A Kothrud, Pune veg-thali restaurant ran the same Shravan Monday surprise for six years — over-prepped sabudana, under-stocked thali components, ₹5k waste discovered by 2pm. This is what changed in the first 30 days of the Forkcast pilot when fasting-day prep was driven by forecast, not gut feel.
The outlet
48 covers, Maharashtrian veg-thali format with a small non-veg add-on menu, Petpooja POS since 2020. Weekday lunch is office traffic; weekends are families. Shravan Mondays and Ekadashi days cut non-veg add-on covers 40–55% — historically discovered at 10:30am when the chicken batch was already marinading.
Onboarding: 11 minutes
- Petpooja API — read-only credentials; no menu changes required.
- Recipe mapping — top 16 dishes by revenue; auto-suggest matched 13; three manual portion tweaks on thali components.
- Fixed costs — rent, three salary lines, utilities entered once.
- First sync — 75 days of order history overnight, including the prior year's Shravan window; forecast live next morning.
Outcomes at day 30
| Metric | Before | After | Delta |
|---|---|---|---|
| Shravan Monday prep planning | 50–55 min | 12–15 min | −70% |
| Fasting-day waste | ₹4.5–6.2k / day | ₹900–1.4k / day | −78% |
| Cover forecast accuracy (fasting days) | ±28% vs actual | 92% (MAPE 8%) | +64 pts |
| Forecast horizon | Same-day gut feel | 30 days, dish-level | +30 days |
Shravan Mondays used to be our worst waste days. The forecast now tells us two weeks ahead which dishes to cut and which fasting specials to scale. Prep takes 15 minutes instead of an hour.
What the forecast caught
- Shravan Monday cover drop — flagged 14 days ahead; non-veg prep cut 45% before the week started.
- Ekadashi within Shravan — deeper drop (55%) modelled separately from regular Shravan Mondays.
- Sabudana/kuttu demand spike — fasting-special SKUs forecast up 50%; kitchen stopped under-prepping by noon.
- Post-fasting rebound — Tuesday after Ekadashi flagged +22% non-veg covers; prevented under-stocking.
What the owner actually does
The brief lands at 7pm with tomorrow's cover forecast, break-even check, and any calendar flags. On Shravan weeks, the owner reads the fasting-window section first — adjusts prep quantities for flagged dishes, shifts one cook from non-veg to fasting-special line. No daily dashboard habit.
Fasting-calendar signal was the single highest-ROI feature for this outlet. Western forecasting tools don't model Shravan or Ekadashi; Indian restaurants lose margin on the days they matter most. See the Hindkesari pilot for a comparable casual-dining outcome on FC Road.