Google Trends Radar

Batch-first Google Trends scraper built for reliability. Feed it hundreds of keywords in one run and get back clean, normalized JSON for:

• Interest over time — the classic 0–100 curve, with average / latest / peak summary fields precomputed for spreadsheets and dashboards
• Interest by region — countries, states, cities or US metro areas
• Related queries — top & rising, the dataset Google throttles hardest
• Related topics — top & rising entities
• Trending now — real-time trending searches with approximate search volume, growth %, start time and related terms (up to ~500 active trends per country)

Google has no official Trends API (the announced one is a wait-listed alpha). This actor speaks to the same internal endpoints the trends.google.com website uses, wrapped in the thing that actually matters: smart backoff, session rotation and per-item fault isolation, so one blocked request never kills your batch.

Why this one

Pain with other Trends scrapers	Google Trends Radar
One keyword per run, or brittle CSV uploads	Native batch: keyword list × timeframe list in a single run
Random 429 crashes	Every request retried with fresh session + exponential backoff; failures isolated per item
Empty datasets, silent failures	Failed items are reported as explicit error rows — and you are not charged for them
Pay for a monthly rental you use twice	Pay per delivered result row

Input example

{
    "keywords": ["standing desk", "ergonomic chair", "walking pad"],
    "dataTypes": ["interest_over_time", "related_queries"],
    "timeframes": ["today 3-m", "today 12-m"],
    "geo": "US",
    "proxy": { "useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"] }
}

Every keyword is scaled independently (0–100) by default — ideal for keyword research. Flip compareKeywords on to get the same shared-scale comparison the Google Trends website shows.

Output examples

Interest over time (one row per keyword × timeframe):

{
    "dataType": "interest_over_time",
    "keyword": "standing desk",
    "geo": "US",
    "timeframe": "today 3-m",
    "averageValue": 62.4,
    "latestValue": 71,
    "peakValue": 100,
    "peakDate": "2026-06-14T00:00:00+00:00",
    "timeline": [
        { "date": "2026-04-21T00:00:00+00:00", "value": 55, "isPartial": false }
    ]
}

Related queries:

{
    "dataType": "related_queries",
    "keyword": "standing desk",
    "top": [ { "query": "best standing desk", "value": 100 } ],
    "rising": [ { "query": "walking pad standing desk", "value": 350, "formattedValue": "+350%" } ]
}

Trending now:

{
    "dataType": "trending_now",
    "query": "alaska airlines emergency",
    "geo": "US",
    "startedAt": "2026-07-19T22:30:00+00:00",
    "approxSearchVolume": 200000,
    "volumeGrowthPct": 1000,
    "relatedTerms": ["alaska airlines", "boeing 787"]
}

Scheduling (what this actor is built for)

Add an Apify Schedule and the actor becomes a monitoring pipeline: track your niche's keywords daily, export to Google Sheets, or push rows to Zapier/Make via the dataset API. Typical setups:

• SEO agency: 200 client keywords, today 3-m, daily at 07:00 — spot breakout queries before competitors write the content.
• E-commerce: category terms + related_queries weekly — mine rising queries for product listings.
• Newsroom / social: trending_now every 30 min with trendingHours: 4.

Pricing

Pay per event: a small start fee per run, then a fixed price per delivered result row (keyword-result for keyword datasets, trending-term for trending rows). Failed items are free and clearly marked. Rows where Google returned an empty result set (e.g. related data for a very niche keyword) are delivered free of charge — you only pay for rows that contain data. No subscription.

Tips for reliability

• Use residential proxy (default). Google throttles datacenter IPs hard, and related queries is the strictest dataset — on datacenter IPs it may fail even with retries.
• Keep concurrency at 2 unless your batch is small.
• Very large batches (500+ keyword × timeframe units) are capped per run — split them across scheduled runs.

Legal note

This actor extracts publicly available data from Google Trends. You are responsible for how you use the data. Numbers are Google's relative indices, not absolute search volumes (except trending-now approximate volume, which is Google's own estimate).