Shop Centricity – A Case Study
Here at Etsy, we’re constantly asking questions about our data: questions about our customers, questions about the items in our marketplace, questions about what works and what doesn’t. As data analysts, we seek to answer these questions that help guide intelligent and informed decisions to make our marketplace better. Today’s marketplace at Etsy contains over 9 million items represented by around 800 thousand sellers. Twenty five million visitors from over 150 countries come to Etsy each month, totaling multi-terabytes of data per month, and we use Hadoop and cascading.jruby to help us analyze this data at scale.
In this post, we’ll be investigating the role of shops on Etsy. Like other online marketplaces such as eBay or Amazon’s marketplace, items are sold by individual shops. If you’ve ever bought from these marketplaces, you should ask yourself if you remember the merchant you purchased from. Compare your experience with that of buying this altered vintage plate from the shop BeatUpCreations who crafts modern portraits onto antique plates. Etsy’s marketplace sells unique handmade goods crafted by unique individuals: here we explore the hypothesis that shops play a central role in the buying experience on Etsy.
Let’s start with analyzing purchase behavior on Etsy. Etsy’s checkout funnel should look very familiar: view an item, add it to your cart, buy it. To learn how visitors discover the items they ultimately purchase, let’s look at referrals to the listing page of the purchased item.
Unsurprisingly, we see that some people find these items via browsing and bookmarking tools like search and favorites. Some shoppers engage in conversations with a seller before finally buying an item: buyers can easily send sellers a private message from their shop page (something we call a “convo” on Etsy). At 47.6%, the biggest faction of purchasers funnel into an item directly from a shop page. On Etsy, every seller has his or her own shop page (like BeatUpCreations’ shop page).
Within shop pages, merchandise is organized into different sections and visitors can browse available and sold items. Buyers who visit a shop see an average of almost three different items from that shop before going on to purchase an item. Just as every listing in Etsy’s marketplace is unique, every seller has a personal style. When buyers see an item they like, they click through to the item’s shop page where they can explore variations of the item (different sizes, etc.), or other items with similar styles, themes, or taste.
Interestingly, we can see from the graph below that the leading referer to shop pages are in fact listing pages. Listings and shop interact in a mutual and symmetric manner: items are a discovery tool for shops, and shops are a discovery tool for items.
By examining the flow of traffic that leads to a purchase on the site, we have learned how shops do in fact play a major role in item discovery at Etsy. In addition to seeking out that perfect item, visitors also tend to focus on finding people and shops which speak to their tastes. We believe that this is due to several factors — the unique nature of all of the products on our site, the emphasis on people and individuals, and a distinctness of style within shops.
Analytics is more than just building dashboards and graphs — it’s asking thoughtful, insightful questions and harnessing a wide-array of tools to answer them. Whether it’s large-scale distributed tools like Cascading and Hadoop or more traditional ones like R and Matlab, it doesn’t matter, as long the job gets done. At Etsy, we not only believe in keeping our data around, but we believe in harnessing it — to gather insight, to learn from our mistakes, and to grow.
Interested in learning more about analytics at Etsy — check out our jobs page as we’re looking for data engineering analysts and hadoop engineers. We’re also looking for a director of analytics to lead and build out the team!