Facebook announced small changes to its news-feed ranking algorithms today, promising more visibility into tweaks that could affect the exposure of organic posts going forward.
News feed has been a sore subject for Facebook over the last year. The event comes in the wake of an outcry last fall, when users and marketers alike reported that the organic reach of their posts had fallen off, in some cases dramatically. GroupM Next, for one, did a study into pages operated by 25 big brands that showed that the number of users seeing posts from a brand they "liked" was down 38% over a five-week period.)
For some marketers, it seemed like a bait and switch. Many had spent heavily to acquire their fan bases in the belief that a high share of those users would then see their content for free. The algorithm changes -- which Facebook characterized at the time as ever-evolving with the goal of showing users the most relevant content and reducing spam -- presaged the necessity of boosting Facebook ad spend to maintain reach.
"You'll still need to supplement with some kind of paid strategy," said Brandon Fischer, GroupM Next's director of predictive insights, last November.
Perhaps buoyed by good news -- its stock price reaching its IPO level after far exceeding Wall Street's expectations in the second quarter -- Facebook is again addressing a mysterious elephant in the room that shows no signs of disappearing. According to one agency executive, the opacity of Facebook news-feed algorithms continues to be frustrating. The executive cited internal research from the first quarter showing that brands' posts with texts and links had begun to get more news-feed exposure than posts with photos, which get much better engagement and thus drive more earned media impressions that marketers don't pay for.
"There's been a lot of speculation, and we haven't done a good job of communicating," said Lars Backstrom, a Facebook engineering manager whose team works on news feed. He announced two changes to the ranking algorithms behind it and another in the works.
'Bumping' the News Feed
After first testing the feature with 80% of its employees, Facebook has rolled out story "bumping," which means that an organic post that users didn't scroll far enough down to see the first time they logged into Facebook on a given day will be eligible to reappear during subsequent visits. The expectation is that the change will boost engagement -- likes, comments and shares -- by ensuring that users don't miss out on content they'll like.
There's something apparently in it for brands too. Based off its test with employees, Facebook claims that engagement for posts from "pages" -- which could be from a brand, an organization, or a public figure -- were up 8%. Among posts by regular users, the engagement lift was 5%.
Story bumping could potentially be aimed at users who are on Facebook infrequently, so they see the most captivating content that could be culled from a period of several days when they do log in, observed Robin Grant, global managing director of the social agency We Are Social.
"Is this a sign that Facebook is so worried about decreasing user engagement that they're changing how the news feed works to try and stem the tide?" he said in an email.
Facebook also announced the rollout of a new algorithm signal called "last actor," which will take into account the last 50 interactions users have made on the site when gauging what posts to present to them. It's significant because the ranking algorithms have historically relied on long-term usage. By way of a simple example, Mr. Backstrom said that "liking" the post of a friend who the given user isn't typically that interested in could have the result of surfacing more of her posts over a short-term period.
"Before we didn't have any of the real-time-ness that this change allows," he said.
Lastly, Facebook announced that it's working on chronological ranking of posts by actor. The notion is to have posts by a given friend that mention a TV show or a sporting event appear sequentially in news feed to provide context and relevance. But the problem is that grouping posts on unrelated subjects -- say a picture of a latte at 8 a.m. and a check-in at a bar at night, posted by the same friend and appearing one after the other in news feed -- could be disruptive to users. It had the effect of reducing engagement in early tests, Mr. Backstrom said, noting that the signal is still in development.