Product Recommendations by Machine Learning: Tools and Examples

Product recommendations can significantly improve the financial performance of your business. According to Wiser’s statistics, they can increase the average check amount by 11%, reduce the number of abandoned carts by 4.35% and increase e-commerce revenue by 31%. You don't need to know each customer personally to make individual recommendations. Entrusting this task to artificial intelligence, you can implement it on any scale — even with dozens or hundreds of thousands of customers. 

At Claspo, we are working hard to introduce personalized product recommendation widgets into our toolkit. But for now, we'll be happy to tell you which brands are using this "you might like'' technique, what benefits it brings to the business, and, for that matter, which tools will help you implement recommendations on your website.

What is a Product Recommendation

Personalized product recommendations are sets of goods and services that are tailored to the needs, expectations, tastes and wishes of individual customers. They are compiled based on customer data such as their behavioral, demographic, socioeconomic and psychological characteristics, history of interaction with a brand, etc. To demonstrate how product recommendations work, let's look at a simple example from offline commerce. A seller sees a regular customer in the store and puts products of his favorite brand on a showcase to increase the probability of a purchase and increase the check amount.

Of course, in the digital space, this process is automated as much as possible. Machine learning product recommendations are created as follows:

1. A visitor goes to an online store page. 
2. The CRM identifies a visitor using an authorization system, a cookie or a linked social media account. 
3. All available information is downloaded from the database - data from the registration form, records of website behavior, key customer characteristics, traffic source, etc. 
4. This information is compared with the typical characteristics of customer base segments. It identifies whether a visitor belongs to a specific group. 
5. The machine learning platform studies customer characteristics and purchase history. Artificial intelligence finds hidden connections that are not visible to humans. Thanks to its high computing power, it is able to perform billions of operations per second. 
6. The product recommendation algorithm comes into play. Choosing the best products and services, it displays them in a special segment of an online store catalog, in a pop-up window or on a separate banner. 

Artificial intelligence can analyze data and make decisions in real time. If a user browses certain products and lingers on certain sections of a catalog after visiting a site, this information can be used to improve recommendations immediately. 

Benefits of AI Product Recommendation

Salesforce provides one of the most powerful arguments in favor of machine learning technology. The connection of the intelligent product recommendation module increases visitor engagement by 50% and increases the time spent on the website by 4.44 times. The technology also has other advantages, which we will discuss in more detail. 

Improved user experience

We live in a dynamic world where time is of the utmost importance. So, no one wants to waste it by browsing dozens of web pages searching for the required product. Applying machine learning-based product recommendations helps customers quickly find and buy what they need. This is crucial for promoting your business and building a positive brand reputation. According to McKinsey surveys, 67% of consumers say that personalized recommendations are one of the most important factors in choosing an online store. 25% will not consider offers from businesses that do not provide an individualized approach.

Increased sales

Barilliance's statistics show that site visitors who click on product recommendations using machine learning are 4.5 times more likely to add items to a cart and 4.5 times more likely to complete a purchase. Because of this, online stores owe 31% of their total revenue to these "you might like" banners.

Improving loyalty indicators

When customers quickly buy a needed product and encounter no obstacles in the sales funnel, they will want to come back to you. Accurate figures confirm this —according to Bloomreach, personalized recommendations increase the probability of repeat purchases by 56%. Along with sales, the audience of loyal customers will grow, forming a layer of brand advocates. 

Top 5 AI Product Recommendation Examples

Work began in the 2010s on using artificial intelligence in personalized product recommendations. Today, this technology is already well-researched, allowing the creation of a variety of application scenarios. It is available not only to large corporations with billions of dollars assigned for marketing budgets but also to small businesses. Here are examples of the effective use of AI for personalized recommendations. 

1. Spotify

Spotify is a leader in developing and using artificial intelligence for marketing tasks. The company has created highly sophisticated machine-learning models capable of analyzing hundreds of song parameters, from frequency characteristics to emotional lyrics and word count. The result of their work is personalized playlists, in which tracks are even diluted with comments from a virtual DJ.

Another great example of product recommendations from Spotify is podcasts. The company entered into an agreement with Google that allowed it to convert all available audio files into text format, analyze them with the help of Big Data services and determine each user's individual tastes. Spotify does not disclose the exact results of introducing new technologies. Still, we can see their impact on the company's overall profit, which has been growing by 35% annually over the past five years. 

2. Amazon

The world's No. 1 marketplace provides the best example of product recommendations in the e-commerce segment. According to McKinsey, 35% of its revenues are generated by artificial intelligence that creates customized product sets for website visitors. It owes its success to Amazon's comprehensive ecosystem, which includes streaming services, food delivery services, smart home appliances and more. With a huge amount of data to train ML models, the company can make accurate predictions about consumer needs and preferences.

Amazon uses a rather simple visual — a single line with recommended products in the catalog. This is a perfect solution for a large marketplace with hundreds of thousands of products.

3. Kappa

The Italian brand of premium sportswear connected multifunctional artificial intelligence to its online store. Its machine-learning model generates content for the following blocks:

• "You have previously viewed";
• "People also buy this product";
• "Add to your look";
• "You may also like";
• "Now in trend ". 

The results of the AI application were a 17% increase in conversion rates and a 40% increase in website visitor engagement. 

Although Claspo doesn't use AI in its functionality (yet!), you can already implement some of the above ideas with our pop-ups. For example, add our slider to product pages and showcase other items from your range that are trending or that would perfectly complement the product currently being viewed. This will help the buyer to create their perfect look and increase the average order value.

Try this template

4. Sambag

This Australian boutique is an excellent example of sending emails with product recommendations. It uses artificial intelligence to evaluate purchased items based on dozens of characteristics: color, cut, style, belonging to specific collections, etc. A certain time after the purchase is completed, it reminds you of the store by sending an email with an offer to complete your outfit with new products.

Sambag uses AI not only for recommendations but also for writing emails. Generative artificial intelligence personalizes emails by positively assessing customers’ tastes, congratulating them on their purchase or simply giving them useful advice.

5. Gym + Coffee

If you are looking for an example of how to apply product recommendations in a small-scale ecommerce business, pay attention to this case study. This Irish clothing manufacturer has entrusted artificial intelligence not only with the selection of products but also with the selection of widgets shown to online store visitors. Visitors can see pop-ups with individualized recommendations, trending products, bestsellers or promotional items depending on their behavior, traffic source and other key characteristics. AI has increased the brand's conversion rate by 18% and the average check amount by 5%.

Claspo also ensures personalization on your website! With our targeting, you can show pop-ups with individual content to visitors who came to your site from a specific traffic source or promotional campaign.

Moreover, if your ad appears in the search engine for a particular keyword, the Claspo web page widget can welcome only those brought to the site by this specific ad! For example, a person searches for "comfortable sneakers" on Google, comes across your ad, and visits your store. In this case, our widget may appear with text like: "Stop searching for comfortable sneakers! Go to the catalog and choose with a 20% discount." The result: you simplify the search for buyers and increase sales for you!

Try this template

Product Recommendations by Machine Learning Tools

Artificial intelligence and machine learning are exciting in theory, but in practice, they can seem complicated, especially for small businesses. Luckily, you don't have to be a math and programming guru to implement these technologies today. We will look at the top tools for creating product recommendations that are available to everyone. 

1. ShopAgain

ShopAgain is a simple service that can be easily connected to the most popular CMSs, including Shopify. It already has several templates for product recommendations based on user characteristics and behavior, sales volumes or purchase history. ShopAgain supports widgets as well as email, SMS or WhatsApp campaigns. The service has functions for generating shortened links and QR codes.

2. Wiser

Wiser is a professional product recommendation software that connects to Shopify in a few clicks and offers ready-made scripts for marketing campaigns: cross-selling and up-selling, welcome bonuses and special offers for regular customers. 

3. Recombee

Recombee is an AI product recommendation service suitable for any e-commerce business, from the smallest store to a giant marketplace. Among its unique advantages are social proof functions — these are widgets that showcase recently purchased products and the most popular products in a particular region. Recombee artificial intelligence learns in real time — every visitor action on your website increases its accuracy.