Etsy is well known and a major player in the handmade and vintage arena with a marketplace that offers curated collections and suggestions.
Etsy search evolution: Etsy uses machine learning to boost search-based purchases
Etsy search is using Google Cloud machine learning technology to return the most relevant products to customers, reports The Wall Steet Journal.
According to Etsy’s chief technology officer, Mike Fisher “getting the first page right” is vital to the marketplace. Etsy’s large inventory of 60 million items from 2.2 million merchants requires the website to accurately match shoppers’ search intent. Etsy say that the majority (80%) of search-based purchases come from the first page of Etsy results. This shows how important optimised listings are on Etsy, found out more here.
‘Search improvements added $260 million to gross merchandise sales’
The idea of ‘the first come, first served’- relevant to search-based purchases – encouraged Etsy to start using Google Cloud in 2017 because of its machine learning capabilities. Etsy are almost three-quarters of the way through the plan which will complete in 2020.
Etsy Q1 earnings conference call saw Etsy chief executive officer Josh Silverman highlighting search and discovery as the key focus for the marketplace. He said that Etsy’s goal for 2019 is to take the vast sea of listings and make it feel more manageable and curated for buyers.
Etsy’s adaptation of the Cloud is already bearing the fruits. Mike says that search and discovery have contributed $260 million in incremental gross merchandise sales over the past two years.
Merchants trading on Etsy specialise in designing handmade artisan products which aren’t stored in catalogues. The discovery of their products relies on machine learning algorithms to display relevant items to shoppers. This move marked a change from Etsy’s reliance on index-based searching which aimed to match the customers’ search enquiries with sellers’ products keywords.
The new process, which Etsy adapted to hone search and discovery, is called ‘context-specific ranking’ which ranks search results in real time to make Etsy’s first-page results display the most relevant products. It uses machine learning to analyse the results from all product searches for the past serval weeks in order to learn shoppers’ buying patterns.
Etsy takes advantage of different search experiments in the Cloud. These trials often include A/B tests which compare different algorithms to see which ones produce the most relevant products in a test.