The best time to send a notification is about an hour before the moment your customer usually decides to buy — not whenever suits you. The same message sent at two different hours produces two completely different outcomes, and timing is the cheapest thing you can improve.
Timing is not a detail, it is half the message
A notification saying "two stamps left on your free coffee" arriving at eight in the morning reaches a customer on their way to work who passes you ten minutes later. The identical message at ten at night reaches somebody in bed, who reads it, thinks "nice", and forgets it permanently.
Nothing changed about the wording, the reward or the customer. What changed is the distance between reading the message and being able to act on it. That distance is the entire difference between a notification that works and one that is wasted — and it is free: sending at the right hour costs you nothing.
The one-hour-before rule
To work out when to send a wallet notification, find your daily sales peak in your point-of-sale system and subtract an hour. That is your send time. Do not send during the peak itself, because by then the customer has decided and has already arrived somewhere. Do not send far ahead of it either, because the intention goes cold.
| Trade | Demand peak | Send at |
|---|---|---|
| Morning cafe | 7:30 to 9:00 | 6:45 am |
| Lunch restaurant | 1:00 to 3:00 | 11:45 am |
| Dinner restaurant | 8:00 to 11:00 | 6:30 pm |
| Barbershop | Thursday and Friday | Wednesday evening |
| Car wash | The weekend | Thursday midday |
Note the last two rows: in weekly trades the rhythm is weekly rather than daily, so the question is not which hour but which day. Wednesday evening precedes the "when am I getting a haircut this week" decision by hours, and Thursday midday precedes cleaning the car before the weekend.
The dead hours
Send
- About an hour before your daily peak.
- On an ordinary working day, Tuesday through Thursday.
- When the customer could actually reach you within two hours.
- Once, then let the result speak before you send again.
Never send
- After ten at night — it is read and never acted on.
- Before seven in the morning — it wakes people, annoys them, and gets deleted.
- During prayer times or your own known closing hour.
- On payday itself, when every other notification lands at once.
The most damaging entry in that right-hand column is the first, because it looks harmless. A late-night notification usually annoys nobody, but it spends your credit of the customer attention for nothing: they read it, did not act, and by the time your next message arrives they have learned that your messages do not call for action.
How often, and at what spacing?
The safe ceiling is two a month for most shops, three in a season. But the number matters less than where the decision comes from: the best send schedule is not a schedule at all, it is a response to the behaviour of the individual customer.
The second case is far and away the strongest. A customer one stamp from their reward is standing at the edge of something they started themselves, and a reminder there is a service rather than an advert. If you send nothing else all month, send that one.
Finding your own hour
- Export a month of sales by hour Most point-of-sale systems give you a table of receipts per hour, and that is all you need.
- Find the hour the climb begins, not the hour it peaks The start of the climb is when the decision happens; the peak is when it is executed and already too late.
- Subtract an hour, then run it for a fortnight Compare the share who visited within four hours of the send, not the share who opened it.
- Change one variable per test Change the hour and the wording together and you will not know which made the difference, and you will repeat the mistake on the next campaign.
The right measure here is visits, not opens. A notification everyone opens and nobody acts on is a failure; one that few open and half of those visit is a success. Do not optimise a number that puts nobody through your door.
Common questions
Should everyone get it at the same moment?
If you serve one neighbourhood, yes — the rhythm is similar enough. If you have branches in different cities or customers with different habits, split them. The simplest useful split is by their own usual visiting time: morning visitors get a morning message, evening visitors an evening one.
What if my trade has no obvious peak?
Look for a weekly peak instead of a daily one; most trades have one even when the day looks flat. If you truly find nothing, start at eleven on a Tuesday morning as a neutral baseline, then move it two hours every fortnight until the results settle.
Does timing change during Ramadan?
It changes completely rather than partially. The peak inverts to the hour before iftar and to after taraweeh, and morning hours go entirely dead in most trades. Rebuild the schedule for the whole month from scratch rather than shifting your usual one by a few hours.