How to use 'Similar Product Recommendations' that connects the offline shopping journey to online.


We were able to systematically manage products through OMNICOMMERCE solutions and enhance online shopping experience through Visual Recommendations.
- THE HYUNDAI, PIC -

Problems the company wanted to solve

  • Continue offline store customers’ shopping experience online.
  • Improve the relevance of similar item recommendations.
  • The Hyundai wanted to solve the problem of poor customer satisfaction arising from the low accuracy of manually managed product data.
  • They also had a problem of having low work efficiency due to the lack of data experts.

Solution 1. Results after integrating the ‘Visual Recommendations’

  • The accuracy of similar item recommendations improved from 50% to 95%
  • The CTR has increased by 75% after improving the recommendation engine.

Solution 2. Results after integrating the 'Automated Product Tagging solution'

  • Significantly reduced attribute tagging errors.
  • Product filters have been diversified.
  • Succeeded in promotion by using highly accurate attribute data tagged by AI.

Amidst the fashion e-commerce boom, one of the three largest department store chains in South Korea, The Hyundai, needed a competitive edge as they looked to transition to the digital space. The brand was attempting to digitally transform their in-person department stores by bringing all of their retail shops online.

<Visual Recommendation>
Point 1. Similar item recommendation accuracy improved from 50% to 95%.

Product recommendation is a common strategy to capture customers for fashion e-commerce, and the most important aspect of this functionality is relevance. The Hyundai, who has transitioned the offline department store shopping experience to online has not reaped good results due to inaccurate metadata and the low-quality recommendation engine they were using previously. However, with “Visual Recommendations’ within OMNICOMMERCE, they were able to increase the accuracy of product recommendations.

  • Improved accuracy with high-quality recommendation engine

As a result of using a high-quality recommendation engine that recommends appropriate products at the right time, The Hyundai’s similar product recommendation accuracy improved from only 50% to 95%.

If the relevance of the recommended similar items is low, the suggestion may lead to churn rather than a purchase. The Hyundai was experiencing a high dropout rate due to inaccurate product recommendations during its online transformation. The ‘Visual Recommendations’ solution of OMNICOMMERCE has reduced the number of errors and helped optimizing the online transition.

  • Problem diagnosis by OMNICOMMERCE, our AI expert.

The Hyundai was having difficulties optimizing the online experience during the offline to online transition due to inaccurate product recommendations. The analysis of OMNICOMMERCE shows that the biggest reason was due to errors in product attributes and related product data, and a problem with the recommendation engine.

  • Improvement in product recommendation errors through 3-week PoC.

The ‘Visual Recommendations’ solution has resolved the errors in related product data and recommended products. They optimized the product recommendation function and improved the shopping experience with a highly accurate recommendation engine.

Point 3. 75% Increase in website CTR

The Hyundai has integrated the ‘Visual Recommendations” solution to provide highly accurate and appropriate similar item recommendations and improved the overall online shopping experience. The CTR has increased by 75% because the recommended products satisfied the customers’ tastes, and as a result, they were able to provide a seamless continuation of customers’ shopping journeys.

<Automated Product Tagging>

Point 1. Resolve product attribute errors

Every e-commerce owner has experienced delivering sub-par customer satisfaction service due to inaccurate product information. The Hyundai was not an exception. They’ve experienced the same difficulty during the digital transition of their customer shopping experience from offline to online due to errors in their product data. To solve this issue, they’ve integrated the AI-powered Automated Product Tagging solution.

  • AI-enhanced data technology.

The Hyundai’s inaccurate and insufficient product information due to the absence of sophisticated data entry technology was resolved with AI-powered technology. With AI, they were able to tag approximately 1,000 attributes within product images.

  • Product tagging in multiple languages

The Hyundai was also able to automatically add product attribute information in multiple languages. Their employees’ work efficiency has been boosted because our technology made it much faster to add new products.

Point 2. Provide various types of product filters

Because product information was registered by inexperienced practitioners, The Hyundai's products often did not contain sufficient descriptions and tags. This resulted in misclassified products and poor search accuracy. Automated Product Tagging has improved their product attribute classification standard and solved issues with inaccurate recommendations.

  • Refined search filters and improved search accuracy.

The Hyundai increased search accuracy using the rich product information retrieved from the solution. The filters were more deeply classified into attributes like ‘material’, ‘pattern’, ‘style’, and ‘detail’ to help customers find items they want easily.

Point 3. Successful promotion planning

Before introducing the solution, the merchandisers had to manually select promotional products while planning events. This was very time-consuming and complicated due to the lack of or insufficient product metadata. However, after automating tagging and registration, merchandisers were able to easily select the right products based on product metadata classified into consistent criteria, and as a result, the resources and time spent on planning promotional events have been significantly reduced.