Notes on Facebook Instant Articles
WTF is taking 8 seconds?
At the core of Facebook’s case is that the average article takes eight seconds to load. We’ve known forever that users don’t like slow loading pages. So why have publishers ignored this? Why did it take Facebook to take action? Is it all the Adware? Poorly optimized images? “Responsive” templates still loading full desktop code? “Personalization”?
Publishers will go where the audience is. But what is the audience?
Facebook drives a lot of audience to content. But who has a good sense for how this actually breaks down? On any given day there is a small number of super-viral articles, and presumably a (very) long tail without so much traffic. It likely varies very differently by demographics too (Newsfeed is the primary destination for news in some countries). Also, how many of the people included in these stats gave up because the content never loaded?
Don’t worry about the data.
One of the oft-given arguments for not putting content on someone else’s platform is that it’s hard to get the user data.
- I don’t think publishers are doing anything very useful with user data right now (except selling it to Outbrain for lots of $$$). The ad networks need it, of course.
- It’s a reality of the multi-platform, multi-device world that tracking (and aliasing) users across them is going to be very hard. We may just need to recognize that and move on.
Newsfeed wasn’t built for media consumption.
Facebook reminded us of this a couple of weeks ago when it announced that it would be prioritizing social updates and conversation over articles. The reality is that the user context of Newsfeed (“what are my friends doing and saying?”) may be unsuited to consuming thoughtful content.
What does success look like?
Good experiments need a testable hypothesis. I’m not exactly sure what the hypothesis is for publishers using Instant Articles. More revenue? More audience? (If the latter, it’s not at all clear how that’s determined.) Mark Thompson is quoted as saying it’s all good “as long as he believes he can bring some of [the audience] to his own site and apps.”
Using the platform for distribution, or building content for the platform?
Which reminds me of an important point. Buzzfeed has figured out how to build and publish content on different platforms, not in spite of itself, but because that’s its core strategy. If you’re blindly hoping that Instant Articles will deliver more audience to your own platform, I think you’re going to fail. Just as Instagram and Snapchat Stories are different platforms, so it’s likely Instant Articles will be too: the publishers that do best will develop and publish their content specifically for the platform.
It’s still an algorithm
For many media commentators, this is the beginning of the end, the straw that breaks the camel’s back, etc. This seems hyperbolic: FB has always had an algorithm and the rules have never been completely clear (with good reason, so they don’t get gamed). In that sense, nothing has changed here. It’s the algorithm (and 1+ billion users) that holds the power.
What is Twitter doing?
Personally, I find Twitter way more helpful for discovery than Newsfeed. What’s going on there? Do articles take eight seconds to load from Twitter? Do they care? In many ways it could be a much much better distribution engine. Perhaps it’s too late.
What could publishers have done differently?
If the real point here was that the content took too long to load, I think there were very many other ways to fix that. Beginning with making the content much faster. Facebook is doing nothing more than hosting the content in a template on some fast servers (likely cached and loaded along with the newsfeed item). Fast servers are a commodity, and templates and caching is pretty straightforward. Publishers could probably have worked together to figure this out without Facebook. And mobile content everywhere would be better for it.
But the ads
Digital publications rely on ad revenue. Ad technology relies on the tracking and profiling of users. That requires scripts and tracking to be installed on the site serving the content. The ad tech slowed sites down (and made them horrible to look at), while creating a dependency that prevented publishers from giving users the best experience. Nothing new there.