eBay’s bundle discount is automated - it’s still guessing

eBay runs a feature called “Bundle Discounts”. It applies a percentage discount across multi-item orders and store-wide by default. Sellers don’t opt in; it’s just on.

Sellers aren’t all happy about it. High-ticket items get discounted along with everything else, and there’s no easy way to exclude complex catalogues from the rule.

A seller’s £400 item ends up in the same bucket as a £4 one, because the rule can’t tell them apart. It just knows “more than one item, apply the saving.”

That’s the risk with any blanket rule: it can’t tell the difference between a shopper who needed a nudge and one who was already reaching for their card. It treats every basket the same, because that’s all it’s built to do.

Automated isn’t the same as intelligent

It’s worth separating the two things eBay actually did here. They automated the delivery of a discount. They didn’t automate the decision of who should get one.

Those aren’t the same job. A rule that fires the same offer at every basket, every time, isn’t reading anyone’s intent. It’s just running on autopilot, and autopilot doesn’t know that the shopper buying five items was always going to check out anyway.

It doesn’t know the difference between someone stuck on the fence and someone who’s already decided.

This isn’t really a story about eBay. Most retailers run some version of the same rule: a sitewide sale, a threshold discount, a banner that fires for everyone who lands on the site that week. It’s simple to set up and easy to explain to the board.

It’s also blind. It can’t see who’s hesitating and who isn’t, so it discounts everyone the same amount, whether they needed the push or not.

The gap between what retailers think and what shoppers feel

That gap between automation and intelligence shows up everywhere, not just at eBay. Research on eCommerce personalisation found 92% of retailers believe they personalise well. Only 48% of their shoppers agree.

That’s a big gap between what a brand thinks it’s doing and what a shopper actually experiences. Most brands believe they’re reading the room. Most shoppers can tell they’re not, because the offer they see has nothing to do with what they were about to do anyway.

Part of the reason for that gap is rules like eBay’s: they look targeted but aren’t. A percentage off a bundle feels like a smart, modern touch.

To the shopper on the other end, it’s the same blanket discount their favourite store has run every month for a year.

What reading intent actually looks like

The same research points to what works instead. Abandoned cart messages, triggered by a shopper’s own behaviour, generate 31.3 times more revenue per 1,000 messages than generic broadcasts sent to everyone.

That’s not because the offer itself is better. It’s because it’s aimed. It goes to someone who showed a real signal, like adding an item to a basket and then stalling at checkout, rather than everyone who happened to be browsing that day. The message matches the moment.

The retailers getting this right aren’t discounting more. They’re discounting fewer people, more precisely, at the exact point hesitation shows up. That’s a different skill from running a sale.

It means watching behaviour in real time, not scheduling a promotion two weeks in advance and hoping it lands on the right shoppers.

Free shipping is telling the same story

Discounts aren’t the only blanket lever losing its edge. Free shipping is still influential; 62% of shoppers say it affects their decision, according to Ryder’s 2026 eCommerce Consumer Study. But that influence has dropped 14% year on year. Shoppers now expect it, rather than rewarding you for it, and 71% will abandon a basket due to an unexpected fee at checkout.

Free shipping followed the same path a lot of blanket discounts are on now: it started as a differentiator, and turned into table stakes the moment every competitor offered it too. A lever stops working the moment it no longer feels earned. The fix isn’t offering more of it.

It’s offering it, or a discount, only to the shoppers who actually need it to convert.

Why this matters more as AI shopping tools grow

There’s another layer coming. Shoppers are increasingly letting AI tools do the comparing and deal-hunting for them. The same Ryder research found 64% of shoppers now use AI to discover and compare products, and 70% use it to hunt for a better deal. McKinsey’s research on US consumers points in the same direction: shopping habits are shifting earlier each year and becoming more automated on the buyer’s side, with more shoppers starting their holiday shopping well ahead of peak season and leaning on AI to do so.

A blanket, store-wide discount is easy for a bot to spot and repeat. It’s published, predictable, and the same for everyone, which is exactly what makes it simple for an AI agent to find and apply on a shopper’s behalf. A discount aimed at one hesitating shopper, in one moment, based on their own behaviour, isn’t something an AI agent can find or trigger for someone else. It only exists because of what that one shopper just did.

Precision isn’t just better for margin. As shopping gets more automated on the buyer’s side, it’s also harder to exploit.

The fix isn’t fewer discounts. It’s better-aimed ones.

None of this means discounts are the problem. Used well, they still convert hesitant shoppers and clear stock that needs to move. The problem is applying the same offer to everyone in the basket, regardless of whether they need it, as eBay’s bundle discount rule does by default.

This is exactly the gap RevLifter closes. Instead of a rule that applies the same discount to every order, RevLifter reads how each shopper is behaving in real time and determines who actually needs a nudge to buy and who was going to buy anyway.

The shoppers who were already converting see nothing, so the margin stays where it belongs. The ones who are wavering at checkout, in their basket, or browsing with obvious deal-seeking behaviour get the right offer at the moment; it changes the outcome. Every campaign runs against a control group, so the extra sales are shown, not assumed.

Automating a discount is easy. Knowing who it should go to is the hard part, and the part that actually protects your margin.