Our test protocol (fixed in advance, applied to every tool)
- Same workflows. Three standard workflows: (W1) lead capture, (W2) branching/approval, (W3) scheduled polling. Identical logic in every platform.
- Same data. 100 runs per workflow, including 10 deliberately malformed records (missing fields, bad formats) to test error handling.
- Same apps. Webhook endpoint, Google Sheets, Gmail, Slack — connected identically on both sides.
- What we record. Billing units consumed per 100 runs; success/failure count; average execution time; timed first-build setup minutes; steps/modules required; error visibility and retry behaviour.
- Evidence. Builder screenshots, run-history and consumption screenshots, and the raw results CSV are archived for every test and available on request.
- What we don’t do. We don’t test on vendor-provided accounts or accept payment for placement. Enterprise-only features we cannot access are marked “not tested” — never guessed.
- Freshness. Pricing is re-verified against the vendor’s official pricing page on the day a report is published, and all flagship reports are re-run quarterly.