4 Major Facebook Newsfeed Factors: Hypothesis
The Facebook newsfeed algorithm has over 100,000 variables which range from the very simple (proper grammar) to “wtf how do I measure this” (social relationship).
Here’s four new variables which I think encompass many under them and act as “master variables.” In math terms, if X = A + B + C + … + n; these four variables are the X and all the smaller ones we do know about fall under them. This isn’t like EdgeRank which may have been mutually exclusive single-use variables but rather four variables that double dip on variable use. In other words, your single reaction is counted more than once when figuring out what news to serve you next.
I’ll cover the master variables first, then a few smaller variable coefficients, and finally a big example at the end.
The less time it takes to read, the more likely it’ll show up in the feed. The exception is live video. Recorded video has a 90%+ view time on the first 15 seconds, 65%+ on the first 30 seconds, and less than 10% view time for anything over 2 minutes.
Positive coefficient: Infographics, video under 1 minute, text under 180 characters
Negative coefficient: Posts with “Read more…,” text only, external links
Reactions matter. Everyone has complaints and drama in their life and there’s been a drop off on activity because the last thing people want to see on social news is complaints about something they like and general drama.
Positive coefficient: Happy reaction, Wow reaction, Haha reaction, keywords to show affection, long time spent viewing a single post, life events
Negative coefficient: Angry reaction, no action on post viewed in feed (i.e. ignoring something is a sign you don’t like it)
Relevant to Goals
This was a tough one until I paid attention to recent news and noticed what affected the feed. If a bit of news isn’t relevant to your goals then you shouldn’t see it. For example, my friends that post about baseball I rarely see their activity; however, those that post about gaming I see quite often even when they aren’t talking about gaming.
Positive coefficient: keywords, events attended, geographic location (closer = better)
Negative coefficient: keywords, geographic location (longer = worse), opposing interests, affiliate links
Range of Influence
Everyone has a sphere of influence. If you like something and everyone knows about it, how many of those people will care and refer their friends? You’d like to hear about a friend’s friend and that friend when they went to a musical event assuming you’re a huge music fan. On the other hand, you could probably care less that a friend’s friend is into mapping behavior using linear regression when you hate maths.
Positive coefficient: friends of friends with similar interests, direct friends with similar interests, sharing news and posts from the same page, belonging to the same FB group as others
Negative coefficient: geographic location (longer = worse), few posts (how do I know what you like if you don’t talk about it), limited reactions to posts, link shorteners, overuse of hashtags, hiding FB group news with keywords.
Big Example (story in text, variables in parenthesis)
A few friends attended a political national convention this week (event, life event, travel, relevant to interests, relevant to goals, range of influence, positive feelings). Normally, I’d see a post a week from their feed where they make 2–7 posts a day. They’re pretty active this week (activity level increase).
However, this week during one visit to FB one friend showed up nine times! In fact, their updates were posted back to back which is unusual given that the feed separates friends into the most likely to engage at the top with one post and mixes with friends less likely to engage falling below.
They had updated their feed with photos (low consumption time), short video (low consumption time), and tagged their other friends (range of influence).
Taking a closer look, I realized they didn’t bother to name drop (keywords) any politicians but rather their body copy was more about excitement (keywords) and people’s reactions were happy/wow (reactions) and comments were encouraging and excited (keywords). Also, people tended to react and then post (high engagement).
I thought maybe this increase in visibility was due to sharing a trending topic (trends); however, my friends who are constantly talking about politics (keywords, relevant to interests) showed up in their normal 1 post per day in my feed despite them ranting 12 hours a day every day (activity level).
Perhaps this was a spike due to a life event (life event); however, my other friends that start new jobs, have kids, or move (life event, geographic location, travel) show up once and if I don’t engage with the post whatever that friend posts later gets hidden (negative reaction, engagement rate).
In any case, I did the usual and reacted and commented on one person’s feed and shared the other’s to my feed. Next day, they only showed up four times before the normal newsfeed sorting kicked in.
What Does This Mean for Business?
Keep it short, make it positive, add value to your client’s lives beyond whatever it is you’re selling, and create some community so if one fan sees your activity the others will as well. Even if you’re brick and mortar and can’t travel, there’s always talking about the local community or weather.
Anyway, I’m rambling and would love to hear your thoughts on what I think is the biggest behavioral model in existence.