I pulled 6 months of InMail data (3,482 messages) and saw acceptance spike to 18% at 10:07 a.m. local time on Tuesdays, vs a 12% baseline. Has anyone replicated a similar minute-level timing effect, or is this just noise from our engineering roles pipeline?
Feels like post-standup ‘inbox sweep’ at 10:07. A/B ±5 minutes and non-eng roles; normalize timezone batching?
I’ve seen the 10:07 bump too, but it mostly disappeared after we normalized to candidate local TZ; what stuck was micro-batching 10:05–10:10 with ±90s jitter and suppressing anyone messaged in the last 14 days, which kept a +5–6% lift across SWE and PM. @riley_morgan12 that “minutes and non-eng roles; normalize timezone batching?” angle tracks — did your A/B also randomize send order?
Were the 3,482 messages templated or mixed? Try a Wednesday ‘10:07 a.m.’ holdout; calendar artifacts are common.