Datalayer

I build actors for people who need the data to arrive every run, not most runs.

Most scrapers are written for the happy path. They work the day they are published, then quietly break — a site changes, a token expires, a rate limit trips — and you find out from an empty dataset three weeks later. That is the problem I build against.

Tested before release. Every actor ships with a test suite covering the failure modes, not just the success case.

Partial failures do not lose your data. If one keyword or page cannot be fetched, you still get everything else, plus a machine-readable record of exactly what is missing. A run that recovers 90% of your data returns 90% of your data.

You are only charged for what is delivered. No billing for empty results or failed extractions.

Flat, tabular output. One row per observation, ready for a spreadsheet, database or BI tool. No nested JSON to unpick first.

Honest nulls. A missing value comes back as null, never as 0 — because 0 is often a real value, and conflating the two quietly corrupts whatever you build on top.

Google Trends Scraper  — interest over time, top and rising related queries, related topics and regional breakdowns. Handles the session requirements that cause most Trends scrapers to fail, so it runs without a proxy.

More in progress.

Open an issue on the actor. I read them and I fix things.