A busy breakfast rush, simulated minute by minute.
Turn the fixes on, one at a time:
A simulation: station times, prices and customer numbers are estimates, not the operator's data. Use it to compare fixes, not to predict one outlet's exact wait. Simulated numbers are the median of 20 mornings; the counter replays the most typical one.
How much of the rush each station was busy. Above 85%, work arrives faster than it gets done and the queue keeps growing.
Every customer who walked up, valued at what they meant to order. Prices are estimates: kaya toast set $6.20, French toast set $7.20, toast + drink $4.60, drink $2.30, add-on $2.00.
Not counted: customers who never return after a bad morning, or who read the reviews and go elsewhere. That loss is real but can't be simulated honestly.
Newest reviews (up to 150 per outlet) from the three best-rated busy outlets, the four lowest-rated, and three in one new town, pulled 25 Sep 2026. Coded against a fixed list of reasons; an independent second coder agreed on 89% of labels (50-review check). Brackets are 95% intervals.
| Reviews mentioning | Best 3 | Lowest 4 |
|---|---|---|
| Staff attitude (complaint) | 13% [10–17] | 31% [27–35] |
| Slow service (complaint) | 9% [6–12] | 14% [11–18] |
| Food quality (complaint) | 13% [10–17] | 15% [12–18] |
| Food taste (praise) | 46% [41–52] | 15% [12–19] |
| Reviews read | 345 | 486 |
Complaints about the food itself are no different. What differs is the experience of getting it.
| Reviewer's wait (Google field) | Avg ★ | n |
|---|---|---|
| No wait | 3.55 | 86 |
| Up to 10 min | 3.74 | 73 |
| 10–30 min | 2.81 | 21 |
| 30 min or more | 1.86 | 7 |
Customers arrive at random at the chosen rate: the rush rate during the rush, the quiet rate otherwise, for the first four hours after opening. A customer who sees more than 6 people in the line to pay may walk away, and the longer the line, the likelier (4% more per extra person, up to 60%). Each order splits into a drink and food items that queue for crew and equipment (toaster 4 slices, griddle 4 slots, egg bath 12). The drink goes to the drinks pickup, food is packed for the food pickup, and an order counts as done only when both are out. Dine-in customers hold a seat for about 25 minutes, and the table is cleared before it is reused.
The live promise looks at the last hour of finished orders of the same kind, learns how the wait grows with the work already queued, and quotes a time that would have been right for 8 (or 9) in 10 of them. Food-quality risk is measured two ways: the drink is ready 5 or more minutes before the food, or cooked food waits 3 or more minutes to be packed. Estimated crew minutes per item (fastest / typical / slowest): cashier 0.5/1/2, drink 0.5/0.8/1.5, kaya toast by hand (scrape, spread, butter, cut) 0.6/0.9/1.3, French toast by hand 1.2/1.6/2.4, eggs 0.2/0.3/0.4, pack 0.5/1/1.5, clearing a table 0.5/1/1.5. Cooking time on the equipment: toast 2/3/4, griddle 4/5/7, eggs 6/7/8. Crew times are multiplied by the new-crew slowdown. The kitchen's hands-on times are set so a busy morning's typical takeaway wait lands near the median wait reviewers report (30 minutes, 65 reviews that state one). The engine matches the textbook queueing formula (Erlang C) to within 0.3% in its tests. Two weeks of the outlet's till, order-ready screen and roster data would replace these estimates with measured values.