Twitter’s Getting Smarter – Here’s Why That’s Important
Twitter’s algorithm – their back-end code which defines which tweets you see every time you log in – is getting better. You might not have noticed it, and if you have, you might not agree that it’s improving. But it is, and that bodes well for the future of the embattled platform.
Twitter’s switch to a more algorithm-defined feed last March was initially met with strong opposition from the platform’s user base. The core functionality of Twitter has long been based in real-time, in what’s happening right now – an algorithm which surfaces tweets behind time is totally against what we’ve come to know the platform for, right? This was the key argument against the change, that it fundamentally and irretrievably changes what Twitter is. We have algorithms on Facebook, we want unfiltered, live updates in tweets.
But the switch to an algorithm has actually been mostly positive – you can still view your Twitter feed in real-time, you still get to see all the updates from all the people you follow. It’s just that now, you also get notifications about tweets you missed that may be of interest, highlights tweeted by connections while you were away.
You might not, personally, like these additions, but the data shows that many users do – Twitter noted in their last performance report that they’re seeing consistent growth in daily active usage, tweet impressions and time spent, with their improved data matching playing a part.
And really, that makes sense – Twitter has access to a heap of user data, the ability to monitor engagement trends and user behaviors, it’s logical that they would follow the lead of other platforms and look to use such insights to improve the user experience.
You can say what you like about Facebook’s algorithm, but the fact is it generates results – The Social Network has seen significant increases in both total and daily active users and time spent on platform since the algorithm’s introduction. Instagram too has grown its audience by the fastest rate in its history since adding a feed algorithm (though in fairness, that time period also incorporates the introduction of Stories).
There are downsides to algorithm-defined feeds, it’s not a perfect system. but it works for the metric that Twitter so desperately needs to lift – it helps boost engagement and interaction on the platform. Which is why we’re going to see more of it.
Twitter’s already testing out additional use cases for their algorithm data – last week, reports highlighted how Twitter was testing out a new system which looks to showcase tweets from the specific people you engage with most to ensure you don’t miss them.
That seems like a pretty basic use of their data – you’d assume Twitter’s engineers must have found that people engage with the majority of these friends’ tweets, yet they only see, say, 70% of them. Make sure they see them all and you boost engagement. Simple.
In another example, Twitter has been sending some users notifications of upcoming events, prompting them to follow along.
This is a particularly clever use of their data, highlighting a relevant issue to certain users – defined by region or the topics they regularly engage with – to keep them coming back. It also works in line with Twitter’s new ‘What’s Happening Now’ approach.
As we noted recently, they implement these things not on a whim, but because the engagement boost is likely worth it. Even if only a tiny percentage of people use these new prompts, there are more than 500 million tweets sent every day. 0.1% of that number would still equate to 500,000 extra tweets.
What’s more, because of engrained user behaviors on Twitter, there’s no way they’re able to maximize tweet engagement without an algorithm. I’m talking here about the “follow back” process – because Twitter places a premium on numbers, people are driven to increase their follower count. The easiest way to do this is to follow as many people as you can, unfollow those that don’t follow back, then repeat. It’s easy, it works – but it’s not conducive to actual engagement. Because of this process, many (many) Twitter users are following thousands of people, and no one can possibly keep up with that many accounts. This means there’s no way for you to see all the tweets that might well be relevant and of interest to you, unless Twitter highlights them for you. The other option is Lists, which is a lesser discussed option – but still, Twitter can’t rely on who you’re following alone as the key determinate of your potential interest.
But where Twitter could really boost engagement is if they can utilize their data to more effectively identify and highlight key trends and happenings, narrowed down to a more specific focus. This was noted as part of CEO Jack Dorsey’s recent call for improvement ideas – one of the most mentioned possibilities was being able to follow specific topics and more easily switch between interests.
The capacity to follow topics would be a start, for sure, but what if Twitter could use their data points to get a more specific view of what you were likely to be interested in? What if Twitter could look at the tweets you engage with, the topics you like to discuss, then cross-reference that against all the engagements on their network in order show you listings of ‘Who to follow’ recommendations or conversations with a significantly higher degree of relevance? People might complain – that’s kind of what Twitter’s for – but improved data matching would enable you to stay up to date with more relevant information. Which is what the platform is all about.
Relevance is at the core of social media, and social media marketing specifically. You come to the platforms that have the most personally relevant information, the content you want or need to see. You might want to see every tweet from every person you follow, but if that can be complimented by smarter recommendations on what you’ve missed, and improved data matching tools highlighting personally relevant topics, Twitter could become a much more powerful platform, and a more important part of your every day process.