Product Recommendations: How to Save $5000 on AI Development

Posted in: Development Hacks
  • Product recommendations make up 26% of e-commerce revenue while accumulating only 7% of the traffic.
  • Buyers who engage with a recommended product have a 70% higher conversion rate.
  • 54% of merchants claim product recommendations to be the key driver of the AOV (average order value).

Looking at these impressive statistical numbers, it seems logical to invest in product recommendation algorithms and engines. So today, we’ll discuss the role of AI and best practices you can use for any product recommendation system.

AI, or not AI, that is the question

Today, AI is quite popular as the base for product recommendations engines. Companies like Amazon or Netflix have used them for a long time now and have proven their efficiency. Moreover, users demand as personalized recommendations as possible. Research shows that 72% of customers want to engage with personalized content only.

To develop personalized product recommendations, you can either rely on experts' opinions and data analysis performed by humans or use AI-based engines. Both these ways have their pros and cons.

Both algorithms and humans can be wrong. So, as a rule, it’s usually recommended to use AI for practical products and human opinion for experiential, sensory, and enjoyable products. For example, robots can help with choosing warm winter coats, but humans will know better about beautiful and fashionable coats.

However, it’s often overlooked that even though AI isn’t a completely new technology, it is still expensive. Development of the MVP of recommendation engine can cost from $5.000 to $15.000. Not every business owner is ready to invest so much money right away. And who knows how much engine maintenance will cost.

As a budget-friendly alternative, you can use tools that allow you to create rules for product recommendations and offer products based on your expert opinion and the statistical data of customer behavior you already have.

Best practices of product recommendations

Say you decided to save on AI development and build product recommendations by yourself. How to keep recommendations on top of effectiveness? Here are 5 best practices.

#1. Place product recommendation block above the fold

Recommendations have 2 main aspects that influence their effectiveness: the products you offer and the position of the widget. We are used to seeing such widgets somewhere at the bottom of the page. But tests show us that if you place this block above the fold, it will almost double its effectiveness (1.7x):


How to move related products block in Magento 2: By default, Magento 2 offers only one position of related products - under the product description. You can move it to a different place using the Automatic Related Products extension:


#2. Offer bestsellers to new visitors

It’s a good idea to set up different product recommendations for different customer groups. But what to do with customers you know nothing about yet?

It is recommended to display your best selling items to new visitors. Moreover, bestsellers are an ideal choice for your homepage:


Thus, you can grab attention and engage with customers that were just browsing without anything specific in mind.

#3. Display “Bought Together” products on cart pages

Cart pages are a great choice for upselling, especially if you want to increase your average order value. If you have such data, it’s better to show “Bought Together” products on the cart, but “Similar products” will have high effectiveness too.

How to create the bought together rule in Magento 2: To display the bought together products on the shopping cart page for any product added to the cart, you need to create a new related product rule using the Automatic Related Products extension:


Note: use conditions if you need some additional filters to narrow the list of the related products.

#4. Offer complementary products together

Product bundles are another type of product recommendation, and a variation on the bought together rule. The main difference is that bundles usually come with a discount. For example, buy sneakers and get a 5% discount on the socks, or get a total look with 30$ off:


How to create bundles in Magento 2: To create a bundle pack using the Automatic Related Products extension, go to Catalog > Bundle Packs and click the Add New Pack button. Then fill in the settings:


#5. Experiment, test, and analyze

Testing different recommendation strategies is the key to success. Comparing them, you can identify what works best for your audience and maximize profit:


Experiment with user data and segment your audience into the groups. Thus, you can provide them with a more personalized experience. And it will directly affect your conversion rate and the amount of returned customers.

To sum up, here are the main tips for you to follow when you add product recommendations, whether you use AI or not:

  • place product recommendation block above the fold
  • offer bestsellers to new visitors and on the home page
  • display “Bought Together” products on cart pages
  • offer bundles with complementary products
  • experiment with different strategies

And remember, it’s not obligatory to use AI to achieve impressive results.

April 28, 2022
April 29, 2022
March 25, 2022
Leave your comment

Your email address will not be published

This blog was created with Amasty Blog Pro

This blog was created with Amasty Blog Pro

© 2009-2023 Amasty. All Rights Reserved.