Screen the participant
- At least 18 and an actual or likely user
- Matches the chosen maker or apparel-production track
- Did not build the site, live with a builder, or receive navigation coaching
- Can use their normal device and assistive technology
Facilitator kit · Study directory-core-v1 v1.1
Recruit consistently. Facilitate quietly. Count honestly.
Moderated baseline gate5 makers + 5 apparel practitioners
Self-guided sessions, internal QA, household members, product-team walkthroughs, and incomplete sessions do not advance this target.
View the public claim gates01 · Build the invitation
One link carries the lane, track, and round.
Send the generated participant-information link before the session. Do not describe the intended categories, labels, or successful routes.
Ready-to-send invitation
Stitch Authority is testing whether people can find and judge textile resources. We are looking for someone who sews, mends, knits, crochets, weaves, dyes, or otherwise works with textiles. This is a 20–30 minute facilitated task study with three realistic tasks. A facilitator will observe quietly; this is not training or a product pitch. The runner collects an anonymous session code, track, task result, one friction category, and timing—no name, email, audio, video, or free-text response. Participation is voluntary, and the session can be stopped or withdrawn. Read the participant information and decide whether to take part: https://stitchauthority.com/research/participate?mode=moderated&cohort=maker&round=baseline
https://stitchauthority.com/research/participate?mode=moderated&cohort=maker&round=baseline02 · Neutral session protocol
The product should do the explaining.
Use this same sequence every time so differences in facilitator behavior do not become fake differences in the interface.
Set expectations
Send the information sheet in advance. At the session, repeat that the site—not the participant—is being tested; participation is voluntary; and they may stop or withdraw.
Explain observers, notes, or recording separately and obtain separate consent. The runner does not grant consent for those extras.
Run without coaching
Ask the participant to operate the site and think aloud. Read one scenario at a time. Stay quiet. If they become completely stuck, give the smallest prompt needed to continue and record the task as assisted.
Close and preserve the boundary
Let the participant review the recorded outcomes. Copy or download the de-identified local record. Keep any consent record or identifying recruitment administration separate from the task record.
A qualifying moderated session must satisfy the report contract.
It must declare moderated mode, include facilitator attestation, finish all three unique track tasks, emit a completed-session event, and have no later withdrawal event. The reporting script rejects legacy, self-guided, incomplete, duplicate-task, and withdrawn sequences from the moderated gate.
After each batch
Synthesize failures, not compliments.
Export the production logs, run npm run report:validation -- path/to/logs.json, compare repeated friction by task, and keep the raw event export out of public artifacts.
Publish only aggregate participant mix, scope, failures, changes, confirmation results, and remaining limits.
Return to the validation ledger →