Reversing the Traffic Slide
Northeastern Global News (NGN) is Northeastern University's award-winning daily news publication, reaching millions of readers a year. When article readership started declining, the team needed a data-driven way to win readers back, grounded in real evidence from real readers.
I designed and built a live, scalable A/B testing system comparing illustrated article headers against photographic ones to find out which format keeps readers engaged longer. From writing the WordPress plugin in PHP to deploying it to production and launching the live experiment in Adobe Target, I owned the project end to end.
Client
Northeastern Global News
Type
Experimentation & Data
Year
2026
Role
Product Designer
Process
Designing the experiment
I started by defining a clear, testable question: does the header format (illustration vs. photo) meaningfully change how long readers stay on an article? I designed a clean 50/50 split test and locked in the success metrics upfront: time on site, measured in Parse.ly, GA4, and Adobe Target. Defining what winning looks like before the data arrives is what separates a real design decision from a gut feeling.
Building and shipping the system
I built a custom WordPress plugin in PHP from scratch, registering two fields directly in the CMS: a photo header upload and a one-click A/B test enrollment toggle, so editors can tag articles for testing themselves, keeping the workflow fast and fully self-serve. The plugin emits a meta tag on eligible articles that Adobe Target reads to serve the correct variant automatically, with the photo version injected accessibly into the DOM using aria-hidden. The architecture went through a full PHPUnit test suite, multiple pull requests, and three merged PRs before going live in production on June 8, 2026. Now in Phase 3, the experiment runs automatically across all tagged NGN articles with zero manual intervention.
Scaling the result
Once the test runs across 20+ articles over two to six months, I analyze the aggregated results in Parse.ly, GA4, and Adobe Target reporting and ship the winning header format as the permanent standard across all NGN articles. One infrastructure investment, one data-driven decision, one site-wide improvement that reaches millions of readers a year.
Key design decisions
The most important decision was to build a reusable system rather than a one-off test. Instead of manually swapping headers on a handful of articles, I designed and built a full experimentation pipeline inside WordPress and Adobe Target that runs automatically, scales across the entire publication, and produces results meaningful enough to justify a site-wide design change. That choice transformed a single header question into a permanent, reusable capability for NGN to make any future design decision with real reader data at its foundation.
Outcome
This project gave NGN a scalable, repeatable experimentation capability: the ability to make any future design decision based on real reader behavior rather than internal debate. By building the full pipeline into the CMS infrastructure, the system runs continuously, grows with the publication, and costs nothing extra to reuse for future experiments.
Once the test concludes, the winning header format rolls out permanently across all NGN articles, directly improving engagement for a publication reaching millions of readers a year. The infrastructure built for this single question now powers every design experiment NGN runs going forward.
