The Algorithm Revisited
This week I’m going to deep dive into the excellent piece of research recently published by Richard Van Der Blom
Richard has put a lot of time and effort into analysing data from 4500 posts and as he freely admits these results can be interpreted in different ways so I thought I would give you my interpretation on what is going on here.
Before we get to that though, a few things I have observed this week…
Are lots of people getting banned because of LinkedIn’s new community policies?
“As far as I can tell, the answer in no”
The only person I’m aware of that has been banned (twice) since the introduction of these new policies is Chris Williams (I’ve added a screenshot of his profile pic in case he gets banned again!)
I like Chris and find much of his content amusing but there is no doubt that he has been walking a fine line for some time. His posts are often in questionable taste and it’s highly likely that many people would be offended by some of his content.
In addition, he has openly criticised some pods (rightly in my opinion). Pods however will group together (their main MO!) and attack someone they take a dislike to. We know from my previous conversation with an ex LinkedIn employee that a trust & safety team investigation will be triggered when 5 members report a person or post – this can easily happen when;
- A pod decides to gang up and report you
- You have published content that could be considered offensive or have broken LinkedIn community policies
Chris has a loyal and very vocal following and this seems to be the only source of the claim that people are widely getting banned.
As far as I can see, there is nothing to worry about.
Potential New features
Jane Wong disappeared from Twitter for a while this week but John Espirian helped out by pointing me in the direction of another Twitter account who does the same thing by spotting things that LinkedIn are testing on their mobile app.
Alessandro Paluzzi has recently tweeted about features allowing more control of company pages via the mobile app as well as the following;
I’m loving the new online meeting feature in Messages using either Zoom, Teams, or Blue Jeans (the only company I know who use that is LinkedIn!!). This doesn’t appear to be a feature integrated in InMail yet – so users who pay to use LinkedIn get less features in their messaging facility – go figure!!
Seeing this reminded me of a feature I was calling for back in my wish list for 2015 when I wanted Skype integration into messages, it’s only taken them 5 ½ years to do it!
In the UK we still don’t have this feature but my gut instinct is that it could be very powerful. From what I can gather, the majority of early adopters are posting complete crap but surely someone must be finding interesting and creative ways to use stories for fantastic content!
If you are in a country that has Stories, please send me any screenshots or better still videos of great use of stories.
New Algorithm Research
This is fantastic work from Richard but as he admits himself, many of the results can be down to interpretation. You can read his article which summarises the conclusions here
The conclusions in this report are not absolute and are therefore freely interpretable. We admit that apart from the elements examined, several other factors play a role in the success of a post.
So I go into some detail in the podcast about my interpretation of these results. In summary:
- SSI – I doubt the algorithm checks our SSI, it’s more likely that some of the components of the SSI closely match those considered by the algorithm.
- All Star profiles: Nothing new in this but it’s not especially significant as it only impacts the initial distribution.
- Hashtags: This is new and very interesting. My advice would be to use 10 popular hashtags from now on as this can only increase views and engagement with no downside.
- Dwell time: New from the last research a year ago but has been with us for quite a while now. Document posts benefit, image posts lose out in my experience. The ‘See more’ element of this has been important for years.
- Posting time: Interesting that Richards research suggests that the ‘golden hour’ is actually two hours!
- Tagging: Nothing new in this but good to see it confirmed in the research. The question I always have is “How long does the tagged person have to respond before the post is penalised”
- External links: I make a very different conclusion here. The algorithm no longer suppresses link posts which explains the increase but they still perform badly because they fail to attract engagement.
- Comments are 50% more powerful than likes: this seems high to me but I don’t doubt the data. All reactions count pretty much the same but I predict this will change – surely a simple Like has to have less value than other reactions.
- Shares: They’re pants…nothing new in that but I’m curious about Richard Bliss’ hack. I have seen some evidence in his feed that it can work (but not always). I suspect both the author and sharer have to have highly engaged followers for it to work.
- Track the performance: This is brilliant! If a post performs well, wait until engagement starts to ease off before posting again. I suspect this happens naturally if you are busy replying to comments anyway but it’s great advice all the same.
- More to come: This is the most interesting slide of all! There does appear to be some contradiction between points 2 and 4 but to be fair Richard is only ‘teasing’ it here.
- LinkedIn Live engagement is lower than native video. I agree it’s much less valuable but the number of comments (which are less substantial than normal comments) are much higher especially as a ratio to the number of viewers!
- Lack of engagement on other posts doesn’t impact your own posts: I seriously can’t buy this one. Virtually every client I have ever worked with has seen better engagement by commenting on others posts. More detail required.
- Followers will see less content than connections because on the whole, they will have a lower relevancy score against those connections you DM the flip side is that by over connecting you are making it harder to engage with great content yourself.
Conclusion: What a fantastic piece of work this is by Richard and I want to congratulate him for putting in the many many hours of work it must have taken.
I may not concur on the interpretation of all of the results but I massively respect the work he has done here.
Post of the week
Thanks to Mark Lee for the nomination.
This post has great structure but it mainly works because it is highly relatable, especially to working mums. Well done to Hayley for a great post…could we be witnessing the emergence of a new Jeri Williams?! 😉
That’s all for this week. I’m on holiday next week so there won’t be an episode.
Catch up again soon.