Upselling and Cross-Selling Without Being Pushy

Every shop assistant knows the two versions of the same sentence. "Would you like anything else?" asked at the door is noise. "That jacket runs cold in November — the liner clips into it" said while folding the jacket is service, and it sells. The words barely differ. What differs is that the second one comes from someone who looked at what you were buying.

Chat is the closest an online store gets to that counter, which is why suggestions land so differently here than in a checkout upsell widget. It is also why they backfire faster: a badly timed offer in WhatsApp feels like a stranger texting you an advert. Our in-depth guide to selling and taking orders on WhatsApp covers the mechanics of the order itself; this piece is about the sentence that comes after it.

Helpful or pushy is a question of relevance

A suggestion is helpful when it solves a problem the customer already has because of the thing they just bought. Batteries with the toy, a filter with the machine, a second cartridge because one runs out in six weeks. It is pushy when it solves a problem you have — a slow-moving SKU, a monthly target. Customers cannot read your inventory report, but they feel the difference immediately: the suggestion has nothing to do with what they said.

One rule keeps you honest. If you cannot finish the sentence "I am suggesting this because..." with something about the customer, do not send it.

Three moments that work

Timing does more work than copy. There are three windows where an extra item feels natural, and a great deal of space in between where it does not.

1. At the order, before payment: the accessory that belongs to the item, added in one message so nothing has to be shipped twice.
2. A few days after delivery: the customer has used the product, and the follow-up doubles as a check-in — "did it fit? the matching case is here if you want it".
3. At the natural reorder point: consumables have a rhythm, and a message that arrives just before someone runs out reads as thoughtful rather than commercial.

Anything outside those windows is a broadcast wearing a costume. Sending an offer three hours after delivery, before the box is even open, is the most common mistake — and the one customers mute you for.

Purchase history is the whole trick

Relevance is not a writing skill, it is a data problem. If the assistant can see what this contact bought, in what size, how often, and what they returned, the suggestion practically writes itself — and, just as usefully, it knows what not to offer. Recommending the same lamp someone bought last month is the fastest way to look automated. Pair that memory with the structured product data behind setting up a WhatsApp product catalog customers actually browse, and the assistant can name the exact matching item instead of a category.

Bundles that make sense

The strongest bundles answer a question the customer would otherwise have to research: which grinder for this coffee machine, which mount for this screen, which shade for this frame. Two or three items, one price, one message. Compare that with a discount on a random second item — the first removes work, the second only removes margin.

Keep the upgrade version simple too. The best upsell is one step up, with one concrete reason: more capacity, longer warranty, the size that actually fits. Offering three tiers in a chat message turns a suggestion into homework, and homework gets ignored.

“

The best cross-sell is the message you did not have to apologise for later.

— be digital ai team

Knowing when to say nothing

Silence is a feature. Do not suggest anything to someone with an open complaint, a pending return, a late delivery, or a warranty case. Do not suggest anything twice in the same week, and never immediately after a customer has already said no. Write these as hard rules in the automation rather than trusting tone to save you; one add-on message on top of a delayed parcel undoes a month of goodwill.

The same restraint applies to volume. One suggestion per order is plenty. Businesses that cap themselves usually see a higher acceptance rate than those that ask at every touchpoint, because the message still counts as information rather than advertising.

Measure basket size, not clicks

Click-through on a suggestion tells you almost nothing, and it rewards the wrong behaviour: the more offers you fire, the better the raw numbers look. Track average order value, items per order, and the reorder rate of customers who accepted a suggestion versus those who did not. Then watch the counterweights — opt-out rate, return rate on suggested items, and how many recovered baskets you would have had anyway, which is where the numbers in recovering abandoned carts on WhatsApp overlap with these.

Start with one product and one moment: the accessory people already ask about, suggested at the order. If basket size moves and nobody mutes you, add the next one.