NewRecommendations on your thank you and order status pages, and moreThank you page and moreSee what's new →
Laptop on a glass desk showing an analytics dashboard with charts
Educational

What the data actually says about product recommendations on Shopify

July 8, 2026 · 4 min read · Updated October 26, 2026

Most numbers used to sell product recommendations trace back to a press release, a single vendor test or nothing at all. A few hold up because they name the source, the sample and the year: Salesforce's 26% of revenue from the 7% of visits that clicked a recommendation, Nosto's 10.2% higher order value in one cart test, Rokt's cart vs checkout comparison and Barilliance's 300-store average. Here is what each one measured, so you can cite it without stretching it.

Search "do product recommendations increase sales" and the same handful of numbers turns up across dozens of posts, usually without a link to a report. Some are real and still checkable. Many have been stretched or separated from the study that produced them. Before you decide what to test on your store, it helps to know which is which.

Why most recommendation stats don't survive a check

Two problems account for nearly every bad stat in this space. The first is segment collapse: a lift measured only among shoppers who clicked a recommendation gets rewritten as if it applied to every visitor. The second is source loss: a number gets attributed to "a study" with no link, no sample size and no year, so nobody can check whether it ever held.

The "Amazon gets 35% of revenue from recommendations" claim

This claim shows up in pitch decks, blog posts and app listings, usually as a current fact about Amazon. Its origin is a 2013 McKinsey article that said 35% of what consumers purchased on Amazon.com came from recommendations. McKinsey published no method for that number, and Amazon has never confirmed it. Over the years "purchases" has often been upgraded to "revenue", which is a different claim.

Recommendations very likely work at Amazon's scale. This particular number just belongs in a sentence like "In 2013, McKinsey reported that...", as ecommerce history, not as a benchmark to plan a store around.

The numbers that do hold up

These come from named companies, with sample sizes and reports you can still find. Each is worth citing with the segment it measured, since that's the part that usually gets dropped.

According to Salesforce, which tracked more than 150 million shoppers across 250 million visits and $550 million in orders, the 7% of visits where a shopper clicked a product recommendation accounted for 26% of revenue (Product Recommendations report, 2017). The same report found that shoppers who clicked a recommendation spent 5 times more per visit, and that 52% of their orders included a recommended product. Read it carefully: revenue was concentrated among visits that clicked, which is not a lift for every visitor who saw a recommendation.

According to Nosto, shoppers at the fashion brand Bandier who saw cart recommendations spent 10.2% more per order in a 25-day A/B test. It's one retailer and one test, but it's a controlled comparison, which is rarer in this space than the marketing suggests.

According to Rokt Aftersell's Revenue Leak Report 2026, across roughly 50 million cart sessions, upsells in the cart brought in 4 times more revenue per session than offers at checkout. It's one of the few data points that compares placements directly.

According to Barilliance, across 300 stores an average of 12% of sales were attributed to product recommendations, and up to 31% on some sites. The range matters more than the average: the real number depends on catalog, traffic and placement.

Two more, with their scope stated. McKinsey's broader personalization research (2019 and 2021) puts typical revenue lift at 5 to 15% (and 10 to 15% in the later piece), but that covers personalization in general (pricing, email, site content and recommendations together), not recommendations alone. Dynamic Yield's published cases show smaller, placement-specific results: 4.2% higher revenue per user from product page recommendations for e.l.f., and 15% from add-to-cart recommendations for LUISAVIAROMA.

What to test first

The data points the same way: the cart has the strongest comparative evidence (Rokt's placement comparison and Nosto's test), so it's a reasonable first placement to try, followed by the product page.

Placement isn't enough on its own. Baymard Institute's UX research, a separate finding about design rather than revenue, found that 52% of sites fail to show relevant products in cart cross-sells and 58% show only one type of suggestion. A well-placed recommendation still fails if it's the wrong product.

How to measure it on your store

None of these numbers will match your store. They tell you what is plausible. Your own numbers come from three counts, in this order: how often recommendations were seen, how often shoppers added a recommended product, and what those products sold for in paid orders. Then compare your average order value and the share of orders with two or more items before and after. The tracking guide explains what each app counts as a sale.

For what goes wrong with apps in practice, see 879 bad reviews of upsell apps, sorted by complaint.

What to expect on a small store

Most of these numbers come from larger retailers with heavy traffic and deep catalogs. According to Rokt Aftersell's 2026 data, stores with under 1,000 orders a month saw $0.20 to $0.28 in revenue per transaction from Aftersell's upsell offers. That's a modest number, and a more realistic one for a small store than the big percentages quoted for enterprise retailers.

That's also why Dropr has a free plan (up to 1,000 recommendation views a month): you can see your own store's number before paying anything.

Frequently asked questions

How much revenue do product recommendations generate?

There's no single number. Barilliance's 300-store study found an average of 12% of sales attributed to recommendations, up to 31% on some sites. The reliable answer for your store is your own data over a few weeks.

Is the Amazon 35% statistic real?

Not as it's usually presented. It traces to one line in a 2013 McKinsey article about what consumers purchased on Amazon.com, with no published method and no confirmation from Amazon. Cite it as history, if at all.

Related reading

Try Dropr on your store

Free up to 1,000 views a month. $19 every 30 days for unlimited views.

Install free

No theme file edits · Pricing · What's new