“Customers also bought.” “You might also like it.” “Frequently bought together.” You’ve seen these widgets on basically every ecommerce site you’ve ever shopped on, which probably means your own Shopify store has one too. But here’s the question that doesn’t get asked enough: is it actually doing anything, or is it just sitting there because every theme comes with one pre-installed?
It’s a fair thing to be skeptical about. Recommendation widgets have become such a default part of ecommerce design that a lot of stores add them and never look back, no tracking, no testing, just an assumption that “personalization is good, so this must be working.” Sometimes it is. Sometimes it’s dead weight taking up valuable real estate on your product page. Let’s dig into what the data actually says, and how to find out where your store falls.
What the Research Actually Shows
The short answer is: dynamic product recommendations can move average order value meaningfully, but the size of that lift varies a lot depending on the study, the industry, and how the recommendations are implemented.
The clearest number worth anchoring comes from research published on ScienceDirect, which found that a single interaction with a personalized recommendation was tied to a 369% increase in average order value, with the effect continuing to build the more a shopper engaged with additional recommended items. That’s a striking number, and it’s worth remembering it represents a ceiling, not a guarantee, your results will depend heavily on your product catalog, your traffic, and how well the recommendations are actually built and placed. But it makes one thing clear: when a recommendation genuinely lands, the impact on AOV isn’t marginal.
Why the Lift Isn't Automatic
Here’s the catch that number glosses over: recommendations only lift AOV when people actually engage with them, and most shoppers never click one at all. That’s really a design and relevance problem, not a “recommendations don’t work” problem, recommendations work incredibly well for the people who click them, so the real challenge is getting more people to notice and click in the first place.
A recommendation widget stuffed at the bottom of a page nobody scrolls to, showing generic bestsellers instead of genuinely related items, isn’t going to move the needle, no matter what the research says. This is where CRO Design actually matters more than the algorithm behind the recommendations. Placement, visual prominence, and relevance to what’s already in the cart all determine whether that widget gets seen at all.
Rule-Based vs. AI-Driven Recommendations
Not all “dynamic” recommendations are built the same way. Some stores use simple rule-based logic, “if someone views Product A, show Product B”, set manually by a merchandiser. Others use AI-driven engines that analyze real-time browsing behavior, purchase history, and patterns across the entire customer base to generate suggestions on the fly.
AI-driven recommendations tend to outperform static, rule-based ones because they adapt to the individual shopper instead of showing the same pairing to everyone, rather than relying on a single fixed rule that treats every visitor the same way. But AI-driven doesn’t automatically mean better for every store. Smaller catalogs with less purchase history may not have enough data for an AI engine to meaningfully outperform a well-thought-out manual “frequently bought together” pairing.
This ties into a bigger shift happening across ecommerce broadly, worth understanding through personalization trends in ecommerce and how AI in ecommerce conversions is reshaping what shoppers expect from a site before they’ll trust it enough to buy.
Why People Click a Recommendation at All
The psychology behind this matters just as much as the technology. Shoppers don’t click “you might also like” because it’s clever, they click because it feels relevant, reduces the effort of searching, or nudges them toward a decision they were already leaning toward. That’s part of why ecommerce psychology triggers to increase sales, things like social proof, relevance, and reduced decision fatigue, explain why “frequently bought together” tends to convert better than a generic “recommended for you” carousel showing unrelated items.
How to Actually Know If It's Working for You
This is the part most stores skip. It’s easy to install a recommendation app, glance at revenue a month later, and assume it’s helping, without ever isolating whether the widget itself is responsible for the change.
Start by checking whether people are even seeing and clicking your recommendations in the first place. Pulling the right essential GA4 metrics for Shopify, like widget click-through rate, assisted conversions, and revenue attributed to recommendation clicks, gives you a real baseline instead of a guess. If the data shows barely anyone is interacting with the widget, no algorithm upgrade is going to fix that; the problem is visibility or relevance, not technology.
From there, it’s worth running a proper CRO Audit to see exactly where the widget sits in the customer journey and whether it’s competing for attention with other elements on the page, or genuinely helping guide the purchase decision.
The Bottom Line
Dynamic product recommendations absolutely can lift AOV, the data supports that consistently, even if the exact percentage varies. But “can” is doing a lot of work in that sentence. The lift shows up when recommendations are genuinely relevant, well-placed, and actually seen by shoppers, not just installed and forgotten.
Before you assume your recommendation widget is pulling its weight, check the data. Look at whether people are clicking, where the widget sits, and whether it’s actually influencing the order or just quietly taking up space on a page that could be doing more for your conversion rate.
