov 2 – 8, 2025", "isPartial": false, "values": { "iphone": 51, "samsung galaxy": 3 } }],
  "interestByRegion": [{ "geoCode": "US-WY", "geoName": "Wyoming", "values": { "iphone": 79, "samsung galaxy": 18 } }],
  "relatedQueries": { "iphone": { "top": [{ "query": "iphone 17", "value": 100 }], "rising": [{ "query": "iphone 17 pro max", "value": 4250, "formattedValue": "+4,250%" }] } },
  "scrapedAt": "2026-11-01T20:02:30.000Z"
}

Per-keyword numbers are under values, keyed by the search term. Values follow Google's 0–100 relative interest scale. Rising values are percentage growth ("Breakout" means more than +5000%).

Pricing (pay per event)

• Actor start: $0.02 per run.
• Trend query ($0.002): one per search term, or per comparison group, that returns interest-over-time data. Includes interest over time.
• Extra section ($0.001): one per additional section (region, related queries, related topics) that returned data.

Notes

• Rising queries are filtered for decoys. Google sometimes slips generic, unrelated "rising" queries (for example "learn golang" or "best vpn service" for protein powder) into results for automated traffic. They change on every request. The actor drops rising queries that share no meaningful word with the keyword or its top queries, and reports how many it removed in risingRemoved. Top queries and interest over time aren't affected. So when Google poisons a list, you get fewer rising queries, not fake ones.
• Related topics: Google currently returns no related topics through its data feed for almost all queries (0 of 1,000 in our November 2026 test). The section stays available and is never charged when empty. Use related queries for keyword research.
• Very low-volume terms may return all zeros. That's Google's data, so it's delivered and charged as normal. If Google returns no timeline at all, the item has a note and isn't charged.
• A failed query has an error field. Common causes are persistent rate limiting or a query Google rejects (for example an invalid category).
• Use it via the API, on a schedule, or from AI agents through the Apify MCP server.

Use cases

SEO and content planning, product research, seasonality forecasting, market sizing, brand comparison, and trend alerts for newsletters and dashboards.

Use it from Python or JavaScript

Coming from pytrends? google-trends-api is an open-source client (MIT) with a drop-in TrendReq that calls this actor:

from google_trends_api import TrendReq   # was: from pytrends.request import TrendReq
pytrends = TrendReq()
pytrends.build_payload(["chatgpt", "gemini"], timeframe="today 3-m", geo="US")
pytrends.interest_over_time()

Pair it with

• Google Trending Now Scraper: what people are searching for right now, with search volume, by country.
• Google Ads Transparency Scraper: every Google and YouTube ad a brand runs, with ad copy and landing pages.