An abandoned cart doesn’t merely mean the customer decided not to purchase. It means the customer decided not to purchase from you. That hurts.

As two of the world’s biggest online marketplaces, Amazon and eBay use very different strategies to convert shoppers who are sitting on the fence. Their approaches show, once again, that the most effective personalization strategies don’t come from a one-size-fits-all playbook. Instead, they’re based on each brand’s unique customer data insights, what those customers value, and the brand’s ability to build a unique and satisfying customer experience for each individual.

Amazon: All Retargeting, All the Time

If a shopper places something in their Amazon cart but doesn’t purchase it, Amazon will follow them all over the web and retarget them with ads for that item. Relentlessly.

Amazon seems certain the shopper will eventually buy the item from them – they just have to get them to push the button. From what we can tell, Amazon uses this strategy exclusively, and doesn’t even bother to send abandoned cart emails.

The obvious downside to retargeting is that it can be intensely annoying for the customer, especially if they’re constantly bombarded with ads for something they’ve already purchased. That’s because every time a shopper adds something to his or her cart, a cookie takes note. The problem is that that cookie doesn’t go away if the item is purchased or deleted from the cart.

Financially, this isn’t a problem for Amazon. The ads are likely bought through an ad exchange, so a third party is buying them on Amazon’s behalf, and betting that the shopper will click. If the shopper has already purchased the item for which they’re being retargeted, they’re not going to click on the ad, and Amazon won’t have to pay for it. There’s no incentive to change the system.

Email works completely differently, giving brands much more control. When they send abandoned cart emails, digital marketers are working off of databases. Once an item is out of the cart, it’s out of the database, and it no longer triggers an email. That ensures that marketers don’t send abandoned cart email for items that are no longer in the shopper’s cart.

Of course, some items seem to just hang around in carts forever. And given the volume of abandoned carts — about 70% of them — some retailers will set carts to auto-empty every 30, 60, or 90 days. Others will transfer languishing cart items to a wish list.

eBay: Unusual Levels of Personalization

Unlike Amazon, eBay has urgency on their side. They can credibly claim that if a particular shopper doesn’t buy the item, someone else will. And often, there won’t be a replacement available. The whole reason shoppers go to eBay, after all, is to hit the jackpot with a steal of a deal or a hard-to-find item.

eBay doesn’t just use abandoned cart emails to remind a shopper that she may miss out on a good buy. If an item goes down in price, eBay lets the customer know. This is most welcome and very rare. Even among the top 25 brands in our Personalization Index, it was unusual for cart-related messaging to be based on a customer’s personalized profile or on specific items left in their cart.

Additionally, eBay provides every little push it can to get that final sale. Shoppers might find some seller info in their abandoned cart emails. eBay also alerts the customer if free shipping is included.

eBay also uses the real estate in its abandoned cart emails to make recommendations for other items. Amazon makes recommendations too, of course, although they do it on their site rather than in email.

But not all recommendations are equal. The two marketplaces use very different types of recommendations to entice customers. Amazon is well-known as one of the first digital retailers to use collaborative filtering for recommendations. That means Amazon’s recommendations are based on what other similar shoppers have bought. That’s the ubiquitous “People like you also purchased…”

“People like you…” doesn’t really work for eBay, whose shoppers are more likely to be looking for one particular item. So eBay uses attribute-based recommendations rather than collaborative filtering. Recommendations are often based on what similar shoppers have purchased; eBay’s are for items similar to those that drew a shopper’s initial interest. The two marketplaces show, once again, why every form of personalization is more effective when it grows out of a unique understanding about one’s customers.