I gotta say, I’m surprising myself with each new post I publish on this issue, as I keep thinking I’ve exhausted all facts and angles. Apparently I now believe I have still got something new to write on this.
Let’s get to it.
“...Unique visiting members grew 11% to an average of 100 million per month, and member page views grew 33%. This has yielded 20% year-over-year growth in page views per unique visiting member, continuing a pattern of accelerated growth throughout 2015”
Mark Mahaney, an analyst and the managing director of RBC Capital Markets, had the following question concerning these figures:
“... Jeff, the detail you pointed out earlier on the rising engagement, the 20% growth in page views per user. Just a little color on what areas are really driving that increase that almost accelerating growth and engagement? And, when do we start seeing that really show up in Marketing Solutions? It seems like an obvious monetization opportunity, yet, there's this disconnect between its acceleration and engagement and a slowdown into Marketing Solutions. Is that just a lag that will close? Thanks a lot.”
Jeff responded:
“With regard to the 20% year-over-year increase in page views per unique,part of that is our greater focus on the quality of the visit and the quality of the unique. So, we're de-emphasizing more transactional flows… in favor of really honing in our value propositions and ensuring that people are getting value from the site. And that's driving much healthier organic engagement on a per unique basis for those folks that are visiting LinkedIn that way.
In terms of the specific products that are helping to drive the dynamic, certainly the publishing platform continues to generate traction... In terms of when you would see that really manifested in our Marketing Solutions growth, I think… you've got these secular headwinds, with regard to display."
To sum up the above:
Jeff said (reading the figures disclosed to the shareholders on Q3 2015 results) that while unique visiting members grew 11%, the total member page views grew 33% and the page views per unique visiting member grew by 20% an obvious pattern of accelerated growth throughout 2015. Jeff mainly attributed this to the traction generated by the publishing platform and the greater focus of LinkedIn on the quality of visits.
When asked to explain the disconnect between the pattern of accelerated growth of engagement and the slowdown in marketing solutions (AKA $$’s), Jeff attributed this disconnect to
secular headwinds with regard to display (I admit, I have no idea what he meant).
I’d like to offer another explanation for this disconnect
First we need to understand that the page views metrics LinkedIn discloses to its shareholders on NASDAQ are not verified by 3rd parties. We know this from the following statement in LinkedIn’s annual reports:
“We track certain performance metrics, including…member page views, with internal tools, which are not independently verified by any third party.”
If you’ve read
my first post on this issue you’d know I was puzzled that, as if governed by a rule of nature,
featured posts on LinkedIn picked by editors get the majority of views here, but have extremely low engagement metrics (likes/comments out of their number of views) compared to non featured posts.
I believe the theory I’m about to propose, explains both this peculiarly low engagement phenomenon of featured posts and the disconnect Mark Mahaney from RBC Capital Markets inquired about on that Q3 2015 earnings call.
If my theory is correct then this would be a very serious issue to LinkedIn as a publicly traded company, so I’ll emphasize the obvious - I call it as I see it and there are possibly facts or explanations I’ve missed. I would be happy to be proven wrong.
A substantial amount of the page views LinkedIn counts are not actually pages members viewed
LinkedIn’s traffic on 2015 was divided fairly evenly between desktop computers and mobile devices, but as unique mobile users spend only
one minute a day on LinkedIn this means that a larger share of the long-form posts read and engaged by members on LinkedIn’s publishing platform, was executed through desktop computers.
One major path in which LinkedIn distributes featured posts on desktop computers is by placing the chosen featured posts on the same page just below any other post a member happens to be reading.
That is why on any given day, no matter how many different posts you read, you would always see the same one or two featured posts each time you scroll to the end of any post you actually read.
I’m suggesting that in most cases, whenever we actually read one post here, LinkedIn counts this for its metrics as 2 viewed posts, one view counted for the post we actually read and another view counted for a featured post most of us don’t read, but were forced to scroll to when we wanted to reach the end of the post we did read, or wanted to read the collapsed comments on the post we read.
You can see supporting evidence for this outrageous theory of mine if you closely watch the URL address bar on your browser while scrolling to the end of any post. When you reach the end the URL automatically changes to the URL of the featured post below, without us clicking on anything.
This false views metric also explains the "low engagement" of featured posts
In 99% of the featured posts I reviewed (any post with more than 10,000 views was most likely featured) the engagement numbers were embarrassingly low (I consider less than 3% likes of counted views and less than 0.3% comments as a low engagement result).
I don’t think our conclusion should be that LinkedIn’s editors are bent on picking to feature only uninteresting and non engaging posts, so the only logical conclusion I have is that the presented views counter of all featured posts does not represent the actual number of members who viewed/read these featured posts, and the faulty UX feature I recorded above explains how that is possible.
Take for instance
this post by
John Battelle featured yesterday and today (also featuring in the screen recording above).
It has now 330,000 views (gained about 100,000 views during the time it took me to write this post)
and only 481 likes and 163 comments.
These engagement numbers are ridiculously low and you can find non featured posts here with "only" 4,000 views that managed the same absolute numbers of likes and comments, hence they “seem” to be 8,000% more engaging than John’s post (and most other featured posts) - of course this is not the case.
If we calculate the average engagement of non featured posts on LinkedIn we would see that the number of “likes” are around 4% to 10% of views and the number of comments are around 0.4 to 1% of views.
Based on these statistics, by inducing from the number of comments John Battelle’s post got, I would roughly estimate that his post was actually read/viewed no more than 30,000 times, making LinkedIn’s presented views counter for that post off by around 300,000 views.
I believe these 300,000 false views were generated by members in the last 48 hours who read other posts, and while scrolling to the end of the posts they read they unknowingly activated the views counter for this one featured post as I’ve illustrated above.
Bearing in mind we are talking about possible false views in 2015 amounting to 70-90% of the presented number of views for all featured posts on LinkedIn, I really hope LinkedIn didn’t consider as true these numbers in its independently unaudited metrics presented to its shareholders.
If it did, as I have to logically assume based on the facts I presented here, then the above quoted statements and figures from disclosures LinkedIn gave to its shareholders, are as radioactive as a North Korean nuke.
About me
I’m the co-founder and CEO of
www.trailvest.com a startup still developing its MVP. We believe we can change the world, and no, we are not developing a rival professional social network :-).
