Scoring methodology
Scores on this site are not a feeling. Seven criteria, fixed weights, applied the same way to every tool. Here they are, so you can disagree with the weighting rather than guess at it.
The criteria
| Criterion | Weight | What it measures |
|---|---|---|
| Time to first working workflow | 25% | Wall-clock time from empty account to a workflow running against real data, including reading docs and fixing what broke. |
| Total cost at real volume | 20% | Monthly cost at the volume a 1-20 person business actually generates, not the headline tier price. Counts operation limits, task overages, and paid connectors. |
| Failure behaviour | 15% | What the tool does when a credential expires, an API rate-limits, or input is malformed. Whether it tells you, retries, and lets you resume without rebuilding. |
| Connector coverage that matters here | 15% | Support for the systems small distributors and service businesses actually run: spreadsheets, accounting, email, e-commerce, and plain HTTP for everything else. |
| Escape hatch | 10% | Whether you can export your workflows, run custom code, and leave. Lock-in is a cost that shows up eighteen months later. |
| Learning curve for a non-developer | 10% | Whether a business owner who is comfortable with spreadsheets but does not write code can maintain the thing without calling somebody. |
| Vendor durability | 5% | Funding, pricing history, and whether the company has changed terms on existing customers before. |
Why these weights
Time to first working workflow carries the most weight because it is the cost that stops most small businesses from automating anything. Nobody abandons automation because the monthly fee was four dollars too high. They abandon it because three evenings disappeared and the thing still did not run.
Vendor durability carries the least, not because it does not matter, but because I cannot measure it honestly from the outside. A criterion I can only guess at should not move a score much.
Thresholds
- A tool scoring below 5 on failure behaviour does not get recommended for anything that touches money or stock, whatever its total.
- A tool with no export path gets a stated warning on every page it appears on.
- Where two tools land within half a point, the page says they are a tie rather than manufacturing a winner.
What a score is not
It is not a prediction that a tool will suit you. A shop with 40 SKUs and a four-person agency want opposite things. Scores rank tools against the criteria above; the "best for" and "skip it if" lines on each review are the part that applies to you.
The process behind the numbers is described in how I test.