Orange hours are floor-driven — the staffing-minimum rule, not the revenue math. Opening is always floor-driven (it's not revenue-based at all); Mid/PM turn orange when the 2-person minimum is higher than what forecast revenue alone would need. Violet marks a major holiday, adjusted from last year's actual vs. typical same-weekday revenue; red means the store had $0 sales on that date last year and is assumed closed — verify or add hours via Extra Sales if you're opening. Teal (shown per week) marks a seasonal trend — last year's same calendar week vs. its own trailing-4-week baseline at the time — applied to every non-holiday day's Mid/PM revenue that week; holiday days in a trending week use only the holiday multiplier, not this.
Loading…
Notes
Hours use an empirical model fit to actual Sling labor hours vs. sales (regression, n=211 days, R²=0.48): regular Mid+PM hours = a 14h/day baseline (net of Opening's hours, so it isn't double-counted) + 10% of regular revenue ÷ $20/hr, split between Mid (11am–3pm) and PM (3pm–9pm) by their historical revenue share and floored at 2 people scheduled. Opening (8:30–11am) is its own fixed staffing rule, not revenue-driven — 2.5h Mon–Fri, 3.0h Saturday, 3.5h Sunday. Catering uses a labor rate that tapers from 10% for small orders (≤$2,000) down to 7% for large ones (≥$10,000), reflecting batch-prep efficiency at scale — no floor, since it's a discrete booked event. Mid/PM revenue is the same-weekday average over the last 4 weeks, excluding catering (including large regular-menu orders that look like catering by their items). Catering hours are for known, already-booked orders only — not forecast. Extra Sales lets you add a manager override for a day (e.g. a known event or promo), charged at the regular 10% rate; it saves immediately but only folds into the official forecast on the next hourly refresh, so it shows a "pending" note until then. Week subtotals run Monday–Sunday (pay week), starting next Monday. 12 major holidays (New Year's Day, Easter, Mother's Day, Memorial Day, Father's Day, July 4th, Labor Day, Halloween, Thanksgiving, Christmas Eve, Christmas Day, New Year's Eve) get their Mid/PM revenue scaled by a multiplier = last year's actual revenue on that holiday ÷ last year's typical same-weekday revenue (other holidays excluded from the "typical" sample so e.g. Thanksgiving can't skew a nearby Christmas Day estimate). If last year's holiday revenue was $0, the day is assumed closed this year — Opening/Mid/PM all zero out with no floor applied. Catering and Extra Sales are unaffected by holiday adjustments either way.
Each forecast week also gets a seasonal trend multiplier: last year's actual revenue for the same calendar week (52 weeks / 364 days back, Monday-aligned — a deliberate simplification, not a true 4-4-5 fiscal calendar) compared to last year's own trailing-4-week baseline at that time, with any of the 12 major holidays excluded from both sides (so a holiday landing in that week doesn't get double-counted — it's already handled by the holiday adjustment above). One ratio applies to every non-holiday day's Mid/PM revenue in the week; holiday days use only the holiday multiplier. If fewer than 4 of the 7 weekdays have clean data after exclusion, the ratio defaults to 1.0 (no adjustment) rather than acting on thin data, and the raw ratio is clamped to 0.7–1.4 either way. Note: the trailing 4-week baseline used for ordinary (non-holiday, non-seasonal) days does not itself exclude holiday-contaminated days from its own sample — that's a separate, pre-existing characteristic of the base model, out of scope here.