UXPILOTstudies / slateboard-client-reports / run 2
FIX QUEUE · 1TUE 22:47
DKDev Kapoor Founder · Slateboard
Study · slateboard.app · v0.9.2
Slateboard client reports, run 2 of 2
FINISHED 22:31 TONIGHT
Run 3 queued for 06:00 tomorrow
43 agents · 5 personas187 agent-hours · $78.54 metered19 stuck points9 agents gave up mid-task
Stuck points
AGENTS DEPLOYED
43
5 persona types
STUCK POINTS
19
2 critical
GAVE UP MID-TASK
9/43
most at the onboarding CSV step
USABILITY (SUS)
58.9
grade D · below the 68 average
Agents stuck, by pagerun 2 · 43 agents
Stuck rate, run 1 vs run 2/clients got worse after v0.9.2
Ranked stuck pointsworst first · click one to inspect
showing 8 of 19 · the rest are below the triage line
Inspector
Session replays
Agent sessions8 of 43 shown
Timeline
Video replay, annotated
every flag links to the exact frame where the agent got lost
Persona coverage
Account manager, first login
inferred from slateboard.app · 11 agents
Task completion64%
Avg SUS49
Findings raised7
WORST MOMENTHunting for PDF export on /reports · median 58s
Agency owner on trial
inferred · 9 agents
Task completion71%
Avg SUS60
Findings raised4
WORST MOMENTTrial banner covering the save button on /billing
Freelancer leaving spreadsheets
custom, written by Dev · 9 agents
Task completion58%
Avg SUS52
Findings raised5
WORST MOMENTEmpty template picker on /reports/new until refresh
Client opening a shared report
inferred · 8 agents
Task completion88%
Avg SUS72
Findings raised2
WORST MOMENTCopied the share link twice, no copied confirmation
Returning power user
custom, written by Dev · 6 agents
Task completion83%
Avg SUS66
Findings raised3
WORST MOMENTRenaming a client silently discarded report edits
Add a persona
plain English, no scripts
Describe a user in a sentence or two and run 3 goes wider. Example already saved as a draft here tonight, "a bookkeeper who only logs in on the last day of the month".
[17:12]run 2 started · 43 agents · 5 personas · budget cap $95.00[17:13]auth check with stored trial credentials ......OK[17:19]prompt-injection gate scanned 214 page inputs ......CLEAN[18:02]agent WREN-27 stuck 94s on /reports/new · template picker empty[18:47]flag F-032 CRITICAL · empty CSV fails silently at onboarding · 4 agents and counting[19:33]flag F-031 CRITICAL · PDF export behind the kebab menu · median hunt now 58s[20:15]regression watch · /clients stuck rate 19% vs 12% in run 1[21:08]9 agents abandoned · sessions preserved with full replays[22:31]run 2 complete · 19 findings · SUS 58.9 · 187 agent-hours · $78.54[22:31]report emailed to [email protected]
You just read Dev's study the way he did, 16 minutes after it finished. His trial user hunted for the export button on a screen-share; 26 of 43 agents proved she was not the exception. Your app has its own list waiting.