Ran a naïve forecast (3-period moving average) for Friday pizza, layered on 20% safety stock per best practice, and — thanks to RSVP oscillations — we ended up with 18 larges for 7 attendees at 12:30. Anyone have a lightweight control policy or kanban signal that tames snack demand variability without driving stockouts or, in this case, cold slices?
What fixed this for us was a simple two-card kanban: each person drops a ‘slice token’ by 10:30, we order ceil(tokens/2) pies with a hard cap of +1 safety pie, then at 12:10 we only top up if tokens still exceed delivered slices — cold leftovers vanished; can your vendor do 15‑min add‑ons?
Building on @nicole29’s tokens, swap the MA+20% for a tiny newsvendor rule: order Q at the 75th percentile of the last 6 token counts (trim the max). Can your shop handle a 10‑minute add‑on? If so, a 12:05 ‘andon’ that triggers exactly one small rush pie when claimed slices hit Q−2 works like a pizza thermostat.
Split the order into two pulls: 70% at 10:30 off tokens, then a guarded top-up at 11:45 only if net adds since 10:30 reach 8 slices (a full pie), cap +1; everyone after that’s a “backorder” to the slice window, not the fridge. If you can’t do the second call, freeze at 10:30 and weight late token changes at 50% so RSVP whiplash doesn’t turn the fridge into a cold-slice warehouse.
Try shrinking the batch: switch to half-pies or slices so your increment is 2 slices, then just order tokens + one “safety slice” at 10:45 and park any extra in a 2:00 snack bin so they don’t die on the table — it’s like swapping pallets for totes. Can your shop do halves/slices?