What should recommendations show when the cart is empty?
Context, not history. With nothing in the cart, lean on the session: the category being browsed, the search terms used, and store-wide social proof like bestsellers and trending items. Save the 'frequently bought together' logic for carts that actually have something in them.
Why empty-cart is its own placement
Most recommendation logic assumes a seed: a cart item, a viewed product, a purchase history. The empty cart has none of that, which is why teams either show it generic bestsellers or, worse, leave the slot blank. But the empty cart is often the highest-intent moment on the site: the shopper opened the cart drawer, which means they are thinking about buying. That intent deserves better than a blank panel.
The key insight is that 'empty cart' does not mean 'no information'. The session is full of signals. What the shopper browsed, searched, and filtered in the last ten minutes is a better predictor of the next click than their lifetime history.
The four strategies that work
- Session-category bestsellers: the top sellers in the category the shopper has been browsing. Specific enough to feel relevant, safe enough to never be wrong.
- Search-aware picks: if the shopper searched, the recommendations should respect the search terms rather than resetting to global bestsellers.
- Trending now: social proof with urgency. 'Trending this week' converts the shopper who wants validation that others are buying.
- Recently viewed: the simplest and most effective. Shoppers open the cart, see the thing they were just looking at, and the path to purchase shortens.
What to avoid
- Do not show 'frequently bought together' with an empty cart. The algorithm will either show nothing or show random pairings, both of which look broken.
- Do not personalize from stale history alone. A shopper who bought a gift last month does not want gift recommendations in an empty cart today.
- Do not overload the placement. Four to six well-chosen items beat a scrolling wall of twenty; the empty cart is a nudge, not a catalog.
- Do not hide the empty state behind recommendations. Acknowledge the cart is empty, then offer the picks. Shoppers distrust a cart drawer that pretends.
Measuring empty-cart placements honestly
The metric is first-item add rate: what share of empty-cart views lead to an item being added. Compare against a control showing static bestsellers, and segment by traffic source, because the empty cart of a paid-traffic shopper behaves differently from that of a returning customer.
Watch for cannibalization. If the empty-cart recommendations lift cart adds but the added items are cheaper than what shoppers would have found anyway, the placement is optimizing for the wrong thing. Track average first-item value alongside the add rate.
The returning-shopper exception
There is one case where the empty cart is not a blank slate: the returning shopper with a rich history. If a logged-in customer with fifty past orders opens an empty cart, their lifetime behavior is genuinely predictive, and ignoring it would be wasteful. The rule is to blend: weight the current session highest, the recent history second, and the lifetime history last.
Even then, keep the session in charge of the ranking. A loyal customer browsing a new category is exploring, and the empty cart should reflect the exploration, not drag them back to their usual purchases. History is the tiebreaker, not the driver. When the session and the history disagree, the session wins.
Reviewed
Published Sep 25, 2026.