Why do online stores keep recommending items you already bought?

A Common and Understandable Frustration

One of the most common complaints shoppers have about recommendation engines is being shown, over and over, an item they’ve already purchased. It’s a natural frustration, since a purchase would seem like the clearest possible signal that the shopper is done shopping for that item. But the reason this keeps happening comes down to how these systems are actually built and what signals they prioritize.

Understanding why it happens also clarifies which cases are genuine mismatches and which are, in fact, reasonable behavior from the retailer’s perspective.

Purchase History as a Strong but Blunt Signal

Recommendation engines treat a completed purchase as one of the strongest available signals of interest, often stronger than a click or a page view. That makes sense in categories where repeat purchases are common and expected, such as household consumables, pet supplies, or personal care products. In those cases, showing the item again isn’t a flaw — it’s a useful reminder that helps a shopper reorder something they’re likely running low on.

The trouble arises in categories where a purchase is typically a one-time event, such as furniture, large appliances, or electronics. Many recommendation systems don’t automatically distinguish between “the shopper wants more of this” and “the shopper is unlikely to need another one soon,” especially if the underlying model treats all purchases similarly regardless of category.

Gifts, One-Off Buys, and Model Lag

Another common cause is purchases made as gifts rather than for the shopper themselves. Since the system generally can’t tell the difference between someone buying for their own use and buying for someone else, it may interpret the purchase as reflecting the buyer’s personal taste and continue recommending similar items. There’s also often a lag between a purchase happening and a model’s next retraining cycle, meaning the suppression of a now-irrelevant recommendation doesn’t happen instantly.

Retailers have been working to refine these systems by building category-aware suppression logic — recognizing, for instance, that a mattress or a major appliance shouldn’t trigger the same kind of repeat suggestion that a coffee pod or skincare product would.

Bottom Line

Recommendation engines keep suggesting already-purchased items mainly because they weight purchases as a strong interest signal without always distinguishing one-time buys, gifts, and repeat-purchase categories from each other. It’s often more a limitation of current model design than an outright error, and it’s an area retailers continue to refine.

Go deeper

• How do businesses budget for the ongoing cost of ai tools versus a one time purchase?