Facebook’s core business model is familiar: the bulk of its revenue comes from advertising. Its most important metric is time on site.
Nearly all digital media companies share this business model. Still, none of them consistently hold people’s attention the way Facebook does: as of March 2016, Facebook had 1.09 billion daily active users. As of July 2014, Americans spent more than 40 minutes on Facebook every single day. As of September 2014, 30% of Americans got their news from Facebook.
Research has shown that getting positive feedback on Facebook taps into our brains’ reward centers. Beside the fact that humans value social approval, though, Facebook’s addictiveness comes largely from a single feature: News Feed. It’s the first thing you see when you sign in, and it’s the first thing people block when they try to curb their Facebook usage.
News Feed also elicits more questions than any other feature. For example: how does Facebook figure out what goes in? Why did I see that some guy from my high school just had a baby, but I didn’t see my friend’s photos from Germany? Why didn’t I see the hilarious video of a panda sneezing three hours ago?
The short answer: numbers. Lots of them.
How Does Facebook Know Who I Care About?
Every one of your Facebook friends has a coefficient, i.e. a number between zero and one that determines how important they are to you. Pages have coefficients as well. Every interaction you have with one of these entities — a like, a comment, a sticker on Messenger — gets added up and averaged out to create this coefficient. Different interactions add different weights.
To see an example of coefficients in action, go to your profile and click on “Friends.” You’ll notice that your friends aren’t listed by any readily apparent criteria, like last name; instead, they’re sorted by coefficients. (The first person is probably whoever you’re dating. If not… I won’t judge.)
Why won’t Facebook let people decide on their own, you might ask? Facebook actually does: users can curate lists of friends, and the defaults are close friends, acquaintances, and restricted (that last category is for the nosy relative you might’ve added out of obligation). The thing is, this feature has a low adoption rate; Facebook’s algorithms work well enough that for most people, creating these lists simply isn’t worthwhile.
How Does Facebook Know What I Care About?
Let’s say you post a picture of Emmy, your beloved golden retriever. First, it’ll get shopped around to your closest friends, the ones with high coefficients. Next, their reactions will determine whether it’ll appear on more people’s News Feeds. The more people interact with that post, the higher its rating gets. (Not all engagement is equal: for example, a share gives the post more of a boost than a like, and a like counts for more than an impression.) Unlike coefficients, rating are computed in real time.
You’ll see nearly everything your closest friends post, but if a lot of people engage with an acquaintance’s post, its quality rating will go up, making it more likely to appear in your NewsFeed. For example, the flame war raging beneath your cousin’s post might not be quality content in the traditional sense, but it’ll definitely cause people to pay attention. The rating of a post decays over time, so if you want yours to last, you can try to game the system by getting all your close friends to interact with it right off the bat. (It won’t work if everyone else ignores it, though.)
In conclusion, to put things as simply as possible:
Rating x Coefficient = Chance of Seeing Content
That, in a nutshell, is what makes Facebook so addictive.
The takeaway for companies running Facebook pages is that quality trumps quantity. If you put out lots of posts but they’re rarely compelling, they’ll get lower ratings, which will lower your page’s coefficient and ultimately reduce your other posts’ organic reach.
The larger lesson for companies relying on ad revenue is that content distribution works best when it’s driven by hard numbers. The most successful digital media companies have taken that idea and run with it: “BuzzFeed is known as a viral content company, and one of basic statistics people look at with viruses is reproduction rate,” BuzzFeed’s head of data science told Contently. “So for every one person who has some disease, how many more get it directly as a result? And there are obviously correlations for sharing. For content, we can tell within an hour or so of publishing what type of stuff we should put prominently on the homepage, promote on Twitter, things like that.”
Until the internet came about, media consumption was almost entirely a one-way affair: companies put out content, and people bought or watched it. Now, media companies get veritable mountains of feedback: they can tell where readers come from, what they engage with, what they share, where they share it, and even which device they’re using. In raw form, that information is messy; making sense of it is difficult, often unsexy work.
But it’s worth it. Just ask Mark Zuckerberg.