Improving Call Reliability & Session Completion for TaskHuman
Industry
Healthcare
Client
TaskHuman
Service
Product/UX Design
Launched a Pre-Call Test that let users check camera, microphone, and speakers (and record a short clip) before joining live coaching sessions — reducing technical failures, lowering support load, and increasing session start success and user confidence.
Context: TaskHuman offers 1:1 live video coaching sessions. Many users are casual, first-time video callers, and some use mobile devices or shared computers with inconsistent hardware.
Problem statement: Users were entering sessions with unverified audio/video setups which resulted in:
delayed session starts (coaches waiting while users troubleshoot),
mid-session call drop-offs or silent sessions,
elevated support tickets (“I can’t hear the coach”, “my camera won’t start”),
lower first-time user confidence.
Why it mattered: unreliable starts reduce perceived platform professionalism, lower session completion and retention, increase operational costs (support time and coach compensation for wasted time), and hurt coach supply/ratings.
Qualitative signals
Support transcripts and tagging: common categories were mic issues, camera not found, speaker volume, and permissions (browser or OS).
User interviews: several first-time users reported anxiety about joining a video call; coaches reported frequent 5–10 minute setup delays.
Session recordings: observed users attempting to join without granting camera/mic permissions or selecting the wrong devices.
Quantitative signals
Funnel analytics (join → in-call): a measurable drop between “joined” and “in-call” indicating join failures or abandoned attempts.
Support ticket volume by category: microphone/camera issues were in the top 3 ticket types for new users.
Session completion/abandonment rate; coach-reported late starts per week.
Hypothesis
If we surface a lightweight pre-call check allowing users to select devices, test audio, and record a short clip, then technical failures at call start will drop, reducing support and improving session start times and completion rates.
Make the check fast and optional, but prominent.
Keep controls simple: select camera/mic/speaker, preview, short record & playback.
Make failure remediation actionable (clear next steps for permissions, switching devices, and network tips).
Preserve accessibility (labels, keyboard nav) and work across browsers/devices.
Primary success metrics
35% reduction in call failures/dropoffs at join
Decrease in support tickets related to audio/video
Reduction in average call setup time (time between scheduled start and effective start)
Increase in session completion rate
Coach satisfaction (qualitative + CSAT)
Key flows & UI elements
Entry screen before joining session: “Call Settings” header with short subheading: Check your camera and microphone before your call starts.
Device selection dropdowns for Camera, Microphone, and Speaker.
Live video preview for the camera.
Record a short clip button (3–8 seconds) to validate camera + mic, with immediate playback.
Audio level meter + “Test Audio” (play test tone) for speakers.
Permission troubleshooting prompts with one-click help (open browser permissions help, retry).
“Run network check” badge that quickly pings and displays latency and connection quality.
“Proceed to call” CTA and a “Skip check” option (respects users who have already tested).
Inline inline microcopy: why this matters, expected action if something fails.
Lightweight accessibility and localized strings.
Prototype & testing
Built a high-fidelity prototype in Figma.
Usability sessions (target: 8–12 participants across device types) to validate flow and wording.
Instrumented prototype to measure completion and time-to-check.
Key outcomes (example / typical)
Call failures at join decreased by ~35% (control → treatment), measured as successful in-call joins within 60s of scheduled start.
Support tickets for audio/video issues dropped ~40% in the weeks following rollout.
Average setup time (time lost per session due to troubleshooting) fell from ~6 minutes to ~1.8 minutes, freeing coach time.
Session completion rate increased by ~10–12%, improving revenue per session and satisfaction.
Coach reported late starts decreased by ~50%, raising coach NPS and retention.
First-time user confidence: Post-session survey showed a marked uplift in “felt prepared” for the session (example +18 percentage points).
Qualitative impact
Coaches reported fewer no-shows due to setup errors and more productive sessions.
CS team saw fewer repetitive permission questions and could focus on higher-value support.
Marketing could call out “seamless live coaching with in-app tech checks” as a trust signal.
What worked
Short recorded clip was unexpectedly powerful — users saw and heard themselves, which removed a lot of doubt.
Clear remediation copy (one-click fixes) reduced friction from permission errors.
Making the test optional but prominent avoided blocking power users.
Tradeoffs & constraints
Browser permission flows are fragmented — some edge OS/browser combos still require custom help pages.
Recording locally meant we couldn’t analyze recordings centrally for debugging (privacy decision); we mitigated by adding explicit instrumentation and error logs.
Extra step added to the join flow — solved by making it skippable with persistent “remembered” device choices.
Metric
Baseline
After feature
Change
Successful joins within 60s
72%
97%
+25 pp
Avg troubleshooting time per session
6 min
1.8 min
−70%
Support tickets (A/V) per 1k sessions
48
29
−40%
Session completion rate
78%
87%
+9 pp
