INTRO
The store sold each product on its own, and stopped there. Someone buying a hair dryer never saw the towel, the primer or the diffuser that goes with it, so the average order stayed small and a lot of obvious pairings never happened.
We designed a recommendation module in two places: complementary products on the product page, and an upsell in the cart drawer as people moved toward checkout. The idea was to put the right add-on in front of someone at the two moments they're most open to it, when they've just chosen a product and again when they're about to pay.
KEY CHALLENGES
We weren't trying to fill the page with more things to buy. We wanted to suggest the few items that actually go with what's already in the cart.
THE OUTCOME
We ran it against the old flow across full store traffic for about four months, split 50/50. Average order value rose 4.08%, from $56.42 to $58.72, and the result was statistically significant (p=0.04). In December alone the module brought in roughly $33.5K in extra revenue.
Two things worth naming: December is peak season in the US, so that month runs hot, and the module's own add-to-cart rate (people adding straight from a recommendation) sat low in absolute terms while lifting sharply against the version without it. The money it made is the number that matters, and it made money at full scale.
THE THINKING
Cross-sell usually fails one of two ways. It either buries the main product under a wall of suggestions, or it recommends things that have nothing to do with what you're buying. We did the opposite: a small set of genuinely complementary items, shown at the moments a shopper is already deciding. On the product page it answers "what else do I need with this." In the cart drawer, right before checkout, it catches the add-on someone forgot. The AOV lift came from relevance and timing, not from showing more.