Five marketing decisions belong to your data, not your instinct.
When to send, which incentive to offer, who to group together, how often to write and when to chase an abandoned cart all vary by store and by shopper. Instinct is useful for choosing what to test. Your customers' behaviour, measured against a holdout group, should decide the answer.
Rules of thumb like Tuesday at 10am, a bigger discount converts better, or do not email too often are averages of other people's stores. They are often wrong for yours, and they are always wrong for some of your customers.
Five decisions are worth taking from data: send time per person, the type of incentive, segments built from behaviour, message frequency by engagement, and the timing of cart recovery by price, category and customer. For each one, state the assumption, test it on a split of your own customers and keep a group that gets nothing, so you measure what the change added and not just what it was credited with.
1. Send time: there is no universal best hour
The idea of one best time to send an email is persistent and wrong. A slot that works for one brand fails for another, and inside one list the right hour differs from person to person: some customers read at 6am, others at 9pm.
Use each customer's own engagement history to pick when their message goes out, rather than one slot for the whole list. Compare it with a fixed send time on a split of the list before you roll it out, and judge the result on orders, not opens.
2. Discounts: the type matters as much as the size
Conventional wisdom says a bigger discount converts better. It is not always true. On premium products a deep discount can make shoppers doubt the quality, while a shipping charge feels like a tax on the order. Free shipping can therefore beat a percentage off. A small, time-limited offer sometimes does as well as a deep one.
Which incentive works depends on your products and your customers, so test two or three before a big promotion, with properly segmented audiences. A dynamic coupon goes one step further: it gives each shopper only the incentive needed to buy, and nothing to those who would have bought anyway.
3. Segments: behaviour beats demographics
Most stores segment by age, gender or city. Behaviour says more: which categories a person browses, the price range they buy in, how often they buy and how they respond to messages. RFM segmentation is the simplest version.
One group is worth finding on its own: category enthusiasts, people who keep browsing one category without buying. Useful content about that category, followed by a modest offer, often moves them where a generic campaign does not. Segments should also update as behaviour changes, rather than being rebuilt from last year's purchases.
4. Frequency: let engagement set the cadence
The fear of annoying customers often leads to writing too little to the people who want to hear more. Your most engaged customers usually welcome more frequent, relevant messages. The least engaged ones are better served by fewer.
So set frequency by engagement, not by a single calendar for everyone: more for those who open, click and buy, less for those who do not. Watch revenue and unsubscribes together, per segment, to see whether the new cadence works.
5. Cart recovery: timing depends on what was left
Recovery messages for an abandoned cart are standard, but the right timing is not. It changes with the price of the item (expensive things need longer to decide), with the category (fashion and electronics have different decision cycles) and with the customer (a first-time visitor behaves differently from a repeat buyer).
A common starting sequence for fashion is a first message about an hour after the cart is left, a second a day later and a last reminder after three days. For high-priced items such as electronics, test longer gaps, for example two days before the second message and five before the last. Run the sequences side by side and keep the one that recovers more orders in your own data.
The assumption, and what to test instead
| Decision | The usual assumption | What to test |
|---|---|---|
| Send time | One best hour for the whole list | Each customer's own engaged hour against a fixed slot |
| Incentive | A bigger discount converts better | Free shipping, a small time-limited offer and a deep discount |
| Segments | Demographics describe the customer | Category interest, price range, purchase frequency, response |
| Frequency | Fewer messages avoid annoyance | More for engaged customers, fewer for the rest |
| Cart recovery | One sequence fits every cart | Timing by price, category and first-time or repeat buyer |
How to move from gut to data
Instinct is still useful: it tells you which questions to ask. Start by listing where each of these five decisions rests on an assumption today. Turn each assumption into a test on a split of your customers. Keep a holdout group that gets nothing, so you measure the incremental effect and not just the orders a message was credited with. Then roll out what wins, and keep testing, because customers change.
These decisions, made per shopper
Timing, frequency and incentive chosen per person
The agent picks when to write, how often and with which offer for each shopper, across email, push, SMS and Viber, from one memory of what they did.
How retention works →Every change measured against a holdout
A group the agent leaves alone shows what its choices added, in your own orders.
How results are proven →Questions about data-driven marketing decisions
Is there a best time to send marketing emails?
Not one that holds for every store or every customer. The right hour differs from person to person, so pick send time from each customer's own engagement history and test it against a fixed slot on a split of your list.
Do bigger discounts always convert better?
No. On premium products a deep discount can make shoppers doubt the quality, and free shipping or a small time-limited offer can do as well or better. Test the type of incentive on your own customers before a big promotion.
How often should a store email its customers?
Set it by engagement. Engaged customers usually welcome more frequent, relevant messages; customers who rarely open are better served by fewer. Watch revenue and unsubscribes per segment.
When should abandoned cart emails go out?
It depends on the price, the category and the customer. A common start for fashion is one hour, one day and three days; for expensive items, test longer gaps. Keep the sequence that recovers more orders in your data.
How do I know a change actually worked?
Keep a holdout group that does not get the change and compare orders between the two groups. That measures what the change added, not what it was credited with.
See which of last week's visitors you missed
One call, with your store and ad accounts open. Then 30 days free. A holdout group decides: if the agent doesn't add orders in 30 days, you don't pay.
See it on your store
You'll talk to one of us.