Analyzing Return User Behaviour helps you optimize your site for better engagement and more conversions.
Data Snacks | Jul 23
John Bonini on February 8, 2018 • 3 minute read
Data Snacks is a new video series from Databox that shares bite-sized tips to help you be more data-driven.
We didn’t spend money on advertising.
We only posted eight times on our company page.
We posted far less than we did on Facebook or Twitter, and yet traffic from LinkedIn grew at more than double the rate of Facebook, while also outpacing Twitter’s traffic growth by 35%.
We’ve been trying a lot of different things on LinkedIn over the last couple of months. It’s become a more interesting place to engage our customers. It’s not as transient as Twitter, it’s not pay-to-play like Facebook, and the addition of video sharing (and other sweeping UI changes) has quietly turned it into a formidable alternative to other channels.
In the latest episode of Data Snacks, I’m sharing what we’ve learned after experimenting with several different tactics over the last few months, what’s worked, and how we’re tracking it.
Want the free template featured in this episode? Get it here.
In 2017, LinkedIn made significant changes to its algorithm wherein it now relies on machine learning to determine the quality of your posts before it even shows them to your network.
And while the specifics around the criteria of LinkedIn’s algorithm remain unknown, here’s how it generally works.
Through LinkedIn’s automation system, anything you post will be instantly classified under one of the following three buckets:
You obviously want to fall into that third bucket.
As mentioned earlier, while the specifics around how LinkedIn categorizes “spam” or “low-quality” is unknown, a general rule I follow is to avoid overly-promotional language like “free”, “discount”, “offer”, etc. This is typically good practice and similar to the way ESPs flag certain words in your subject line which increases your spam score.
Next, your post is then shown to a small segment of your audience in order to gauge engagement quality before determining whether or not to show your post to a larger subset of your network and beyond.
(This is why you might notice it takes a bit for your posts to pick up steam.)
During the audience testing phase, specific engagements (likes, comments, shares) are scored in order to determine whether or not to:
At this stage, automation yields to human intervention.
When your post starts picking up traction through natural engagement, it’s passed along to actual humans who work at LinkedIn in order to further assess your post and whether or not it should be shown to more people.
Depending on the level of engagement your post is seeing, this assessment could even result in your post being shared with users that are outside your network, either through “trending content” or through the natural flow of the newsfeed.
It sounds trite, but natural conversation wins the day.
While shares are great, LinkedIn’s algorithm values a long thread of engagement on your posts. It’s not enough to seed a bunch of likes from your coworkers anymore. If you want your content to be exposed to more people, you need to inspire conversation on your posts.
Treat every post as a conversation-starter. Ask questions. Solicit feedback on your content. Tag people that had a hand in the success of the post, like contributors, customers, vendors, etc. Respond to everyone that leaves a comment.
If anything, this should inspire your content strategy to include more collaboration with your customers, vendors, industry friends, coworkers, or whoever else could play a hand in improving the scope and balance of your content.
Data Snacks | Jul 23
Data Snacks | Jul 16
Data Snacks | Jul 9