Although Zoot’s 2 million monthly visits (+Bibloo’s 2.5 mil monthly visits) are impressive, more importantly, the brand itself managed to gain the “love-brand” status in the fashion community.
So, when the client decided to take its site search to a new level, it made this project a big challenge and committed us to improving the precision of the search results and enhancing the user experience.
The results quickly showed Luigi's Box's superior robustness
Search is a key part of our e-shop, so we carefully selected a new solution. We A/B tested Luigi's Box against our own and other tools. The results quickly showed Luigi's Box's superior robustness, and further A/B tests didn't find a better alternative.
David Šandera
Chief product and technical officer, Zoot (digital people)
The main challenges:
- Increase the overall search results accuracy
- Manage to provide relevant autocomplete search suggestions for a huge number of products (1 million) while maintaining a fast load time.
- Facilitate the product discovery process with customized autocomplete showcase and dynamic faceted filters.
Analyzing the user behavior
Data showed us that Zoot’s visitors are three times more likely to convert when using search. With around 10-11% search usage, it’s been more than clear that it would be a missed opportunity to only provide the technology without adapting it to the specifics of the fashion industry.
That’s why our first step aimed to understand how users interact with the website and the search itself. To do this, we implemented Luigi’s Box Analytics, which helps us understand how users are searching for different types of products by analyzing:
- The most important product attributes (e.g., colors, gender, style/cut, brand, etc.)
- Usual synonyms and typos (jeans, denim, blue pants, etc.)
- The most common reasons for the no-result searches
Fine-tuning the search
Afterward, we started to fine-tune our search to reflect the data in the search results relevancy and product rankings.
One of the main targets was to improve the accuracy of color-related search terms (e.g., blue jeans, white cardigan, etc.).
Identifying the most common synonyms, slang words, and misspellings was crucial to decrease the number of no-search results and increase accuracy.
Last but not least, we cooperated intensively with the client to improve the product feeds to get the most out of our search by providing full and accurate product data based on our findings and requirements.
Design and usability
The design of autocomplete, search results, and faceted filters aims to adapt to website design based on the client’s brief.
As the client thought of the search as the fashion showcase, we created the full-width autocomplete bar design with a focus on a compelling visual presentation of the products.
Autocomplete showcase also contains recommended categories, top brands, and similar queries related to the search terms.
Since data showed us that the most important filter attribute in search results is the brand, we prepared a quick visual filter (logos) of the most popular brands matching the particular search term along with the regular faceted filter.
We also added recommended categories to the search results page since a lot of the search terms (e.g., “white top”) are often relevant to several product categories.
A/B test and the results
Since Zoot was using a competitive solution at the time, they decided to launch an A/B test with the 50:50 split.
The A/B test started on August 17th and was evaluated after one month, but even after the first weeks, the results started to show an improvement in favor of Luigi’s Box.
After several prolongations, the test was stopped after three months. The results revealed that Luigi’s Box Search and Autocomplete:
+9.14%
Increased add-to-cart conversion rate
+2.57%
Increased purchase conversion rate
This case study showed us how important it is to choose an individual, data-driven approach to get the most out of the technology’s potential.