A shopper buys a camera from your store. A week later they email asking which memory card to use, because the camera did not come with one. They buy the card from a big marketplace. You had the right card in stock. Nobody mentioned it.
That is a missed upsell, but more importantly it is a missed chance to help. The best bundles and add-on suggestions are not sales tactics. They are answers to questions the shopper would ask if they thought of it.
What makes a good suggestion
Good add-on suggestions share three features:
- They complete the purchase. A case for a phone, batteries for a toy, a care kit for leather boots.
- They are specific. "This 64GB card works with this camera" beats "you might also like."
- They are easy to decline. One mention, not a series of prompts.
Suggestions that fail these tests feel like pressure. Suggestions that pass them feel like service.
Script 1: the missing accessory
Visitor: Does this camera come with a memory card? Assistant: It doesn't include one. It takes SD cards, and the 64GB card on our accessories page is a good fit for photos and short video. Would you like me to point you to it?
The shopper asked a question. The suggestion answers it.
Script 2: the bundle that saves money
Visitor: I'm buying the shampoo and conditioner, is there a deal? Assistant: Yes, the hair care bundle includes both plus the leave-in treatment for less than the shampoo and conditioner bought separately. It's on the bundles page if you'd like to compare.
Only mention bundles that exist and prices that are published.
Want a 24/7 AI Employee for Your Website?
See how SpideyChat can answer customer questions, capture leads and handle enquiries while your team is offline.
Script 3: the compatibility question
Visitor: Will this lens fit my camera? Assistant: This lens fits cameras with the X-mount. If yours is one of the models listed on the product page, it'll work. If not, the adapter on the same page lets you use it with most other mounts.
Compatibility questions are some of the most useful moments for suggestions, because the shopper needs the answer to buy.
Script 4: when not to suggest anything
Visitor: Just checking this arrives by Friday. Assistant: If you order before 2pm today, standard delivery should arrive by Friday.
No suggestion. The shopper asked a delivery question; pushing an add-on here would be irritating.
Accuracy comes first
An assistant can only recommend accurately if your product pages describe compatibility and what goes with what. SpideyChat for ecommerce stores answers from your pages, so preparing them matters, as covered in how to train a store assistant on your product catalogue. When compatibility is unclear, it should say so, as explained in what happens when a store assistant does not know.
Measuring whether it helps
Watch average order value for conversations that included a suggestion, and returns on suggested items. If returns rise, suggestions are not accurate enough. If order value rises without more returns, they are working.
Fit questions in other businesses
Recommending the right add-on is a form of product fit advice, the same skill used when developers ask whether an API suits their needs, as in API questions from developers evaluating your product. For more on retail fit questions, see the AI shopping assistant and product fit.
To see how it would recommend from your own catalogue, try an assistant on your store, or look at the other features.