Select Page

Screen Shot 2014-11-19 at 11.23.47 AM

This is the first time in this blog’s short history that I’ve had a guest post and it gives me great pleasure to introduce Gericke Potgieter and his firm Artifex Knowledge Engineering to my readers. I learned of Gericke’s work via Quora when I got a notification last week that someone had added an answer on the LinkedIn Publisher topic. I was thrilled to learn that he had succeeded in finding out more about how the LinkedIn algorithm functioned. Though it is very unlikely that we’ll ever know exactly how the LinkedIn algorithm works, I believe Gericke has done the LinkedIn publishing community a big favor by shedding light on what triggers the algorithm. His book “How to Feature on LinkedIn Pulse: The Definitive Guide to Publishing Successful Long-form Posts on LinkedIn” is the best book yet on the topic. Thanks to him, we can now apply insights about hidden metrics like “Velocity” and “Fame” to our publishing, marketing and engagement strategy. Anyway, enough from me, here’s Gericke in his own words:


15 million members now have access to Publisher, but LinkedIn has over 330 million members and climbing, the reality is that we are only in the early phase of what will soon become a flood of LinkedIn member posts. From the first day that LinkedIn decided to roll out the publishing platform to all members, one of the most frequently asked questions we see on the web is: “How do I feature on LinkedIn Pulse?


LinkedIn shares very little information about how they select posts to feature on Pulse – they tell us that you can be featured as an Influencer, as an Editor’s Pick or as a result of being chosen by an impenetrable and mysterious algorithm that auto selects member posts for featuring in a Pulse Channel.


When I first started publishing posts my results were disappointing – maximum views typically hovered around 100 with maybe one like and if I was lucky, a comment.  And then it happened – I wrote an article and without warning, the views started flooding in, along with likes and comments. I also gained hundreds of followers!  In fact, after this one article I grew my follower base by almost 50%. Buoyed by this success I wrote another article – and bingo! it received over 4500 views with a large number of likes, comments and shares.  Followers again grew by more than 50%.  This time however we were prepared – Artifex gave our audience a free gift on our company web site that converted very well.  Of the roughly 4500 views we got 89 downloads – close to a 2% conversion rate which these days is unheard of in online marketing.


As a data scientist I was naturally curious about the LinkedIn Publisher system. I decided to try to figure out what had just happened. What could I find out about the algorithm? How could I improve my chances of being ‘featured’ on a Pulse Channel? The thing about a computer algorithm is that you can always get some insight into its workings if you look at the results of the algorithm through the right lens.  I conducted a study which analyzed 561 Top Posts featured in 48 Pulse channels with a single goal in mind: I wanted to know why the algorithm chose certain posts and ignored others.


8 key areas:
1. Titles:
I measured titles in terms of keywords, length and whether it contains numbers or not.
2. Base metrics:
I looked for connections between post length, days active, views, likes, shares and comments.
3. Subscriber activity:
I measured how active subscribers were in a given channel in terms of the base metrics.
4. Audience activity:
I measured how engaged audiences are in relation to base metrics on specific posts.
5. Word correlations:
I wanted to see if there are any links between word count and the base metrics.
6. Daily performance:
I measured base metrics daily to understand any patterns related to activity over time.
7. Weekday analysis:
I wanted to find links between weekday publishing and their relative success.
8. Keyword analysis:
I wanted to figure out how the Pulse algorithm uses keywords to classify posts.



What I found was surprising – the Pulse algorithm seems to be less sophisticated than I had anticipated. It essentially relies on two key things namely: a small, but evolving tag cloud and audience interaction.  I’ll focus on audience interaction here since my research indicates that the degree of audience interaction contributes significantly to the likelihood that a post will be picked up by Pulse. I identified two “hidden” metrics – every author can see the base metrics of views, likes and comments, but the algorithm does some magic with these metrics to determine the potential popularity of a post.  I christened these two key metrics “velocity” and “fame.”



Velocity seems to be a trigger for Pulse to recognize posts with potential.  Some LinkedIn bloggers have previously stated that x number of views will get your post into Pulse, but that’s not true – if 10,000 people view a post we don’t know whether they read it in depth, skimmed it or left after a quick glance.  We call this a high bounce rate in SEO terms because people literally bounce off the page before engaging and actually reading the content. There are currently only two ways an author can see the number of shares their post received. If they log out of LinkedIn and google and find their post or once it is featured in a Pulse channel. I believe that sharing plays a very important role in calculating velocity. This makes complete sense because a post that is shared more often will move faster through the LinkedIn network as it gains popularity.


So how do we know when a post is popular?  A post gains velocity if there is a healthy ratio between the views, likes and shares.  I haven’t been able to identify precisely what this ratio is (yet), and I know that other factors may play a role (like the bounce rate). However, I do know that the algorithm measures in terms of ratio rather than actual quantity.  This means that an article with only 30 views stands as good a chance of being featured in Pulse as one with 300 views, as long as the audience interacted with it. So having an engaged audience, regardless of size, matters a great deal if you want to build that audience.



According to LinkedIn engineers the Publisher platform selects posts based on relevance, freshness, and diversity. The research showed that relevance is related to how a post aligns with a channel, freshness aligns with the length of time it has been published and diversity seems to be a natural result of selecting posts from members as a result of velocity.  When I looked at why a post becomes a Top Post I realized that there is something else going on – Top Posts were clearly not selected on the basis of a single metric such as views or likes.  I realized two things: 1) Top Posts are selected relative to other posts in the same channel and, 2) since this is seemingly the case, there must be something against which these posts are compared.  It seems as if another hidden metric was at play, one I called “fame.”


Fame is a metric that seems to play a role in differentiating Top Posts from Latest Posts.  Top Posts are featured on the Top Posts page of Pulse which means they automatically get more exposure.  My research shows that the Pulse algorithm seems to compare the ratio of the base metrics of posts in a given channel to qualify a post as a Top Post.  This means that an article with relatively few views can in fact be a Top Post if its ratio between views, likes, shares and perhaps even comments are higher than those of other featured posts. In other words, once a post has been selected for a channel, if it performs well on that channel, it can get boosted by having prime positioning on LinkedIn’s top shelf.


Some tips to get you started

So now that you know about Velocity and Fame, how will you use this knowledge? Understanding some of the science behind LinkedIn publishing is useful but there are obviously many other factors to consider when trying to get your post featured in a Pulse channel.  To get you started here are some of my best tips:


  1. Write for a specific channel – if you want to be featured, then your content must match one or more specific channels.  When you plan your post make sure that you know for which channel you are aiming.
  2. Write for your audience – the Pulse algorithm isn’t very sophisticated because it relies to a certain extent on how your audience interacts with your post.  If your post has lots of keywords, but no real value then it won’t be shared.  If it isn’t shared your post won’t succeed.
  3. Interact with your audience – many authors forget that a post is simply the start of a conversation.  When you respond to comments, or invite readers to go beyond the content of your post (to your web site for example), then you will gain more loyal followers.  This is very important because with enough loyal followers you won’t need to feature in Pulse to have success on LinkedIn.


About the Book

My research was so successful that I decided to publish a Kindle book on the subject called “How to Feature on LinkedIn Pulse: The Definitive Guide to Publishing Successful Long-form Posts on LinkedIn”.


This book answers questions like:


  1. Why are you failing on LinkedIn Pulse?
  2. How do you get more views, likes and comments?
  3. What is the best day to post your article?


This is what you will learn:


  1. How to prepare your Profile for success.
  2. How to build a loyal audience.
  3. How to write a great post and optimize it to feature in a Pulse Channel.
  4. How to successfully publish and distribute your post.
  5. How to capture and retain your audience outside of LinkedIn.


The book also has the following bonus features:


    1. A quick reference guide for the anatomy of a great LinkedIn post.
    2. A full research report explaining how we performed our study and what our findings were.
    3. A free keyword analysis tool that will help you write better long-form posts.


About Artifex Knowledge Engineering (Gericke Potgieter & Amore Potgieter)

ARTIFEX 0Our company specializes in making complicated knowledge practical.  We do this by creating automated expert systems, making sense of Big Data (and Small Data) results, and by conducting break-through research.  You can learn more about us on our website.