We’re piloting two 20-minute microlearning sessions per week (9:15 and 3:45) with a single floater on inbox duty, and our first-response time slipped from 2:40 to 3:10. What training cadence and coverage model are you using to protect SLAs — staggered seats, rotating floaters, or after-hours — and how do you size headcount for those blocks?
We shifted to staggered seats (10‑min offsets) plus a tiny surge pool sized with Erlang C (https://www.callcentrehelper.com/tools/erlang-calculator/) to keep occupancy under 85% during sessions — think zipper merge, not a roadblock — and it settled at about 1 floater per about 6 on-queue with “no-train windows” around our top 30‑minute peaks. What does your arrival curve look like 9–10 and 3–4; any spikes you could dodge?
I’d push those to 9:40 and 4:10 and split into two 15‑min cohorts with a 5‑min buffer; that kept our FRT from drifting like your “2:40 to 3:10.” For coverage, we staff one extra floater per about 12 seats pulled, sized off the last 4 weeks’ arrivals at those minutes and capped at 80% occupancy. If you have to keep 9:15/3:45, pause low‑urgency email for 25 minutes and add a “running about 30 min behind” banner; if your trough hits later in the hour, shift the blocks there.
Building on @logan62: during those blocks, cap chat concurrency by one and disable proactive chat, then run two quick “triage sweeps” (minute 0 and 10) where a cross‑trained lead sends a 1–2 line first touch + ETA to every new ticket; that kept our FRT about 2:40 without adding heads. It’s like tapping the brakes before a speed camera — do your metrics count that human triage as the first response?