A guide to product recommendations

If you shop online, you’re familiar with product recommendations. Every time you visit your favorite brand’s website, you’ll see a list of products recommended just for you based on your search history, your past purchases, and countless other data points. Behind the scenes artificial intelligence (AI) and machine learning algorithms are crunching these numbers, trying to determine exactly what you want to see in real time.

Product recommendations have played an outsized role in the rapid acceptance of ecommerce. Amazon and Netflix were early adopters of AI-powered product recommendations, investing millions to build and test their own proprietary solutions. Of course, this strategy paid off — and now a growing number of businesses have made product recommendations central to their digital business models.

Research suggests this is a winning proposition, whether you sell to consumers or businesses. When used correctly, product recommendations can increase the average number of items in online shopping carts by 68% — and conversion rates by an incredible 320%. Overall, product recommendations may account for up to 31% of ecommerce revenues.

Nevertheless, a large number of online businesses still don’t use product recommendations at all. And many of those that do manage them manually, hand-coding lists of top sellers on the home page and in other key locations. These lists have to be updated throughout the year as customer needs change, making the process extraordinarily time-consuming.

The good news? Significant strides in AI and machine learning technology mean automated product recommendations are more accessible and easier to use than ever, even if you don’t have an Amazon-sized budget.

What are product recommendations, and why are they important?

Product recommendations are product listings that are customized for individual website visitors based on data about them, their behavior and preferences, or the behavior and preferences of similar shoppers. You can provide product recommendations even for customers who do not have a profile or an account with your business. Common types of recommendations include:

• Bestsellers
• Hot right now
• Recommendations for you
• You might also like
• Trending
• Products like this

Product recommendations are important because they improve metrics like conversion rates, revenue per visitor, and average order value (AOV) while also improving the customer experience. In a survey of executives at more than 600 retail and travel industry companies, those that implemented personalized recommendations saw 10 times improvement in conversions, and 9 times improvement in AOV, while increasing revenue per visitor by 800%.

Adding product recommendations to your online store

Until recently, any business that wanted to harness AI to power product recommendations had two choices — build their product recommendation algorithms from scratch with a big team of developers or pay huge fees to third-party vendors and solution providers. Thankfully, things have changed.

Most ecommerce platforms now offer product recommendation engines right out of the box. There are also a wide variety of extensions and add-ons to choose from. If you already have an ecommerce platform, you’ll want to take a closer look at its support for product recommendations. If you haven’t chosen one yet, you’ll want to include personalized product recommendations — or compatibility with third-party recommendation engines — on your list of requirements.

Here’s a simple way to choose the right ecommerce platform or extension:

Source: Product Recommendations for Beginners (Adobe eBook)

Adobe Commerce, like other leading ecommerce platforms, comes with AI-powered product recommendations built in. Powered by Adobe Sensei, a framework that applies AI and machine learning to predictive challenges, our product recommendations feature automatically suggests relevant products based on shopper behavior as well as specific product attributes, popularity, trends, and more. It’s an easy way to add many different types of product recommendations to your entire website — and it eliminates the manual effort of identifying relevant and timely product affinities.

10 strategies for product recommendations that sell

Product recommendations can be used in almost any location on your site to connect with customers and encourage them to buy. Ten of the most effective product recommendation strategies used by Adobe Commerce customers are:

1. Highlighting best sellers

Promoting your best-selling products is one of the simplest and most effective forms of product recommendations. If many other customers like them, chances are your new customers will too. Customers also like to see that other people are buying a product, so a “bestseller” or “top seller” label is alluring. Plus, it can be great for your bottom line — 80% of most companies’ revenues come from 20% of their products.

The one caution we have is to avoid highlighting bestsellers that are out of stock or nearly out of stock.

How it works

Bestsellers recommend products most frequently purchased by shoppers within a predetermined period. It’s most effective when used on the home page, category pages, product detail, cart, and confirmation.

Suggested labels

These headlines and copy snippets can be used to describe recommendations on your website to attract customers’ attention:

• Most popular
• Bestsellers
• Top sellers

Helly Hansen uses product recommendations in Adobe Commerce to show off its bestsellers.

2. Focus on trending products

Highlighting products customers are likely already looking for during certain seasons or holidays can be highly effective. And customers love new arrivals. Returning customers are exposed to something they haven’t seen before. New customers see that your inventory is constantly updating, and it’s worth checking back regularly.

How it works

This tactic involves featuring recommended products based on the recent momentum of a product’s popularity across your site. AI aggregates browsing and purchase data across your site to determine and rank which products are the most recently popular with your shoppers. Because AI analyzes recent product momentum, this is an effective recommendation type for catalogs with high turnover. If your catalog is more static, it might not be as helpful unless the shopping patterns of your audience are highly variable.

Suggested labels

• Trending
• Trending now
• Recently trending
• Hot products
• Trending related products

BeerHawk uses product recommendations to surface trending beers.

3. Placing sales and discounts front and center

Customers are always looking for a deal. By recommending discounted products, you’ll give them a reason to purchase. You can also suggest a sale page (as opposed to a single item) or simply provide a pop-up with your current deal.

This strategy is great for encouraging impulse buys. Two-thirds of consumers say they have made an unplanned purchase because of a coupon or discount.

How it works

Beyond featuring discounted products in other areas of the site, you can recommend items with a low view-to-purchase or view-to-cart conversion. Your customers may get stuck on the price when it’s time to purchase, so promoting the same things with a discount may be what they need to convert.

4. Social proof and ratings

Showing social proof via quotes and ratings is a great way to build trust and prove value to a customer. Like many previous techniques, this tactic offers customers products that other customers are already happy with, encouraging the customer to not miss out on a great item.

How it works

Social proof works by recommending products most frequently purchased by shoppers within a predetermined number of days. Alternatively, you can recommend products with the highest view-to-purchase conversion rate. You can also feature “highly rated” products and display the star rating or have customer review snippets show up underneath a product recommendation.

Suggested labels

• Highest rated
• Popular products
• Hot products
• Most loved

Catbird uses product recommendations in Adobe Commerce to connect customers with their most-loved jewelry.

5. Further personalize your recommendations

Personalized recommendations begin with recommendations based on what shoppers like. Once you’ve made progress building out individual customer profiles, the personalization starts. Powerful personalization is based on transactional and behavioral data, plus financial, operational, and third-party data. This information delivers the most relevant, tailored product recommendations that increase sales.

Here are three ways it’s done today:

1. Location-based recommendations

Customers appreciate — and expect — relevant recommendations. A great way to provide that is by knowing their general location area. Depending on your franchise, industry, and location, there are plenty of ways to use this information once you have it.

How it works

Use macro-location details your customers provide — such as time zone, ZIP code, or even latitude and longitude coordinates — coupled with other visitor data to deliver product recommendations in their general location. For example, recommend a store near them, suggest a specific product that might be trending with shoppers in their area, or even recommend a product based on local weather.

Instacart is a great example. Its successful business model revolves around making locally relevant grocery recommendations.

2. Browsing history data

This is one of the most popular methods for making product recommendations, leveraging customer browsing data to provide targeted products.

How it works

Shoppers see recommendations based on their current and previous onsite behavior. These recommendations are most effective on the home page, where most shoppers begin their journey on a site. Based on browser history, you can also use recently viewed data to display products most recently viewed by the shopper.

You can highlight bestsellers and trending products for first-time shoppers who haven’t generated enough information to have a personalized experience. When the shopper interacts with the products on the site, you can adjust the recommended products to their behavior.

Suggested labels

• Just for you
• Recommended for you
• Inspired by your shopping trends
• Take another look
• Recently viewed

Catbird provides personalized recommendations for shoppers who regularly browse its website.

3. Customer purchase history

Using a customer’s purchase history is a powerful way to recommend products. It can be even more effective than browsing history because now you have a concrete data point for what the customer has bought. You don’t need to do much guesswork about how they’re browsing.

How it works

Recommended products are based on each shopper’s current and previous transactions and purchase history. They see items that are related to their previous purchases and interests.

Suggested labels

• Just for you
• Recommended for you
• Inspired by your shopping trends
• Frequently bought together
• Customers like you also bought

Lafayette 148 uses Adobe Commerce to surface product recommendations based on customers’ purchase histories.

6. Similar products, upselling, and cross-selling

There are a few methods for recommending products that are similar to what the customer is currently viewing. These recommendations are either social-proof-driven (based on what others liked) or product-driven. They work by helping shoppers see more options for what they’re looking for without having to navigate multiple pages.

Similar products

Recommend products based on similar metadata such as name, description, category assignment, and attributes. By evaluating the characteristics of the products being viewed, you recommend similar products in the same category.

Suggested labels: More products like this, Similar to this

Honeywell’s GoDirect Trade is a B2B marketplace for aerospace parts. The site offers recommended products that are similar to what the customer is viewing.

Cross-selling

Suggest products that complement shoppers’ original purchases. Highlight items that shoppers view disproportionately more often with the currently viewed product or recommend products that shoppers tend to buy disproportionately more often after considering the current product.

Suggested labels: Customers who viewed this also viewed, Customers who viewed this ultimately bought, Customers ultimately purchased

Olam Nuts highlights products that are related to the item on the page.

Upselling

Highlight a better (and more expensive) version of the product customers are viewing or have added to their cart. Shoppers see products that others buy disproportionately with the product they’re currently viewing.

Suggested labels: Get everything that you need, Don’t forget these, Frequently bought together

Visual similarity

Surface visually similar products to what the customer is currently viewing. This recommendation type is most helpful if images and the visual aspects of products are essential to the shopping experience.

Suggested labels: You may also like, We found other products you might like, Inspired by this style

The JCB B2B sales website offers similar — often more expensive — models to complement the product on the page.

Sugarfina uses product recommendations in Adobe Commerce to increase AOV.

7. Take full advantage of email

Product recommendations should live outside your website, touching other channels like email. Using the information in a customer profile — behaviors, transactions, demographics, and so on — you can send relevant product recommendations via email campaigns.

You can translate any of the product recommendation types we’ve discussed into email campaigns. For example, you can send recommendations based on what a customer previously bought or see if they want to purchase the same item again.

Flora and Fauna’s email campaigns are timely and full of relevant product recommendations.

8. Cart page recommendations

Product recommendations don’t end at the shopping cart. These pages can be another opportunity to add suggestions based on the products in the cart or items the customer viewed while browsing the site.

Recommending additional products in the cart is a great way to cross-sell related products or use other customer purchase data to suggest a “frequently bought together” item. You can also upsell here or even show off a similar product on sale.

Of course, some businesses report that product recommendations can distract shoppers from completing a purchase, so be sure to test different types of product recommendations and track your results.

Suggested labels: Frequently bought together, Still interested in these?, Customers also bought

Sephora gets great results by including product recommendations on the checkout page.

9. Optimize location and web design

You can have the best product recommendations — but if they’re not placed correctly or don’t look appealing, it will be hard to be successful. Ensuring that recommendations look good and are well-located on the best pages is crucial to their effectiveness.

The best locations for product recommendations are:

• On the home page
• In category pages
• Featured in individual product details
• In the customer’s cart
• On the confirmation page

The location on the page itself also matters. Recommendations on the home page are more successful when highly visible, while you want to be more subtle on a product page or a cart.
You can also feature recommended products:

• In a hero banner or carousel
  • These recommendations should activate and deactivate based on weather conditions, location, or inventory levels.
• Down the page
  • Include 3–6 individualized recommendations using any of the methods we’ve talked about. Any more than a few is too many.
• In search
  • This one can be a bit tougher to get right and relies heavily on AI. Intelligent search can understand shopper intent with natural language processing (NLP) and start providing suggested queries the moment the user begins typing. Your AI can deliver related recommendations to the search query.

Helly Hansen integrates recommendations into its product page designs so they don’t distract from the primary product listing.

10. Test and experiment with your recommendations

You won’t know which product recommendation strategies will perform best until you test them. Once you’ve added product recommendations to your web store, tracking their performance over time and making adjustments as needed is critical. As you introduce new products and customer expectations change, your product recommendation strategy needs to as well.

Here are some key metrics to track for each recommendation:

• Impressions. The number of times a suggestion is loaded and rendered on a web page.
• Viewable impressions. The number of times a recommendation is shown to a shopper and is visible.
• Click-through rate. The percentage of a recommendation’s viewers who click on it.
• Viewable click-through rate. The percentage of those shoppers who’ve seen a viewable suggestion and then click on it.
• Conversion rate. The percentage of shoppers who buy a product after interacting with a recommendation.
• Revenue. Revenue driven by a product recommendation, equal to the total value of products that were clicked on in a recommendation and were ultimately purchased.
• Lifetime revenue. Lifetime revenue driven by a product recommendation, equal to the total revenue generated by a suggestion from the time it was created.

Besides monitoring performance, use A/B testing to experiment with different product recommendation methods to see which strategies perform best in each page category.

Getting started with product recommendations

Personalized product recommendations are quickly becoming an expected part of the customer experience for everyone. And advances in ecommerce technology mean they’re within reach for virtually any business, including yours.

If you’re ready to build a product recommendation strategy — or just want to learn more — check out Beginner’s Guide: Personalized Product Recommendations for simple, step-by-step guidance.