The UK Rank algorithm optimizes search results in real-time based on customer behavior
Until now, most search engines only displayed product names that match queries (entered search keywords). These engines prioritize a “matching score,” a score that changes depending on matches with information such as the product title and descriptions, with scores for product title matches being high and scores for description matches being low. Thus, when site administrators wanted to control the search results, they had to rewrite product titles and descriptions.
Search results with this type of primitive search algorithm do not change at all depending on the pages seen by customers, the links clicked, or with products being placed in carts or purchased. The search results only change when products are added, deleted, or when product titles or descriptions are changed; that is, only when text information is changed. The operation of such primitive search engines have no relationship whatsoever with customer behavior. It is, therefore, natural that eCommerce sites using such search engines face difficulties in achieving higher sales.
We felt that customers could not optimally find the products they were looking for in this way. That is why Universal Knowledge’s search algorithm, “UK Rank,” logs all customer behavior and automatically optimizes search results through a comprehensive data analysis of what customers want. What results is a search engine in which manual changes to product titles and descriptions are no longer necessary, and seasonal and popular items are automatically displayed at the top of search results, making the search process easy as well as increasing sales.
Revenue Optimization
Customer search behavior is the behavior that they seek the products they want.
It makes intuitive sense that sales would increase if the exact product a customer is searching for were to be displayed in the search results. Universal Knowledge’s search algorithm, UK Rank, optimizes searches to make it easier for customers to find the products they want, while simultaneously increasing sales for eCommerce sites. UK Rank is an original algorithm developed by Universal Knowledge and based on the fundamental concepts of which search results should be provided in order to increase relevance and sales.
Additionally, depending on the eCommerce site, variables such as product quantity, price range, and purchasing frequency also differ. This is why Universal Knowledge is constantly fine-tuning its basic algorithm to have it adapt to the real-world characteristics of each eCommerce site. As no two eCommerce sites share the same set of parameters, the results of fine-tuning are then confirmed through bucket tests to make further improvements.
It is Universal Knowledge’s policy to conduct bucket tests without fail. While the relevance of queries to products can be determined in advance, sales and conversion indicators cannot be known until new search results are actually displayed to customers. Furthermore, as customer behavior differs for each eCommerce site, there is no guarantee that behaviors from one site will be observed on another. This is why we always perform bucket tests when improving or fine-tuning our algorithm.
Big Data + Real-time
In order to provide search results that have high relevance and increase sales, customer behavior must be recorded in a log. Universal Knowledge begins by filling the main pages such as “Search Results,” “Product Details,” “Shopping cart,” and “Checkout” with JavaScript tags (UK tags). Customer behavior is recorded on Universal Knowledge’s log server through these tags, and the search engine is then fine-tuned repeatedly through analysis of this big data.
Universal Knowledge’s unique algorithm, UK Rank, utilizes customers’ purchasing behaviors (logs of accumulated data) to the fullest extent. Through this, constant and real-time changes in search results can be achieved despite there being only changes in the purchasing behavior of customers without any changes to the product mix.
This has major implications for eCommerce sites. Search results for products whose sales normally change depending on the season will change automatically through the data analysis of seasonal purchasing behaviors. Furthermore, popular products such as those that have suddenly become popular or have just gone on sale are automatically displayed at the top of search results. Displaying the exact product that customers are looking for at the top of search results achieves both high relevance and increased sales.