Slingshot

Directing people to where they really want to go in the Priceline ecosystem.

Company

priceline.com

Function

UI Design, HTML, CSS, SCSS, Onboarding

Team Members

Liqun Chen (principal engineer)

View Project

Due to an issue with my hosting service (that I'm actively resolving), you can view assets from project here.

Overview

Small disclaimer: Slingshot was not originally a project I should've been involved in, but like many things I ended up getting involved with, it just happened to be related to some work my team was doing...so naturally, we got involved.

One of the more important focuses for my team was creating a seamless and pain-free experience on both the promotional and transactional side of our program. As the products became more robust, so did their features. Unfortunately, the tool we had created to help route our traffic was not being taken into consideration. Luckily, just as we decided that something had to change, another team was currently considering fixing the same issue. After deciding to pair-off and tackle the problem, Slingshot was born.

What is Slingshot?

Slingshot is a routing application that intercepts incoming URLs to its' application and then sends them where they need to go, as defined by the user. It also helps with collecting incoming data from those URLs, to ensure proper attribution (another issue we were having). The benefit to Slingshot over our old platform was the flexibility and also the URL sanitiation. Before, links would come in with maybe a dozen or more query strings of varying length, and all we would do is redirect that URL with a new header. Slingshot goes a step farther to convert query strings, data formats, construct query strings, append or remove data, etc, on the fly.

That had a few benefits:

  • URLs could be shorter, so lower file sizes could mean no inbox clipping
  • We could perform more lucrative testing
  • URLs can be configured to handle edge cases and potential mistakes on its' own
  • It removed a lot of redundancy

Creating the base

Luckily, to help with the process, I had been in the process of converting a couple of tools over to an internal application design system that I built for a couple of other projects. Since most of what we would need was already pre-defined, and because we wanted a sharper UI, my system was deployed. This meant we could focus on development rather than having to make decisions on the fly.

Quick results

Noticeably, there were positive results after the first attempt at testing the new platform. In just a small test with a ~1m customer list, we saw:

  • Fewer clipped messages, which raised CTR and activity with some seldom-used links
  • A slight life in conversion due to the lack of clipping
  • Another observed lift in conversion due to removing redundancy in inbound URLs
  • In a 10% test of that group, a very confident increase across the board in personalized deep-links

That's just a small subset of things we noticed, and as we expanded the platform, it made those data points much more significant.