in ,

Decoding the social media algorithms. The ultimate guide

Unwrapping the riddles in a mystery.

social media algorithms


We now live in a world where bite-sized chunks of mathematical data have become central to how we live our daily lives.

When you search online, scroll through your social feeds or receive song recommendations from Spotify, you are being guided by an algorithm that understands your consumption habits perhaps more than you do.

Each day you are influenced by an algorithm in guiding your decisions and choices.

Mathematical calculations are influencing your purchasing decisions on Amazon, your flight in an airport and even whether your supermarket has your favorite cereal in stock.

As if by magic, you may be completely unaware of when an algorithm is at work. They act seamlessly between yourself and the task at hand. And while this might sound sinister, algorithms are, by and large, there to help you.

Social media algorithms are becoming central to everything you do. Often misunderstood and occasionally inaccurate but always on and always learning.

How you and most of society consume news and information can be attributed directly to algorithms.

Whether searching on Google or scrolling through Facebook, the news and information presented come from a mathematical equation based on two factors:

  • Algorithmic quality – the quality standard of the content available
  • Your previous history – the actions and reactions you’ve taken to specific chunks of content in the past

This means whether by design or by personal choice, social media algorithms have allowed us to create filters to see content we want and remove everything else we don’t.

While this guide does not claim to lay out the exact inner-workings of the social media algorithms, it tries to include everything we know about them that is publicly available.

What’s presented here is based on a combination of publicly disclosed information by the social networks, third-party research, some basic assumptions and a little common sense.

Secondly, the diagrams presented are not visual representations of the algorithms. That would be impossible. They are more of a process and checklist you can follow to ensure your content and messaging has the best opportunity to receive maximum impact.

The diagrams below are more Decision Problems rather than algorithmic equations.

Consider this a starter for ten and a guide that can be expanded upon and added to.

Social media algorithms for information and news flow

The most influential kind of algorithm is one that controls the flow of information people receive. It helps shape their thinking and understanding of news and events. Perhaps there is no greater power than being able to control the kind of information a group of people consume.

Over the years, social media platforms have become central to how people both communicate and receive information.

Whether that’s keeping in touch with long-distance family members, organizing events, networking, running a business or keeping up with the news.

Social networks have evolved from being a place where people connect to an information distribution platform.  

Many people now depend on social networks to be informed and up-to-date on news events. In fact, research by the Pew Research Center found that in 2017 two thirds of American adults reported getting at least ‘some of their news’ from social media.

As more people have started consuming information on social platforms, many publishers have taken advantage of the viral effect they can receive by having their content spread across them.

Indeed many publishers built their entire business models based on the viral effect platforms like Facebook, Twitter and LinkedIn create. The content is optimized for shareability using persuasive techniques to create an emotional response in readers.

Understanding how social media algorithms work empowers us to know when we are perhaps being manipulated by these publishers.

Content creators – brands and individuals alike – need to understand how best to work with the social media algorithms to generate the best return for their content creating efforts.

Time and again we see content go down a black hole because no consideration is given to the algorithm. What works on one social network may not work for another which is why it is important to understand how each algorithm works to create the content to suit it.

Not that the social networks give us the ‘secret sauce’ to their algorithms. Most are notoriously hush about the inner workings of these often-complex mathematical equations.

They do, however, provide tidbits of advice and information on occasion either through their official channels or by their employees on social media.

The changing tide of the social media algorithms

In the early days of social networking, the intention was to connect people online who may not have been able to do so in real life.

Family and friends on Facebook; colleagues and work acquaintances on LinkedIn and anybody and anyone on Twitter.

As they grew and as each platform explored ways to monetize they began to diversify their offerings. News distribution was a key component of this.

The social networks, and especially Facebook, began to see themselves as a conduit for news distribution. They changed their algorithms to suit this shift especially as more people began to receive their news via them.

News publications worked on growing their audiences on these platforms and increasingly made content that was both snackable and shareable.

In 2015, Facebook surpassed Google as the number one traffic source for news sites solidifying its place as a social networking behemoth.

Then something happened.

Clickbait articles, fake news, bots, trolls and political bickering began to dominate the platform. The Facebook algorithm, perhaps the most famous algorithm in the world, was at the center of it.

People were reporting higher levels of anxiety and unhappiness after visiting Facebook and Twitter. They had turned into purveyors of outrage news and instead of connecting people they had created divisive groups.

In 2018, the tide is beginning to change once again for the social networks and their algorithms.

Facebook announced that it intends to start prioritizing “meaningful conversations’ instead of news articles.

Twitter is killing automation on the platform meaning any bots that create fake engagement via likes, follows or retweets will be quashed.

LinkedIn is prioritizing status updates by its professional users who dare to be more personal and open about their lives.

This is the beginning of a new era in social networking. One that we’ve seen before and one that is more about human connection as opposed to information consumption.

It’s one reason why the influencer marketing industry is on the rise. Brands are looking to work with social media influencers who have developed deep connections with their audience.

The algorithmic levers have been pulled once again. For many publishers, it spells an end to their business model. For brands looking to engage in social networks, it requires a mindset shift.

It’s about building a community, not an audience. It’s about making content conversational instead of attention-grabbing. It’s ultimately about taking a more honest approach when engaging with people on social networks. 

Why we need to understand social media algorithms

A lack of understanding of how social media algorithms work is like driving in the dark with no lights on.

It’s possible and you may get to your destination but it’s an unnecessary risk.

In a nutshell, you need to understand them for the following reasons:

  • Impact: To ensure that your content creates the greatest impact possible
  • ROI: To ensure the time and effort spent publishing to these platforms has the greatest return
  • Reputation: To become a long-term trusted source of information for the algorithms
  • Wider societal impact: Communicators need to be able to ‘fight fire with fire’ to prevent false information spreading online

The social networks are constantly changing, refining and testing their algorithms. It’s important we keep up with them.

Facebook Algorithm

Perhaps the most famous (or infamous) algorithm in the world, the Facebook algorithm is constantly evolving in the attempt to bring more value to its users. The more value it brings, the more eyeballs Facebook has for advertising dollars.

Formerly known as EdgeRank, it has received its fair share of negative press in the last few years.

The endless decline of organic reach

In 2014, Facebook page owners started noticing a decline in organic reach. While numbers vary, the average organic reach dropped from around 16% to 6.5%. A page with 10,000 fans would only reach just 650 of them.

Page owners with more than 500,000 fans saw an event steeper drop to around 2%.

The reason for this, Facebook said, was because there was simply more content being created on the platform and people should see only the most relevant to them.

Over the years, Facebook has continued to tweak the algorithm further decreasing the organic reach. Around the same time, its advertising revenues have grown substantially.

The Facebook facilitation of fake news and clickbait headlines

The ability for information to be shared among Facebook’s billions of users is what is appealing to news publishers and brands alike.

It’s this viral effect of social media and Facebook, in particular, that has caused the spreading of false information and lies.

There are many reasons for this including political groups with an agenda, trolls looking to spread fear and hate for fun and hackers looking to exploit naive users into giving up their login details.

The algorithm has played a big part in this which Facebook is fully aware of. Both Brexit and the 2016 US presidential elections were largely played out on Facebook and other social media. Both were susceptible to fake news from fake websites.

As well as this, the use of clickbait headlines by well-known publications to entice users into clicking through to an article diminished the credibility of those involved.

The psychological manipulation of people using this method is straightforward. Develop an article headline that pulls people’s emotional triggers like fear and outrage which makes them click through and share it with their friends.

The new Facebook algorithm

The accumulation of negative media coverage along with research finding that people feel less happy after visiting Facebook has prompted the company to make swift moves to eradicate the sharing of content that make people either mad, bad or sad.

facebook algorithm
Meaningful interactions will be prioritized. Source: Facebook. Via: Matt Navarra

Facebook’s new approach is to drive, what Mark Zuckerberg called in a recent announcement, “meaningful social interactions.”

In Zuckerberg’s post, he outlines a new direction for Facebook with the algorithm at the centre of it.

“We’re making a major change to how we build Facebook. I’m changing the goal I give our product teams from focusing on helping you find relevant content to helping you have more meaningful social interactions.”

This means the algorithm will now prioritize content that drives discussions among friends and family as opposed to clickbait and links to third-party sites.

An example of meaningful interactions given by Facebook include:

  • A person commenting or liking another person’s photo or status update
  • A person reacting to a post from a publisher (or brand) that a friend has shared
  • Multiple people replying to each other’s comments on a video they watched or an article they read in the news feed
  • A person sharing a link over Messenger to start a conversation with a group of friends

The move has prompted concern by businesses and organizations that depend on Facebook as a communications platform.

With organic reach at an all-time low and the prioritization of communication between friends and family, many organizations question if Facebook is worth the effort at all.

Given its sheer size and dominance online, Facebook is worth the effort but this algorithm change requires a new approach.

What do we know about the Facebook algorithm?

  • A post is served to a small percentage of users to measure initial engagement
  • The Facebook algorithm will prioritize content that stimulates a conversation between friends and family
  • It will prioritize links shared over Messenger
  • The credibility of a user (completeness of their page, history of sharing etc) is a ranking factor
  • Brand or publisher content that is shared by a user and generates further discussion will be prioritized
  • It will prioritize live video because it receives more interactions
  • Engagement is based on a points system
  • Posts with long-form comments will receive a higher weighting
  • Native content takes precedence over links to other sites
  • Clickbait and asking people to ‘like, comment or share’ your content will receive a markdown

The Facebook algorithm ranking signals

facebook algorithm
Click for higher resolution

Working with the Facebook algorithm

  • Create content with the purpose of driving discussion among Facebook users

This is the new normal on Facebook and anyone looking to work with the algorithm will have to take this into consideration when developing content for the platform.

  • Take an 80/20 approach to native and owned content

Facebook wants you to create content for the platform, not just to post links to your own sites. But if you have a credible profile and do it sparingly the algorithm may help you

  • Use live video to drive engagement whenever possible

Facebook has specifically said they will prioritize live video because they know it generates six times more engagement than any other content. Anything that drives meaningful engagement is worth doing

  • Aim to be reputable in the eyes of the algorithm (be choosy with what you share)

Don’t use Facebook as a means to share every piece of content you have. If it doesn’t generate engagement then your credibility score will decrease. Only share content you believe is suitable for the platform and audience you’re trying to engage.

  • Absolutely no clickbait, engagement bait or overly promotional content

Any clickbait articles, statuses that ask users to like, comment and share, or promotional updates will be marked down by the Facebook algorithm. Don’t do it.

Twitter Algorithm

The introduction of the Algorithmic Timelines

The Twitter algorithm, or the “Algorithmic Timeline” as Twitter calls it was introduced in 2016. Prior to this, when you logged in to Twitter your feed was in reverse-chronicle order with the latest tweets from the people you follow at the top of the page.

When you log into Twitter now it’s a different story. It’s no longer in chronological order and your timeline may show a tweet from 30 minutes ago at the top of your page whereas there may be more recent tweets further down.

This is the algorithm at work and is designed to provide you with the most relevant content first based on your previous use of the platform. These are the tweets that the algorithm thinks you will find most important. When you’re developing a Twitter strategy you have to take this into account.

Trolls, bots and elections

Like Facebook, Twitter has been on the receiving end of controversy in recent years.

It has been accused of providing anonymous trolls and extremist groups a platform to spout vitriol and hate. And allowing Russian accounts and bots to meddle in the 2016 US presidential elections.

In January of this year, Twitter said it had removed 50,000 Russian-related accounts that used the platform to post automated material relating to 2016 elections.

This has prompted a new change to the Twitter algorithm announced in February this year.

From now on, automation (or bot usage) on the Twitter platform won’t be allowed. Twitter has sent a clear warning sign to services that facilitate the usage of these bots with the following statement:

“The use of any form of automation (including scheduling) to post identical or substantially similar content, or to perform actions such as Likes or Retweets across many accounts that have authorized your app (whether or not you created or directly control those accounts) is not permitted.

Working with the Algorithmic Timeline in 280 characters

The intention of the Twitter algorithm is to make the timeline more relevant so users can catch important tweets that they would otherwise miss from people they engage with the most.

At the same time, Twitter still considers itself as a live news site as well as a social platform. It reinforces this by asking “What’s happening?” when you log in to your account.

It won’t show you tweets that are days old because, by Twitter’s standards, they’re too old. The intention of Twitter is to help people find out ‘what’s happening right now’ as opposed to Facebook or LinkedIn which is more about ‘what’s going on this week.’

So while the algorithm will show you older tweets first, time is still an important factor. And if you want your tweets to get as much reach and engagement as possible, you have to work with the algorithm.

In September 2017, Twitter made one of its boldest changes yet by going from the 140 character limit to 280.

There was outrage from users who believed this move would spell the demise of the platform.

In hindsight, Twitter had a good plan based on analyzing engagement data on languages that could fit more context into a tweet. They found that tweets in Japanese – where the characters allow you to say more – received much more engagement than tweets in English.

What do we know about the Twitter algorithm?

  • Timing is heavily weighted in the Twitter algorithm
  • Profile credibility is favoured in the algorithm
  • Using the 280 character limit increase engagement rates
  • Native content is likely to take precedence over links to other sites
  • Tweets from people you engage with the most will show first
  • A tweet is served to a small percentage of users to measure initial engagement
  • Likes, replies and retweets are likely to have a weighting score
  • Time spent reading someone’s tweets or visiting their profile will impact the content you see even if you don’t engage with it
  • The more you engage with people and the more they engage back makes the algorithm show any further tweets in their timeline (if they are following you)

The Twitter algorithm ranking signals

twitter algorithm
Click for larger version

Working with the Twitter algorithm

  • Timing: Tweet when your followers are online

Use a third-party service like Tweroid or ManagerFlitter to tell you what time of the day your followers are usually online. This is the best time to post to ensure that your tweets get as much visibility as possible among your followers. The more visibility a tweet has, the greater chance of engagement it will receive.

  • Develop a follower-base around a specific topic or subject area to be more relevant and drive more engagement

Some people buy followers to look more influential than they are. Or they use a bot to follow and unfollow lots of accounts to inflate their follower numbers.

The problem here is when they tweet and receive little to no engagement. Why? Because the majority of their followers are either spam accounts or irrelevant.

Twitter has now thankfully put a stop to this manipulation of numbers but the lesson in it for us all:

To create engaging tweets you must have a relevant and responsive audience. Don’t add any kind of account in the hope they’ll add you back. Instead, be highly specific with the accounts you follow and over time those types of accounts will follow you back.

  • Create tweets based on what your followers like

Your Twitter Analytics provides you with a wealth of data not just about your tweets but about the demographics of your followers. It will tell you their subject interests and break them down by percentage.

Once you understand your follower interests you can create tweets for them. The more relevant they are the more likely they will engage.

  • Use the 280 character limit to provide further information and context

Twitter has stated that tweets that use more characters get more engagement. Why? Because you can add more information and more context. You can also be more creative. Do this and the Twitter algorithm will reward you for it.

  • Ensure your profile is credible

Ensure your profile is up-to-date and you have all the relevant information (name, bio, location, links, photo etc) filled out.

The algorithm favours accounts that it deems credible so do not be spammy, post broken links or use automation software.

  • Show up every day

If you’re tweeting every day and providing value to your followers, the algorithm will reward you for it. If your tweets from the past week received engagement the algorithm will prioritize you further. It’s about showing up every day and adding value to a relevant audience.

  • Engage with your followers

When you retweet and @reply other people and they respond, the algorithm will place any further tweets of yours in their timeline (provided they are following you).

Engaging with other Twitter accounts ensures that the algorithm sees you as a contributor to the community. Don’t just tweet out and expect people to engage with you – go out and engage with others.

LinkedIn Algorithm

While the LinkedIn algorithm hasn’t been on the end of as much controversy as Facebook or Twitter, it has certainly had a few hiccups along the way.

In September 2016, LinkedIn was accused of showing a preference to men over women when looking for candidates using its search feature.

A few months before that, LinkedIn users had fun and games with people who have names that are double entendres. Bart Simpson played no part in any of this.

LinkedIn has perhaps been one of the most open social networks about how its algorithm works.

In March last year, it’s data team published a blog post titled, Strategies for Keeping the LinkedIn Feed Relevant which included an algorithm diagram of how it combats spam.

What is telling from this post is that LinkedIn uses human intervention as well as its algorithm to determine content quality.

If a post begins to receive a lot of engagement “real people at LinkedIn” will analyse it and decide whether it’s good enough to be seen by a wider audience on the platform.

Insider tip: keep your LinkedIn network relevant to the types of content you share on the platform. It will have a greater chance of spreading if most of your contacts are in the same industry as you and value the same kind of content.

The viral effect is with your 2nd-degree connections

The LinkedIn algorithm gives your content virility by showing it to the connections of those who engage with it. If someone engages with your post either by commenting on it or liking it, the algorithm will show that comment or like to their 1st-degree connections (your 1st or 2nd).

So, let’s say someone comments on your post, the LinkedIn algorithm will show it to five of their connections. This is how your content is amplified outside of your 1st-degree network to your 2nd and 3rd-degree connections.

linkedin viral effect
Click for larger version

Use ‘Broetry’ to generate a viral effect with LinkedIn statuses

If you use LinkedIn regularly, you may have seen status updates what BuzzFeed calls ‘Broetry.’

These are “go-to post format for marketers, social media mavens, and growth evangelists.”

They are one sentence paragraphs designed with a narrative, emotional hooks and social triggers.

The best-known creator of ‘broems’ is growth hacker, Josh Fechter, who has created some serious viral posts on LinkedIn using this method wracking up 200 million views in the process.

While broems are not to everybody’s taste, there’s no denying this style of posts is LinkedIn algorithm friendly.

As we know, however, algorithms never stay the same and already users of the broem are acknowledging that they don’t receive the same traction as they once did.

Fechter has recently published a follow-up post outlining his new approach to working with the LinkedIn algorithm. In short, mix your content up. Don’t create posts that are all broems and instead use a variety of content styles.

What do we know about the LinkedIn algorithm?

  • Native content takes precedence over links to other sites
  • Likes, comments and shares are likely to have different weightings
  • The key to going viral is to get the algorithm to show your content to your 2nd and 3rd-degree network
  • A post is served to a small percentage of users to measure initial engagement
  • One sentence paragraphs of personal narratives (currently) do exceptionally well as a viral mechanism
  • Content with high engagement will be analysed by LinkedIn staff and potentially opened up to a wider audience (though human analysis can be subjective)

The LinkedIn algorithm ranking signals

linkedin algorithm
Click for larger version

Working with the LinkedIn algorithm

  • Create a relevant network of LinkedIn connections

If your LinkedIn network is full of people from a myriad of industries and sectors then your content won’t resonate with all of them. Instead, keep your connections relevant to the industry you’re in so your content has more chance of appealing and being shared by them to your 2nd and 3rd-degree networks.

  • Use the status box over blog post to stand a better chance of going viral

Users report that statuses receive more visibility than actual blog posts. This is surprising given LinkedIn’s focus on publishing content and the greater effort it takes to write a blog post. While this might change in time, this is the way the LinkedIn algorithm prioritizes written content for now.

  • Use native imagery and video

While LinkedIn hasn’t specifically said the algorithm will prioritize native imagery or video, given its focus on the latter and how other social networks have prioritized it, we can assume that if it currently doesn’t, it will.

  • Mix up your content style broetry with regular writing styles and paragraph structures 

While broetry isn’t dead yet, it’s not as viral as it once was. Take a leaf out of Josh Fechter’s book and mix up your writing styles. Use broetry alongside regular writing styles and paragraph structures.

  • Take an 80/20 approach to native and owned content

Most brands and people prefer to use the social networks to drive traffic to an owned channel like a website or a blog. These days it’s increasingly harder to do so as the algorithms prioritize native content over external links. The LinkedIn algorithm, however, will still drive a lot of traffic to your owned channel provided you don’t abuse it.

Instagram Algorithm

Instagram announced it algorithm in mid-2016 as a way to provide users with the type of content they will engage with the most.

Prior to the Instagram algorithm, a post had a half-life of 72 minutes. Now it can receive engagement days after it is been posted.

Regardless of whether the people you engage with the most post an image the day before, you’ll see it at the top of your timeline when you next log in.

Engagement is a key ranking factor for the Instagram algorithm. The more likes, comments, comment likes, post saves, DM replies and sent via DMs a post receives, the more weighting is placed on it by the algorithm.

What do we know about the Instagram algorithm?

  • When first published, a post is shown to a select group of followers to gauge engagement
  • Posting regularly will help you feature higher in users’ timelines
  • Posts shared via DM will be ranked by the algorithm
  • The genres of content you interact with most will be shown higher
  • Hashtags still work in the algorithm but mainly for the Explore page
  • An established and engaged community will improve each post’s credibility
  • Actively engaging with other people’s content (via likes and comments) helps drive people and further engagement to your profile
  • The longer users spend on the post the more algorithmic credibility it has
instagram algorithm
Click for larger version

Working with the Instagram algorithm

  • Post regularly

If you post irregularly the algorithm will treat you as a bit player and won’t prioritize your posts among your followers. And why should it? Those who post regularly and contribute the most will be rewarded for it.

  • Engage with the types of content you post on 

Engaging with content and users that are relevant to your own Instagram profile will help drive people to your own account. The more you engage with these accounts the more the algorithm will show your content their timelines.

  • Don’t worry about the timing of posts

Unlike Twitter, timing is not as important on Instagram as the algorithm can show content that is a few days old.

  • Use hashtags with the intention to be found via the Explore page 

If you can get to the top of the Explore page via hashtag usage it can generate hundreds, if not thousands, of likes and followers.

YouTube Algorithm

The YouTube algorithm is perhaps one of the hardest to crack these days given the popularity of the platform. With 1.5bn global users and statistically the second largest search engine in the world, YouTube is a platform to be reckoned with.

The YouTube algorithm has been developed to serve those that contribute the most to the site. This is reflected in some of its ranking factors which are based on posting consistency and the number of subscribers a user has.

Unless you’re famous, superbly talented in some way or have a completely different angle to creating videos than anyone else, building a substantial audience on YouTube from scratch will take work.

Why? Because the quality of content is exceptionally high and every topic imaginable has been covered already. On top of this, it requires posting of around 2/3 times a week to gain algorithmic traction.

This is why YouTube SEO is a growing industry and many ‘traditional’ SEO experts like Brian Dean are beginning to focus on and hone their skills on the YouTube platform.

What do we know about the YouTube algorithm?

  • Total watch time and audience retention are important ranking factors
  • Upload frequency is an important factor
  • A recently uploaded video is served to a small percentage of users to measure initial engagement
  • The more subscribers you have, the more priority the algorithm places on your videos
  • Videos that are 7-16 mins are the optimal length
  • The YouTube algorithm is AI. It learns, understands and expands
youtube algorithm
Click for larger version

Working with the YouTube algorithm

  • Consistency is key 

Channels that are sporadic in their video content uploads will be marked down by the algorithm. The best approach is to publish videos consistently and preferably at the same time and/or day each week.

Not only will this help you from an algorithmic perspective but your subscribers will be more receptive to new videos when they know they are due out.

  • Building a subscriber base

The more subscribers your channel has the more credible you are in the eyes of the algorithm. This is a catch 22 situation for anyone starting from scratch and looking to grow a YouTube channel making it all-the-more difficult for newbies.

  • Hook your viewers in on the first few minutes of your video

The longer people watch your videos, the more interesting the YouTube algorithm will assume it to be. The aim is to hook the viewer in as soon as possible and many YouTubers use specific tactics to do so.

  • Make your videos around 7-16 mins in length 

Matt Gielen found that videos which are 7 – 16 minutes in length receive the best retention, most engagement and best viewer-to-subscription rate.

Decoding the social media algorithms requires a collective effort

Social networks continually A/B test and change their algorithms to accommodate new features, revenue streams and provide more value to users.

While we’ll never completely understand the inner workings of each social media algorithm, we can take cues from company announcements and adjust our approach accordingly.

We can decode the social media algorithms piece-by-piece through our own experiences and by sharing learnings with one another.

Using crumbs of information, basic assumptions, some common sense, continuous testing and the sharing of data, these mystical bite-sized chunks of mathematical equations can be cracked.

Much like the SEO industry was born out of a collective desire to crack the search engine algorithms, we too can work together to make sense of their powerful social media counterparts that are shaping society and culture.

This means continually testing content to discover what works, what doesn’t and, in either case, why.

It requires speed and agility when it comes to social media communications. It requires a collective sharing of knowledge and experience.

And perhaps it requires a collective sharing of clout to pressure the social networking companies to provide more information on the inner workings of these mysterious algorithms that are having a greater influence on how we all live.

This is my starter for ten.

If you’ve anything extra to add, feel free to let me know.