What type of stories
go viral on LinkedIn, and how does the algorithm impact the visibility of
certain posts? We take a closer look at the content process on the business
professionals’ network. For some business-focused sites such as Forbes and
Inc.com, LinkedIn engagement is beginning to rival, or even surpass, their
shares on Facebook.
Here’s how engagement,
measured as shares of links on LinkedIn, looked for English language publishers
on the platform from January to September:
According to
Executive Editor Dan Roth,
the platform had three million writers and around 160,000 posts per week as of
the end of 2016. Those articles either get distributed by LinkedIn’s in-house
editorial team, made up of around 25 editors based around the world, or
algorithms. LinkedIn claims that 87 percent of users trust the platform as a
source of information, making it an attractive location for gaining people’s
attention.
But what sort of
messaging works on LinkedIn, and how does it get distributed? Unlike Facebook,
there isn’t a whole lot of discussion about the influence of LinkedIn’s
algorithm on what their users see when they log on.
As with most
algorithm-based news feeds, we can divide the question of why certain stories
go viral into two sections. First, we need to analyze the actual substance,
tone and presentation of the stories themselves. Second, we need to consider
the distribution particulars of LinkedIn, the role of its algorithm and the
influence that a writer or publisher can have on that process.
The content: aim for
high quality
Let’s consider the
types of stories that are seeing high engagement on LinkedIn. LinkedIn is
actually quite explicit about the types of stories that it sees as being likely
to go viral on the platform. In a guide, they note that articles should “share
professional expertise,” and suggest titles such as “What will (or should) your
industry look like in 5, 10, or 15 years and how will it get there?” and “What
advice do you have for career advancement?”
Looking at the most
popular stories of the last few weeks on LinkedIn in NewsWhip’s Spike tool, we
can see that these type of stories also resonate when they come from
publishers. Career advice and professional development insights are extremely
popular.
In presentation
however, LinkedIn makes an effort to distinguish its content from more
mass-appeal platforms. The platform discourages the use of listicles and
obvious clickbait and recommends that writers “keep articles appropriate for
the LinkedIn audience. Don’t post anything obscene, shocking, hateful,
intimidating or otherwise unprofessional.”
Being able to maintain
these editorial standards is something that LinkedIn takes very
seriously and to reasonably good effect. You won’t have noticed LinkedIn
mentioned very often in ongoing discussion about the spread of “fake news,” and
the platform is not known as a place where viral publishers go to thrive.
It’s also important
that articles aren’t seen as overly promotional; it’s fine to mention where you
work, or the product you’re building, but going overboard results in risking
spam status and a visibility downgrade. LinkedIn isn’t trying to compete with
Twitter for the breaking news audience, or Facebook for mass appeal. Its strengths
lie in allowing users to develop thought leadership and sharing content
relevant to their careers. Developing a genuine persona on LinkedIn with
expertise around a specific topic is a surefire way to build an audience base
on the platform. LinkedIn also recommends that articles are at least three
paragraphs long.
Distribution: The
algorithm at work
Distribution of
content on LinkedIn is an algorithmic process, and that algorithm is
theoretically designed for engaging, interesting stories to go viral. In this
sense, the algorithm isn’t all that different from the type that bigger
platforms employ, but it’s aimed at a more niche user base. LinkedIn is open
about the effect its algorithms have on content visibility in the news feed,
using a “man+machine” approach to classifying content in real time based on
signifiers such as early engagement, previous reaction to content from the
page and more.
LinkedIn uses a
feature called “FollowFeed” to help determine what gets prominence in users’
feeds. FollowFeed aims to provide high precision and recall, or relevance. For
a technical explanation of how FollowFeed works, see this great detailed explanation from
LinkedIn engineer Ankit Gupta.
LinkedIn has a
three-stage process for identifying and dealing with low-quality content. As
the post is being created, a classifier buckets posts as “spam,” “low-quality”
or “clear” in near real time. Next, the system looks at statistical models
based on how fast the post is spreading and the networks engaging with the
post, in order to spot low-quality posts. Finally, human evaluators review
posts flagged by users as being “suspicious.”
There are some factors
that help determine how much preference articles secure for algorithmic
distribution based on factors related to the personal details attached to the
author’s LinkedIn profile.
Here’s what LinkedIn
recommends regarding article distribution factors from individual writers:
(Stories are) shared
with a subset of your connections and followers. This is determined by
connection strength, your connection’s notification settings, and notification
state (i.e. number of unread notifications). Members who aren’t in your network
can choose to follow you and by
doing so they will receive your articles and posts in their feed.
1. Followers may receive notifications when you
publish an article. Your articles may be available in their LinkedIn homepage
feeds and can be included in news digest emails.
2. In an effort to simplify the notifications
experience, we often aggregate notifications to your connections.
So as with any news
feed, there’s quite a bit at play behind the scenes in determining how many
people will see and share your posts. Analyzing other success stories and
changing techniques learned on other platforms can help in boosting your own
signal.
Ultimately, LinkedIn’s
editorial mission statement is to provide timely and professional content to
users. Those users can be divided into different cohorts – engineers,
salespeople, executives and countless other – but timely relevance remains the
key consideration.