Retail send-time advice optimises for opens. Restaurant send-time has to optimise for something narrower: arriving inside the window in which the meal decision is actually being made. An email opened at 3pm and an email opened at 4: 45pm may have identical open rates and completely different order rates.
Households decide dinner in a fairly narrow band, and it is later than most marketing calendars assume.
| Occasion | Decision window | Send window |
|---|---|---|
| Weekday lunch | 10: 30–11: 45 | 10: 15–11: 00 |
| Weekday dinner | 16: 00–18: 30 | 15: 45–17: 15 |
| Friday / Saturday dinner | 15: 30–19: 30 | 15: 30–17: 00 |
| Late night | 20: 30–23: 00 | 20: 15–21: 30 |
| Weekend lunch / group | 10: 00–12: 30 | 09: 45–11: 15 |
| Catering and office orders | Two to five days ahead | Tue–Thu mornings |
Every customer has a personal ordering time distribution, usually tight. Someone whose last ten orders all landed between 18: 10 and 19: 40 has a window, and sending to them at 16: 00 puts the message an hour and a half before they are thinking about food.
Compute the median order time per customer and send 45 to 90 minutes before it. Where history is thin, fall back to the store’s dominant window for that day of week. The lift from this is consistently larger than the lift from subject-line testing, and it takes a fraction of the effort.
Two practical limits override the theory. First, kitchen capacity: sending a large offer to arrive at 17: 30 on a Friday can produce a rush the store cannot serve, and a marketing win that becomes an operations failure is a net loss. Stagger large sends over 30–60 minutes.
Second, quiet hours. SMS has legal and practical quiet-hour restrictions, which in the United States generally means no marketing messages before 8am or after 9pm in the recipient’s local time zone. Multi-market operators must send on local time, not on head-office time.
A promotion that arrives at the perfect moment for a kitchen that cannot cook it has not helped anybody.
For a weekday dinner offer, roughly 16: 00–17: 00 local time. But a per-customer window beats any global answer, and the gap between the two is worth more than most creative testing.
For broadcasts, yes: Thursday and Friday consistently outperform for weekend occasions, and Monday is the weakest day in most pizza markets. For triggered journeys, day of week barely matters because the trigger is the timing.
Yes, over 30 to 60 minutes, for two reasons: it protects the kitchen from an unservable spike, and it improves email deliverability by avoiding a sudden volume burst from your sending domain.
Keep reading
Setting send frequency per segment rather than per calendar, global frequency caps across channels, and the fatigue signals that appear long before unsubscribes do.
ChannelsConsent, 10DLC registration, message construction, cost per order, and the discipline that keeps an SMS list alive for years rather than months.
Audience & dataTreating dayparts and occasions as separate audiences: the lunch rush, the family Friday, the late night, game day, and the office order.
Campaigns & offersBuilding triggers from the weather forecast and the local fixture list: thresholds that work, lead times, and how not to promote into a night the kitchen cannot cover.
marketing.pizza runs all of it, every night, across every store you have.