Enhancing Retail Sales Through Applied Data Science
An e-commerce startup that offers members free two-day shipping at online fashion retailers engaged SPR to help them experiment with new technology, flesh out service offerings and take advantage of the tracking data that their data science team was collecting. The lure for online retailers to partner with the startup was increased customer traffic and tracking data, which they could use to customize their offerings and drive greater sales conversion.
SPR’s first job was to use a cross domain toolkit, Zoid, to insert shared micro-apps on partner websites. The app collects a list of products the data science team tagged as “trending” based on page views; the client then presents it to users with some customization requested by the partner, such as font, scroll arrow design, etc.
If you’ve ever been shopping online and seen a “Low Stock!” or “8 people have item in cart” message, you’re seeing the magic of data tracking. SPR’s second app was an inventory counter that collected stock numbers from partners’ data feeds; if a user clicked a size that was under a stock threshold, the cross-site app would inject “Low Stock!” in the UI. All interactions and events were tracked and sent to data science. By hooking up a quantity-counting data stream, SPR was able to develop the Low Stock app to display things like “Low Stock, only 3 left!” It was shown that this “Low Stock!” app provided about $3M/year in incremental revenue for partners.
The next projects increased member traffic back to the client’s partner sites. First, SPR helped modernize a Chrome extension that injected banners or buttons on site navigation. When it detected the user was viewing a product that was also available at a partner’s site, it provided partner images and links to the customer.
Next, SPR assisted in modernizing their member website and signup funnels. The team that SPR worked with used Kotlin as a backend serving up React sites compiled and minified with Webpack. The experience was customizable based on user activity data collected through Snowplow, as well as metrics and alerting driven by Datadog. These sites were able to interact with cookies and implemented authorization exchanges using Auth0 and Spring Security to enable passive logins between sites.
Ultimately, SPR was able to help the client take advantage of their extensive data science collection, driving more user engagement and more purchase conversion. Tracking data was used to understand user flow and drive metrics/alerts when the data unexpectedly changed, and provide evidence to retailer partners that the client pushed a sale to success (and should get paid as a result).
- React TypeScript