Reliable Trends Scraper

Every other Google Trends scraper on the Store shares the same complaint in its reviews: "ran for 20 minutes, didn't return anything", "took my money for zero results". That's not a Google Trends limitation — it's what happens when a scraper doesn't handle rate-limiting and IP blocking properly, and bills you anyway.

This one is built around a single rule: a keyword is only ever billed if it actually returned real data. If Google blocks every attempt, you get a clear line in the run log explaining why — not a silent empty row you paid for.

How the reliability actually works

• Residential proxy by default. Google Trends aggressively blocks datacenter IPs — this is the #1 root cause of empty/blocked runs on other scrapers. Every request goes out through Apify's residential proxy pool.
• Session rotation, not just retries. If one proxy identity gets rate-limited, the actor doesn't just retry on the same IP — it opens a fresh proxy session (new IP, fresh cookies) and tries again, up to max_proxy_sessions times per keyword.
• Exponential backoff with jitter inside each session, so retries don't hammer Google in a tight loop (which is itself a fast way to get blocked harder).
• Silent-empty detection. A 200 OK with a technically-valid but content-free response is treated as a failure, not a success — the exact failure mode the incumbent's bad reviews describe.
• No charge on failure. A keyword that fails every session/retry is logged as a warning and simply not pushed to the dataset — you are not billed for it.

Input

• keywords (required): search terms to check
• geo (optional, default worldwide): 2-letter country code, e.g. US, FR
• timeframe (optional, default "Past 3 months"): now 7-d, today 1-m, today 3-m, today 12-m, today 5-y
• category (optional, default 0 = all categories): Google Trends category ID
• include_related_queries (optional, default true): also fetch top/rising related queries
• max_proxy_sessions (optional, default 3, max 6): how many fresh proxy identities to try per keyword before giving up on it
• proxyConfiguration (optional, default: Apify residential proxy): can be overridden, but a datacenter/no proxy will get blocked far more often — expect a lower success rate if you turn it off

Output (one row per keyword that returned real data)

{
  "keyword": "bitcoin",
  "geo": "US",
  "timeframe": "today 3-m",
  "data_points": 93,
  "latest_value": 29,
  "peak_value": 100,
  "average_value": 51.2,
  "top_related_query": "bitcoin price",
  "top_rising_query": "spacex stock price",
  "timeline": [{"date": "May 15, 2026", "value": 50}, "..."],
  "related_top": [{"query": "bitcoin price", "value": 100}, "..."],
  "related_rising": [{"query": "spacex stock price", "value": 38850}, "..."]
}

Values follow Google's own 0-100 relative-interest scale. timeline/related_top/related_rising carry the full data; the flat fields above them are a quick-glance summary for the dataset table view.

Known limitation — read before you buy

Google Trends' internal API is unofficial and can change shape without notice — true of every scraper in this category, not specific to this one. What's different here is what happens when it does: a keyword that stops working gets skipped and logged, not billed as if it succeeded.

Pricing

Pay per result — no charge for keywords that come back empty.