🟡 MOCK PROTOTYPE — Meet quality analytics · Mirrors the Grafana “Meet Dashboard” · Figures are REAL (StarRocks, iOS daily, 2026-04-15→06-21) · See repo
Meet
Call & meeting quality · iOS daily · version groups old / 4.10.* / 4.11.* · iOS-only SDK change · source: ads_botim_meet_call_quality_stats_di
iOS · Bad-call rate (4.11.0)
31.9%
vs 4.10.x 59.4% · ▼ 27.5pp
iOS · Call success (4.11.0)
73.1%
vs 4.10.x 78.7% · ▼ 5.6pp
iOS · Calls (4.11.0)
559K
4.11.1 1.29M · ramp continuing
Android · Bad-call (control)
63.3%
no SDK change · ~ flat
iOS · call-volume share by version group · 100% / day rollout / migration
iOS · daily bad-call rate · by version group one point/day · ≥1,000 calls/day floor
iOS · daily call-success rate · by version group
All anomalies · daily trend
Each panel = one bad-call sub-signal. A call is counted "bad" if it trips any of these; bad_call_total is the de-duplicated union.
old (4.4–4.9)4.10.*4.11.* (SDK change)iOS · daily rate (anomaly / total calls) · ≥1,000 calls/day floor · y-axis auto-scaled per panel
Join failed
Socket failed
Socket slow
Slow join
High RTT
Very high RTT
Audio freeze
Video freeze
Audio loss
Video loss
Meet level · all platforms · by app version source: ads_botim_meet_quality_stats_di
Version Total meets Participants / meet Bad-meet rate
4.10.x6.33M2.4353.2%
4.11.0217.5K1.8728.5%
4.11.1596.5K2.1030.1%
Platform code follows the warehouse convention device_type 0 = iOS, 1 = Android. The Grafana panel SQL labels these inverted — figures here use the corrected mapping.