Google Trends Scraper

Pull Google Trends data for a list of keywords: interest over time, interest by region, and the top and rising related searches. It's the same data trends.google.com shows, for any country, region or worldwide, and it downloads as JSON, CSV or Excel or comes through the Apify API.

Google rate-limits Trends requests quickly, and that's where the scrapers I tried before building this one kept failing. This one keeps a small pool of IPs, each with its own Google session. When Google answers "too many requests", it drops that IP and retries on a fresh one, and if that keeps happening it switches to residential IPs by itself. A 50-keyword run on 26 September took 58 seconds and every keyword came back.

What you get for each keyword

• Interest over time: one value from 0 to 100 per day, week or month, depending on the time range
• The average interest over the whole period
• Interest by region: countries, states, cities or US metro areas, sorted from highest to lowest
• Top related queries and rising related queries, with growth such as +350% or Breakout
• A link to the same view on Google Trends, so you can check any number yourself

Every result also carries a status. Keywords that Google has no data for come back as no_data and cost nothing.

Input

{
    "keywords": ["mortgage rates", "home insurance", "solar panels"],
    "geo": "US",
    "timeframe": "today 12-m",
    "compareKeywords": false
}

• keywords: up to 100 per run.
• geo: a country code like US, GB or DE, a region like US-CA, or empty for worldwide.
• timeframe: past hour, 4 hours, day, 7 days, 30 days, 90 days, 12 months, 5 years, or everything since 2004. customTimeRange takes exact dates, for example 2024-01-01 2024-12-31.
• compareKeywords: off gives each keyword its own 0 to 100 scale. On puts up to 5 keywords on one scale, which is Google's limit, so you can say one is searched twice as much as another. More than 5 keywords are split into groups of 5.
• searchType switches between web, image, news, YouTube and Google Shopping searches, and category narrows the results to one Google Trends category.
• language sets the language of region names and queries, for example de for German names.

Output

One dataset item per keyword. Here is a real one, trimmed:

{
    "keyword": "chatgpt",
    "status": "ok",
    "geo": "US",
    "timeframe": "today 12-m",
    "averageInterest": 79,
    "interestOverTime": [
        { "date": "2026-09-13T00:00:00.000Z", "value": 78, "isPartial": false },
        { "date": "2026-09-20T00:00:00.000Z", "value": 75, "isPartial": true }
    ],
    "interestByRegion": [
        { "geoCode": "US-CA", "geoName": "California", "value": 100 },
        { "geoCode": "US-DC", "geoName": "District of Columbia", "value": 94 }
    ],
    "relatedQueriesTop": [{ "query": "what is chatgpt", "value": 100, "formattedValue": "100" }],
    "relatedQueriesRising": [{ "query": "chatgpt caricature", "value": 3350, "formattedValue": "+3,350%" }],
    "trendsUrl": "https://trends.google.com/trends/explore?date=today+12-m&q=chatgpt&geo=US"
}

The dataset has four views. Overview has one row per keyword. Interest over time, Interest by region and Rising queries have one row per data point, which is the shape you want for a chart or a spreadsheet.

Pricing

You pay $0.002 for each keyword that returns data, so 50 keywords cost $0.10. Keywords with no data are free, and Apify adds its standard $0.00005 per run for starting the Actor. You can set a maximum charge per run in the run options, and the Actor stops cleanly when it reaches it instead of failing.

Limits

• The numbers are Google's relative interest from 0 to 100, not search volumes.
• Related topics are not included. The Google endpoint this Actor reads returns an empty list for them.
• Keywords with very little search volume, or small regions, often have no data at all. That's Google's answer, not an error, and those keywords are free.
• The Google SERP proxy doesn't work for Google Trends, so the Actor refuses it. The default Apify proxy is the right choice for almost everyone.
• Up to 100 keywords per run. For more, start several runs or a schedule.

Changelog

• 2026-09-26: first version.