From a Hackathon Prototype to Launching with Facebook at F8
Most people don’t know what Instant Articles is or why they should care. Instant Articles eliminates load time of mobile articles and keeps you inside the Facebook app. While it’s arguably a better user experience it does potentially mark the end of the open mobile web. Beyond conversations of Internet morality, this poses a problem to publishers—how can they reach their readers inside the Facebook walled garden?
While Instant Articles will fundamentally change the way the Internet consumes mobile content, as the first app approved for Facebook instant articles, (announced during its annual developer conference, F8), Social A/B by ShareThis will enable publishers to not only capitalize on this fundamental shift but also capture revenue.
What is Social A/B?
Social A/B allows publishers to automatically test different headlines to increase traffic to their articles from Facebook. Our product is seamlessly integrated directly into the publishers’ existing workflow—our algorithms determine the winning headline, updates the publishers’ articles with the winning headline (directly on their page), and provide reporting to the publisher. Matt Harvey, Senior Product Director of USA Today, has said that “[Social A/B] drives up to 150% more engagement on popular content, directly impacting our audience and revenue goals.”
He also added that, “Before using Social A/B, our editors would spend too much time trying to pick the best headlines for their stories. This was inefficient and not always effective. Now, they can run multiple headlines on every story and allow Social A/B to pick the best one based on data, not just intuition.”
From hack-a-thon to the big stage: F8
Building Social A/B has been a journey. Traditionally, engineers work closely with product managers, who work closely with clients and stakeholders, who then communicate the requirements back to the engineering team. We take a different approach at ShareThis – we believe that engineers and product managers are equal, integral members at all parts of the production. Social A/B was born from a prototype built for one our publishers, USA Today, at a joint, on-site hackathon down in their offices in Los Angeles, CA, with engineers and product managers from both companies.
We work on what matters by focusing on client/end user value driven development. From inception, engineers at ShareThis worked side-by-side with USA Today to build and iterate quickly through prototypes. We flew across the country to sit next to an editor to better facilitate communication and understand product fit. We vetted every feature by quickly testing and validating the feature prior to any technical development. If the data suggested big improvements in engagement, we built out the whole feature. We used data and A/B testing to quantify the success of each feature.
Though it seems arduous, this process actually creates greater efficiency in the long run by allowing us to quickly identify which features to prioritize and build, and which features to drop. During the development of our Social A/B product, it guaranteed that our engineers were only spending time on features with the greatest potential to drive engagement. The graph below shows the number of Social A/B experiments our publishers created over the last quarter in our alpha release.
The Technology That Got Us Here
This is a tough engineering challenge for ShareThis, as we see billions of events daily with a reach close to 80% of the US population. Today, we’re focusing on getting Meteor to a point where it can handle our large publishers as well as leveraging our existing data pipeline. Once our entire network is scaled out using Meteor, we could be the largest production Meteor deployment in the world.
We built a deployment pipeline for continuous delivery, delivering features as soon as they’re merged to the master branch. We rely heavily on code reviews, pair programming, unit and integration testing. This gives us confidence that we deployed code without need for regression. That’s not to say that regression doesn’t happen – when it does, we just roll back, address the issue, and then create the right test to prevent it from happening again. At ShareThis, we believe that failure doesn’t exist – it’s just a learning point to draw from to ensure future success.
We’re looking forward to tackling the hard technical challenges ahead. We’ll be scaling out the system to handle on boarding our massive network of publishers; help the publishers convert the success of Social A/B campaigns into increased revenue; and identify business value for ShareThis and revenue opportunities.
We have an amazing collaborative process and deep relationship with our publishers to take Social A/B beyond our F8 launch. We have an opportunity to build out a service that will prove to be a strategic asset to our publishers driving increased readership and revenue.
Interested in joining the team?
Send an email directly to Isaac Mosquera, SVP of Engineering, at firstname.lastname@example.org or check out any of our open positions here at: http://www.sharethis.com/hiring