What One A/B Test Taught Me About Recommended Products

In one of my eCommerce product roles, I had a goal every online retailer shares: get customers to add more related products to their cart. Bigger baskets, higher average order value. The question wasn't whether to show recommended products — it was where and when.

My team had two ideas, and honestly, I had a favorite.

Idea A: As soon as a customer added an item to their cart, we'd automatically scroll them down the Product Details page to a Recommended Products section. Impossible to miss. Maximum visibility.

Idea B: Show a small set of recommended products near the Checkout button on the Cart page, so customers would see them right before heading to checkout. Quieter. Easy to scroll past.

On paper, Idea A looked like the winner. More eyeballs should mean more add-to-carts, right? Instead of debating it in a conference room, we split traffic between the two and let it run for two sprints — about a month.

Idea B won. Customers in that group added more products to their cart.

The lesson: don't interrupt the customer's flow

Looking back, the reason is pretty human. When someone taps "Add to Cart," they already have a next step in mind — keep browsing, check the cart, head to checkout. Idea A hijacked that moment. We moved the page out from under them to a section they weren't expecting. Even with good intentions, it felt like being steered, and being steered is annoying. We weren't helping them shop; we were forcing them to look.

Idea B met customers at a completely different moment. Right before checkout is when people pause and ask themselves: Did I pick the right one? Am I missing something? Sometimes they genuinely second-guess their choice. A few well-chosen recommendations at that point weren't a distraction — they were useful. Some customers added a complementary item. Others swapped the product they originally picked for one that fit them better. Either way, we were helping with a decision they were already making.

Less obvious turned out to be more effective, because it showed up when the customer actually wanted it.

Then we tested the next assumption: how many to show

Once Idea B won, the natural instinct was to show more recommendations. If a few work, more should work better.

It didn't. We tested different quantities, and past a certain point, more recommendations created more confusion. Too many options at the moment of decision made customers stall. Some got stuck comparing, and some abandoned their cart and checkout altogether. The extra choices didn't add revenue — they cost us orders we already had.

That second test was just as valuable as the first. Without it, we would have "optimized" our winning idea right into a conversion problem.

Why this matters

Every product team has ideas that seem obviously right. I've had plenty. The problem is that we are not our customers. We know where every button is and why every section exists. Our customers are just trying to buy what they need and get on with their day.

A/B testing is how you close that gap. It turns opinions into evidence and lets customers tell you — through what they actually do — what's on their minds. In this case, they told us two things we wouldn't have guessed: don't interrupt me, and don't overwhelm me.

A few principles I took from this and still use today:

• Test the "obvious" idea too. The idea everyone agrees on is the one most likely to go unexamined.

• Design for the customer's moment, not just your placement. Where someone is in their decision matters more than how visible your feature is.

• Keep testing after you win. The first test tells you which direction works; the follow-up tests tell you how far to take it.

• Give tests enough time. Two sprints gave us enough traffic to trust the result instead of reacting to a few good days.

The takeaway for founders and product teams

Good product decisions don't come from being right about your own ideas. They come from being willing to find out you're wrong — quickly and cheaply, before you commit. When you're willing to experiment, your customers show you what they really need, and your job is to help them get it.

If your team is debating features based on gut feel, or you're not sure where to start with experimentation on your eCommerce site, I'd be glad to talk it through.

Share Your Pain Points & Let's Chat!

© 2026. All rights reserved.