Last week, I wrote to you about the death of SFC. SFC was an URL shortening service I unceremoniously shut down after technical debt and lack of unique features made maintaining it more painful than it was worth. Except, that wasn’t entirely true.
In actuality, as I was writing SFC’s eulogy, I had already finished a total rewrite of the service and was setting up a new hosting environment. See, it’s hard for me to let go of projects. Especially one of the first web services I finished. And I’ve had an idea that SFC would be perfect for.
What if on a blog, you could set up dynamic Open Graph tags, titles, and other meta data and get meaningful insights on how those variations impacted the conversion rate of your articles on third-party platforms like Twitter and Facebook? Open Graph is the social protocol that determines what information gets loaded when you paste a link on a social network and a dynamic card gets populated.
If you click to view the below tweet on Twitter’s website, the card that generates below the caption is an Open Graph card.
These changes wouldn’t be hard to measure. With access to how many sessions your article gets, using UTM tags accurately, and building the right management system into your backend to fairly distribute your variants per session it would be possible to implement an underutilized statistical exercise that could supercharge your viewership: split tests.
A split test, otherwise known as an A/B test, is a statistical test that compares multiple changes to an element of a website to determine what changes, if any, have improved a site metric (like conversion rate). Building a suite of tools to split test Open Graph data, or even just to change an article’s deck and header image dynamically, like how changing an existing YouTube video’s thumbnail can increase the total video’s views, is a great way of relieving a typical stressor from writers. By allowing more flexibility in how we distribute writing, we can optimize any given pieces performance, or produce an evergreen title card that dynamically updates given certain criteria.
This is all an idea I’ve been mulling over today, but in the context of SFC, I think adding the ability to combine the data of multiple links and run split tests on what pages a SFC link points to is an easy way to bring split testing to more people and provide a service worth paying for.
What do you think? If you’re a writer, would being able to test what decks and images increase your articles performance be something you or your organization would pay for? Leave a comment or if you’re reading this in an email reply.
If you know a writer, why not share Ham Typed with him or her?