AI Product Recommendations Explained + How to Set Them Up

Non-personalized marketing communications, including product recommendations, can hurt your ecommerce business, big time.

It’s simple—

No one can afford to give generic recommendations when customers stick with competitors offering personalized and positive shopping experiences.

For example, 67% of new visitors to your store prefer relevant recommendations and targeted promotions—a share too significant to risk losing:

While traditional rule-based product recommender systems can work relatively well for stores with a very limited product range, they just can’t compete with personalized AI-based recommendations.

This guide will be useful if you'd like to learn more about AI-based product recommendations in ecommerce and how to get started with this effective technology.

Go to sections:

• What are AI-powered product recommendations

• How AI recommendations work

• Rule-based vs AI-generated recommendations

• Examples of AI-based product recommendations

• How to add AI product recommendations to an ecommerce store

What are AI-generated product recommendations?

AI-powered product recommendations are suggestions tailored to each ecommerce website visitor based on their visited pages and product preferences. Using machine learning, they analyze billions of data points to show the right products at the right time—adapting to each visitor in real time.

AI product recommendations can appear in website popups, blocks embedded in web pages, dedicated sections in ecommerce store themes, and onsite feed, helping to boost revenue (various sources suggest a lift of 5% to 30%). By delivering personalized product suggestions across the entire onsite customer journey, businesses can increase the average order value and customer satisfaction.

Here's an example of AI-based product suggestions on Pierre Hardy, a French fashion store, shown in the onsite feed:

How do AI product recommendations work?

AI-powered product recommendation software analyzes what shoppers do—like what items they click, view, or buy—and finds patterns in that behavior.

To make the recommendations as relevant as possible, AI algorithms also process and analyze historical data on purchases, products, customers, prices, etc., to suggest items that similar shoppers liked or that go well with what someone’s already interested in.

This process has a few stages:

• Collecting information. A profile of a customer is created using data like demographics, website browsing history, purchase history, and product preferences

• Analyzing and learning. AI algorithms filter and analyze the most important data in customer profiles to make predictions of what they might be interested in

• Providing personalized recommendations. The suggestions are displayed to the customer on the website on every visit

• Learning and adapting. When a customer returns to the website, the algorithm dynamically updates the profile based on their latest interactions

A few machine learning algorithms are commonly used to predict future purchases:

• content-based filtering

• collaborative filtering

• deep learning

• hybrid

Each of them takes a unique approach to analyzing data and generating results:

How do AI-generated product recommendations differ from traditional ones?

There are two main types of product recommendation systems: rule-based (traditional) and AI-based. Both help customers discover products, but they work in different ways.

Let's compare.

Rule-based systems follow fixed rules set by the user. For example, they might show best sellers, trending items, or products that share specific tags or attributes, e.g. "Show products with the same tag as the item being viewed."

AI-powered systems autonomously analyze visitor behavior on the website and combine it with historical data to generate personalized suggestions. Each time a customer interacts with the site, the AI collects data, spotting patterns over time, like suggesting jackets to a visitor who buys casual t-shirts or recommending shoes based on past purchases.

Best practices for implementing AI product recommendations in ecommerce

• Use different campaign formats. Consider displaying recommendations in a few formats, such as popups and embedded blocks. They help you reach different sections of your website and engage shoppers at different stages of the buying process

• Run A/B/n tests. Using them on product recommendation campaigns (including tests against a control group) helps you understand their true impact by comparing layouts or situations where users who don’t see recommendations

• Consider using multiple categories of recommendations. Your store needs to display bestsellers, trending items, "others also viewed," personalized picks, recently viewed items, and "frequently bought together" categories to cover the entire onsite customer journey from discovery to purchase

• Target returning visitors, too. These visitors are already familiar with your brand, are more likely to convert, and may appreciate personalized suggestions based on their past visit(s). You can use what they’ve browsed or purchased to show more relevant products and boost average order value

• Track revenue. Use this feature in your CRO tool to know the impact of your recommendations in terms of the monetary value they bring to your business

Examples of AI-powered product recommendations in ecommerce

Pierre Hardy — personalized picks

Personalized ecommerce product recommendations based on the browsing history are one of the best ways to improve shopping experience because they give relevant items similar to those viewed by customers.

Pierre Hardy has a good strategy for us to review—this store displays personalized picks to new and returning customers, only after they visited at least four pages.

On the fifth page a customer visits (in my case, it was the homepage), they get a message with three recommendations:

What's notable about this AI-based product recommendation example is that it appears only on exit, meaning Pierre Hardy doesn't interrupt the browsing experience when we browse the products and may not need help with choosing.

Here's a closer look:

This type of campaign is ideal for thoughtful, independent shoppers who like to browse a lot of products before making a decision. These customers appreciate relevance and personalization but prefer to explore at their own pace without being interrupted.

émoi émoi — recently viewed items

Recently viewed products help increase online sales by reminding shoppers of items they showed interest in, making it easier for them to return and complete a purchase.

émoi émoi uses AI to display recently viewed products in real-time and encourages visitors to check them out again with the onsite feed.

The display order of items is based on an analysis of their sales performance combined with the visitor's browsing history:

The feed is housed within the website's header, so shoppers get a familiar, easy-to-spot "unread message" style notification:

Clicking the "bell" opens the feed with the messages.

Product recommendations are always displayed on the top to ensure easy access:

This way of displaying recommendations has some advantages:

• shoppers can add the recommended products to the shopping cart directly from the feed in one click

• the list is updated in real time and always has the best-performing items at the beginning based on AI analysis

• the feed saves the visited products, so a customer can find the list in the feed even after they leave the website and come back later

• customers can access the list of recommendations from any page—all they need to do is click the "bell"

And, last but not least, this strategy is effective—

Over 11% of customers who clicked the AI-generated product recommendations in the feed at émoi émoi made a purchase.

As a result, the brand also increased the average order value by 23%.

By delivering personalized suggestions in real-time with AI and maintaining easy access to these options even during the next visit, émoi émoi made the shopping experience smoother and more enjoyable.

Read the case study if you'd like to know more.

How to create product recommendations with AI for an ecommerce store

Getting started with AI product recommendations can be easy—and you don't need any coding skills or experience. Plus, so many product recommendation platforms available, you'll definitely find one that suits your business.

For this tutorial, I'll use Wisepops to show you how to create an AI product recommendation popup campaign.

If you'd like to follow along, you can start completely free—no credit card needed. Or, get our native Shopify app if you prefer.

Steps:

• Choose a template

• Add relevant products

• Choose when to display the campaign

• Customize the design

• Track performance and adjust

Step 1: Choose a template

In the main dashboard, go to Campaigns > Popups > New popup campaign.

Choose the "product recommendations" category in the template library:

I selected a template featuring similar products ("Others also viewed") to show recommendations based on what a customer is currently viewing.

But you can easily choose a different category anytime by going to the AI Recommendations section under Blocks:

Step 2: Add relevant products

No matter which category you select, click on the product recommendation block to start customizing:

Once you do that, you'll see two tabs in the left panel: Products and Add to cart.

In Products, you can add recommended items and set their total number of items.

To add products, enter their IDs, separated by commas:

id:4567770, id:4572093

Here’s how you can exclude products (no apostrophes):

Product ID: e.g., "id:435342655" (include "id:")

Exact title: e.g., "White summer shirt"

Partial title: e.g., "sweater" (matches "Yellow summer sweater" and "Red winter sweater")

More on adding products.

Next up—

In Add to cart, you can choose to show the Add to cart buttons on the campaign for easy access, and even customize the confirmation message:

With the recommended products added, AI will take care of the rest.

Step 3: Choose when to display the campaign

To give the algorithm some browsing data to improve the quality of the recommendations from the very first day, go to Display rules > Triggers and add a four-page delay.

That means that the visitors will see the recommendations on the fifth page they visit on your website:

Step 4: Customize the design

Now, we need to add some copy and colors to the campaign to align its appearance and tone with your website.

Here's how you can customize the design—

First, the texts. Change the heading and the subheading by choosing that element and using the personalization menu at the top of the preview:

Next, do the same with the colors of the section with recommendations:

And lastly—

Let's choose a screen position.

As you remember, Pierre Hardy's product recommendation campaign was displayed in the top left corner of the screen—you can go with the same position or choose another one.

Go to Design > Position to view all the positions:

The only thing left to do is activate the campaign by adding a short code snippet to your site.

Here's a quick video that'll walk you through this process:

Step 5: Track performance and adjust

If you're using Wisepops' native Shopify app, it will track the revenue from AI product recommendations automatically.

You'll see the revenue indication in the popup dashboard immediately after you publish the campaign, along with displays, clicks, and orders:

Thanks to this info, you'll know exactly how impactful your AI product recommendations are for your store.

This help doc for goal and revenue tracking will be helpful if you have any questions.

You'll also track your results in a dedicated dashboard:

If you'd also like to try different ways to display product suggestions, you can create an A/B test by clicking the corresponding button in the popups dash:

To set up the test:

• add a copy of the existing campaign as the second variant

• choose the percentage of visitors to show each variant to

• customize the second variant

All these settings will be available in this window that appears after you choose the A/B test button:

Have questions? This guide on setting up A/B/n experiments will help.

Summary

AI product recommendations are a crucial strategy for converting visitors into customers. By integrating tailored suggestions into your ecommerce site, you can significantly enhance user experience and increase sales.

Consider these resources from our blog if you'd like to keep learning about how to grow your business:

• How to add a popup campaign on Shopify

• Ideas for coupons and discounts for ecommerce

• How to promote a new product effectively