The need for personalized service implementation increases year by year, so we believe personalization should satisfy both customers and marketing specialists. Our team is constantly working on platform functionality improvement to meet their needs.
In June Retail Rocket experts focused on platform stability improvement and minor bugs fixing. There were a few points to brush up, so we go into them in a number of thesises below:
- The method of analyzing user interests to related products has been finalized and improved. The fact that algorithms have learned to distinguish differences between related subcategories including highly specialized and complex ones (for example, adult goods) made possible to read user interest in more detail and provide more accurate recommendations. The mechanics is based on analyzing positions customers view and buy together, as well as whole sets of products generally purchased by users.
- Sales report has been improved. Minor bugs on numbers rounding in partner account statistics have been fixed.
- The search-based personalized recommendations algorithm has been tuned. Now it corrects typos in the query as well as identifies cases when there is no need for correction. Sometimes users saw products with similar IDs in recommendations entering an exact SKU. We revised it and now search algorithm offers products that exactly match the request of each customer.
- The stability of loading images has been increased. We brought to minimum errors in mass images transferring from the client’s server. Minor issues causing rare fails of images download for markups rendering were also fixed.
- Segmentation module has been refined. Now we form a separate segment for categories with the same names if it’s the case for a particular store:
- The segmentation speed, efficiency and fault-tolerance have been increased. Now, regardless of calculated segments size, the page loads instantly, even if there are more than 150 thousand categories in a segment. Before the update the speed of page load could take about a couple of minutes in such cases, now it takes 6 seconds and less.
- Unsubscribing alerts module has been adjusted. Its compatibility with other protocols, such as SOAP (Simple Object Access Protocol), has been also improved. Now the store immediately receives a notification if a customer unsubscribes from Retail Rocket emails. Thanks to this, you can now easily set up a joint user unsubscribing.
- Product details update module has been improved. Now it’s real-time product specifications and other additional parameters tracking, as a number of goods in stock and others.