Maze runs task-based unmoderated tests against a prototype or a live product and returns quantitative task metrics from what people actually did. Reynard has no participants at all. The difference is not better or worse — it is what you have available at the moment you need an answer.
Once you have people and a task, this is the tool that answers the question properly.
A completion rate from real users doing a real task is evidence. Everything Reynard produces about the same task is an expectation. If you can get the measurement, take the measurement.
'Can someone find the pricing and start a trial' is a testable task with a start and an end. Unmoderated task testing is the correct instrument for that question.
If you have visitors or a recruited panel, sending a test to them is cheap and the sample grows without you doing anything. Reynard's panel size is fixed by design.
Structural differences only. Where a figure would be a guess, it is left out.
| Aspect | Maze | Reynard |
|---|---|---|
| Method | Task-based unmoderated testing against a prototype or live product | An agent run through the live product plus a read by eight weighted population groups |
| What the numbers mean | Measured behaviour of real people on defined tasks | A predicted reaction from a simulation |
| What you need before you start | Traffic, a recruited panel, or a list to send the test to | A URL or a description — nothing else |
| Task metrics | Quantitative, per task | None; the scan reports where the path breaks, not completion rates |
| Right moment | There is a specific task to measure and people to measure it on | Before the page is published or the flow has been seen by anyone |
| Population coverage | The people who took the test | Eight groups weighted on published population data |
| Cost per iteration | Depends on how you source participants | A per-run price, free for the first check |
| Standing of the evidence | Real behavioural data | A simulation — no respondents, no significance, not fieldwork |
Every run returns the same shape of output, whichever market you point it at.
Stated plainly, because it is the whole difference: a Reynard result is a prediction of reaction. It is not a measurement of behaviour. Nobody completed a task. No time on task was recorded. No one dropped out of a funnel.
That is why the two fit different moments. Reynard fits the page you are about to publish and the flow nobody has seen yet — the point where there is nothing to measure. Maze fits the point after that, when there is.
Reynard is a simulation. It is not fieldwork, it carries no statistical significance, and it cannot replace talking to a customer. Treat it as a fast, structured second opinion that tells you where to look — not as evidence.
The agent run, step by step, and what the report contains.
Read the comparisonThe narrow version: one page, one agent pass, eight groups on whether the offer is understood.
Read the comparisonReynard has published no accuracy figure. This page states the validation experiment we will run, and what a good result would and would not prove.
Read the comparisonOne check costs nothing and needs no account. You see the format of the output before you decide whether it is worth paying for a full run.
Start the free check