In this case study, we will discover how home appliances and electronics retailer Comfy boosted its email newsletters performance with personalized product recommendations. You will be able to learn about:

• How to set up web tracking and develop a personalized recommendations algorithm

• What algorithm did Comfy use in their newsletters

• How to set up channel attribution and track the results

• Comfy’s plans for the future

The peculiarity of this case is the customization of issuing recommendations based on the priority features of the goods. The functionality of Yespo allows Comfy to show customers first priority products ("Best") and to adjust their quantity.

The Results

2 months after the completion of setup (April 2023), Comfy achieved the next results:

• The average share of orders from personalized blocks in general newsletters of the "Product of the Day" category is 49%.

• The average share of clicks on personalized blocks is 41% of all clicks.

• The share of recommendation blocks in trigger-based letters income was 13% of the total income from the email channel.<

• The share of orders from personalized blocks in triggers is 20% of the total number of orders from these letters.

About the Project

Comfy is one of the largest Ukrainian retailers of household appliances and electronics. The company constantly introduces new technologies to remain the market leader. With this goal in mind, they decided to add product recommendations to triggered letters and promotional emails, and in this way, increase sales.

To achieve it, Comfy needed:

• Subscribe to the Yespo Professional pricing plan.

• Create the personalized algorithms request considering their business specifics and transfer it to the Mercury Team through the customer success manager.

• Compile letter templates and hand them to the Yespo integration team for customization.

Web tracking connection and algorithm setup

General newsletters to a wide audience bring in a significant share of sales, but they can also cause a growing churn rate, so Comfy decided to improve customer experience with personalized product recommendations based on the customer’s preferences. Yespo CDP collects and enriches this data thanks to the web tracking technology that works on the Comfy website.

Web tracking is a technology for personalizing communication with users. In order for it to start working, it is necessary to install a script on the site, upload the feed to the Yespo account and configure events such as "Product View", "Cart Update", "Sale", etc.

Product recommendations are suggested by artificial intelligence. The script, in turn, is needed to bind the visitor's cookie to a specific contact and send a message to the media channel available to them in the case of a certain event. Web tracking combines collected data with contacts, and AI suggests the necessary product recommendations based on selected algorithms.

Yespo offers more than 150 ready-made algorithms for common business purposes. For even more precise recommendations, there are filters in the algorithm settings. For example, it is possible to set up:

• The products that should always be shown

• The products with specified properties that should never be shown

• Priority products that should be shown more frequently

• The products that can be shown for expanding the range

For example, if you need to show more products from the "Accessories" category with a price from UAH 999 to UAH 13,999, the set of rules will look the next way:

Comfy’s Product Recommendations Algorithms in Email Letters

Triggered letters

A personal algorithm is an individual product recommendation mechanism based on the needs of a specific business. This service is available to every client subscribed to the Professional pricing plan. For a quick connection, you are welcome to contact your Yespo customer success manager or email us by sales@yespo.io.

Mass Mailings

We have added product recommendations to the regular "Product of the Day" and "Day/Night Sale" newsletters.

AI algorithms substitute products into recommendation blocks when sending the email. The selection of products is based on the following data about the customer's behavior on the site:

• views,
• search queries,
• purchase history, etc.

Analyzing this data, the neural network detects behavioral patterns and preferences of a person, and then predicts what will potentially interest them. The algorithms work according to the following rule: always add 4 products with the "Best" property to the recommendations.

"We prioritized the products with the “Best” property in order to encourage customers to buy top products of the network. Comfy offers the best discounts, additional bonuses, and delivery for UAH 1 on these items."  — Oleksandr Feller, CSM Yespo

Each recipient receives truly personalized content with relevant offers. If behavioral data is not enough, instead of AI-generated personal recommendations, a block with bestsellers (up to 6 products) is automatically added to the email.

Already in the first two months of use, personalized product recommendations created in Yespo showed good results. The average share of clicks for these blocks was 41% of all conversions from the email to the site. The average share of orders from blocks in general newsletters of the "Product of the Day" category reached 49%.

The Comfy and Yespo teams continued to analyze the effectiveness of personalized product recommendations in bulk campaigns. In the fall of 2023, for a month, we compared the indicators of two variants of promotional emails:

1. With blocks generated by Yespo AI algorithms.
2. With standard product selections manually added by Comfy marketers.

Both options of bulk campaigns were sent to the same segment of subscribers. During the analysis, we focused on how many contacts from all users who switched from email to the site clicked on the recommendation block. It turned out that blocks with personalized product offers bring more potential buyers to the online store than standard selections.

Thanks to the Yespo recommendation blocks, mass email mailings to large segments, which usually lacked personalization, are filled with unique content selected for the special requests of each client.

Advantages of using product recommendations in email newsletters:

• automation of the process — there is no need to manually create product selections for different customer segments;
• relevance of offers — artificial intelligence can more accurately identify the needs of each person based on their behavioral data.

The time that marketers will save on AI tasks can be used for creation and A/B testing.

Setting Up Attribution and Tracking Results

In the "Revenue from campaigns" section in the Yespo account, you can set up attribution for all communication channels. 

These are the standard default settings where income is always credited after a set period of time. However, Comfy's experience shows that customers tend to order in a longer period of time after interacting with an email.

In order to correctly assess the impact of letters with personalized product recommendations, UTM tags were used. In particular, unique values are transferred to UTM-term, which allows for tracking sales from personalized letters in Google Analytics, if the last touch was a week, a month, or more time ago.

Comfy’s Future Plans

"We aim to make more personalized offers for our audience, and the relevance of these offers is the highest priority for us. That’s why we are moving towards ensuring brand-customer interaction at every touchpoint with our store. Therefore, the next step will be setting up mobile tracking to spot triggers and deliver product recommendations in the Comfy app." — Tetyana Novosadska, Comfy's leading specialist in personal communications. 

Get professional expertise