How do I sell more to existing customers?

Everything here depends on treating a customer who bought once six weeks ago differently from one who has bought four times and is due again. A single list, sent a single message, is why most stores conclude that existing customers do not respond.

Four groups do most of the work. People who bought once and are inside their normal reorder window, who need a nudge and nothing more. People who bought once and are well past that window, who have effectively lapsed. People who buy repeatedly, who are your most valuable audience and should almost never see a discount. And people who spent a lot on a single order, who behave differently from frequent small buyers and often want the premium or complementary item rather than a refill.

Your ecommerce platform can produce all four from order history without any extra software. Do that first and look at the sizes. A store that discovers most of its customers sit in the once-and-lapsed group has a different priority from one where the repeat group is large and under-served.

Then write for each group separately. Four short, specific emails beat one long, general one every time.

Take your own order data and work out the typical gap between a customer's first and second order. That number, not a round figure someone recommended, sets the timing of everything.

For a consumable, the cycle is close to how long the product lasts. A container that lasts about eight weeks means the useful message arrives at week six or seven, while the customer still has some left and before they have gone looking elsewhere. Arriving after they have run out means you are competing with whatever they bought in the meantime.

For products that are not consumed, the cycle is driven by something else: a season, a growth stage, a wear-out point, a matching item, a gifting occasion. A store selling children's clothing has a cycle set by how fast children grow. One selling equipment has a cycle set by accessories and replacement parts.

Work the number out per product group rather than for the store as a whole, because averaging a fast-moving consumable with a durable item produces a timing that is wrong for both. Then set the replenishment message to fire on a delay from each customer's own order date, which every serious email platform can do.

The highest-return message to an existing customer is usually about a different product, chosen from what people who bought their item actually go on to buy.

You can get this out of your order data without any modelling. Look at customers who bought product A, then look at what appeared in their next order. The two or three items that come up repeatedly are your cross-sell for product A, and they are usually not the items you would have guessed. Then say why the pair goes together in plain terms, because a recommendation with a reason converts and a grid of unrelated products does not.

The same data builds bundles. Products that keep appearing in the same basket can be offered as a set, with a small saving that costs less than the discount you would otherwise use to win the order.

And it tells you what to leave out. Recommending the item someone bought a fortnight ago, when it lasts a year, reads as a store that does not know who it is talking to. Suppressing recently purchased items from recommendations is a small setting that changes how the whole programme feels.

Reduce the effort between the message and the order, because a repeat customer has already decided and any friction is pure loss.

A one-click reorder link that puts the previous order straight into the cart removes the whole browsing step. Saved addresses and payment details, and a customer account they can actually use, remove the typing. For anything consumed on a schedule, a subscribe-and-save option is the strongest version of this: the customer decides once and the reorder happens without a message at all.

Subscriptions only work where the product genuinely runs out and the interval is predictable. Offered on the wrong product they get cancelled after the second delivery and cost you goodwill. Where they do fit, the details matter more than the discount: easy skipping, easy interval changes, easy pausing, and a reminder before each charge. Subscriptions that are hard to pause get cancelled outright instead.

For products bought at intervals nobody can predict, the equivalent is a back-in-stock alert and a saved wishlist, so the store gets in touch when the reason to buy appears rather than when the calendar says so.

Change what the messages are about, not just who receives them. The most common mistake is sending existing customers the acquisition message: the brand story, the introductory offer, the case for the category. They have heard it and they bought.

What works on someone who already owns the product is help getting more out of it. How to use it properly, what to do with the part everyone gets wrong, what other customers pair it with, when to replace it, how to look after it. That is a message people open, and it makes the next purchase feel like a continuation rather than a pitch.

Keep discounts for the moments where they change a decision, mainly the lapsed group. Sending a code to customers who were going to reorder anyway is paying for orders you already had, and it teaches your best customers to wait.

Measure this properly by holding a portion of each segment back from the send for a period and comparing. Without that, a replenishment flow gets credited with every order that would have happened anyway, and you end up scaling something that is doing less than the report says.