Route one: a defined task and structured path evidence
Useberry documents a Figma connection with click tracking, user flows and recordings. Maze’s integration is described by Figma in terms of task success, misclicks, duration and screen-level heatmaps. These capabilities make both relevant starting points for a specified path question, but they do not establish that the tools calculate every metric identically. [S05] [S18]
Write the intended start state and observable end state. For example: “Starting from the account page, find the invoice for the latest completed order.” Do not put the label of the correct control into the task. This sample is an editorial example; it is not a validated universal script.
Route two: watch the task unfold
Userbrain documents a workflow where you provide a public Figma prototype link and tasks, then watch think-aloud recordings. That is a useful route when the spoken explanation and visible behavior are the primary output. Do not call a recorded video a structured path-analytics export unless that output is actually documented for the chosen product and plan. [S10]
Recordings and metrics can be complementary. A completion count identifies an outcome; the session may help you understand how that outcome happened. Neither justifies inventing the participant’s motivation. Keep an observed action, a spoken comment and your interpretation as different notes.
Access settings can change the whole test
Maze’s current import instructions describe supported Figma permissions and a separate Enterprise restriction for password-protected prototype imports. Figma-side “available on all plans” language refers to that side of the integration; it is not a promise that every Maze capability is free. [S16] [S18]
Create a dedicated test copy containing only material you can share. A public-link prototype is not appropriate for confidential customer content just because the study is intended for a small panel. Preview the actual participant route outside the editor’s signed-in session.
Missing screens are not user confusion
Maze documents that imported coverage depends on reachable screens and flow configuration. A screen omitted from the imported flow is a setup problem, not evidence that participants cannot navigate to it. Its prototype-block documentation also calls for refreshing a linked prototype after changes and describes one imported prototype per study. [S17] [S38]
Freeze the version used in a comparison. Write down the flow, device context and date. When the design changes, keep results from the earlier version identifiable rather than blending them into a fresh-looking chart. This is our reproducibility rule, not a claim about a platform ranking algorithm.
A practical selection
Select a structured prototype route when the decision depends on a specified task, route or first interaction. Select a narrated-recording route when you need to inspect the attempt in context. Add a live moderator when adaptation during the session is part of the method; do not confuse a video recording with a live interview.
The tool picker uses those distinctions. Compare the Useberry and Userbrain workflows for a closer example, then calculate the actual audience and plan costs separately.