What Twitter's Search Changes Mean for Marketers
Twitter has quietly announced a change to how their search results are displayed, moving from reverse chronological order to being listed by 'relevance'.
The change has reportedly been in place for the last few months, but Twitter has only made an official announcement this week. What it means, in basic terms, is that Twitter's now looking to show users the most engaging tweets, as opposed to the most recent.
Here's an example of the old (on the left) and new results side-by-side:
As you can see, the new results are not based on when they were tweeted, but how much engagement they've generated - you can see the retweet and Like counts are much higher on the right-hand side results.
But these aren't the only factors taken into consideration - Twitter software engineer Lisa Huang has explained their process of training the new model in the announcement post, detailing how they're using machine learning to provide search results that users are more likely to engage with. Which is a complex process - the problem with showing Tweets with more engagement is that in doing so, you're actually encouraging more engagement as a result, which skews the algorithm.
For example, if a tweet with four re-tweets comes up high in search results, that tweet will then be seen by more people via search, giving it more re-tweets, while a more relevant tweet might be available for display but it won't get the same opportunity for exposure over time.
Twitter's sought to resolve this potential bias by examining various engagement models.
"A person's behavior on Twitter, for example, provides an invaluable source of relevance information. This data contains both attributes about the Tweets shown and the consumer's reactions to them. Using this information, we can train machine learning models that predict how likely a Tweet is to be engaged with (Retweets, likes and replies). We can then use these models as scoring functions for ranking by treating the probability of engagement as a surrogate for the relevance of Tweets."
So not engagement factors alone, but they are significant indicators.
The change presents both an opportunity and a challenge for marketers.
On one hand, it gives businesses an extra way to get more exposure for their tweets - if search is no longer focused on chronology, but engagement, that means your best tweets are likely to have a longer life span as they'll keep coming up in search results for that topic over a longer period of time.
But then, of course, that also means you need to be producing engaging tweets in the first place. No doubt this is already a key aim for all Twitter marketers, but this change will put more emphasis on generating tweet engagement.
But the key factor here is how much traffic the change will actually drive - will it really make much of a difference to discovery volume? Twitter reportedly facilitates more than 2.1 billion search queries per day, so there are a lot of people looking for information on the platform via search. You'd expect many of those would be users looking for the latest updates about news and events, and TV shows, but even so, there is clearly a high amount of search activity to tap into - and maybe newsjacking will also now become more valuable as an engaging tweet that taps into an evolving trend could get significantly more attention by also seeing a search exposure boost.
It's worth considering how this could play into your Twitter marketing efforts.
Follow Andrew Hutchinson on Twitter