This data was originally featured in the July 19, 2023 newsletter found here: https://www.trustinsights.ai/blog/2023/07/inbox-insights-july-19-2023-innovation-prerequisites-twitter-vs-threads/.
In this week’s Data Diaries, let’s take a look at whether Threads has had any material impact thus far on Twitter. Threads was clearly intended to be a competitive service for Twitter. Now that it’s been almost two weeks since its launch, has it had an appreciable impact on Twitter? Let’s find out.
First, there’s no way to know in real-time just how much impact Threads has had globally on Twitter; we can only make that assessment from the accounts we have control over. Because of this, your mileage will vary; some Twitter audiences are firmly loyal to Twitter, while other audiences will want to participate elsewhere for any number of reasons.
So, how would you go about determining Threads’ impact? Essentially, you’d conduct a retroactive A/B test, where you export your Twitter data for the past few months and use a statistical technique called propensity score matching to find tweets with similar performance (to rule out co-variates), with the differences being what’s happened.
You can do this in a programming language like Python with its Causal Inference library or in R with the MatchIt library. The way these software packages work is that they look at the treatment period – the period you want to measure – compared to other similar periods of time, based on variables you provide. This allows you to control for things like other activities, other promotions, other interfering data that could otherwise contribute to an anomalous test result. That’s what makes it a retroactive A/B test and better than simpler tests like taking the average of the post-Threads period versus the pre-Threads period, or using a t-test.
So let’s take a look at three different Twitter accounts to see how they’ve fared since the launch of threads. First, we’ll look at the Trust Insights account:
What we see is… unexpected. Instead of a decline, we see pretty significant gains across almost all metrics. Okay, that’s unexpected, but that’s a sample of 1. Let’s move onto my personal Twitter account:
As with the Trust Insights account, we see an increase in many metrics for Twitter. All right, let’s move onto the third account:
This is a significant change; in this account, since the launch of Threads, almost every metric is down substantially compared to the previous period.
So what gives? Why does this look so strange? Every audience is different. Every audience participates on social networks for their own reasons, to be parts of communities that serve their needs. In the case of Trust Insights and my personal Twitter account, these are established audiences that have, in many cases, built up long-time followings of people who have been on Twitter for years. They may have some inertia – certainly, many have tried Threads, but they may still stick with the network they know, especially if they’ve built up a following. Some may even be doubling down on their Twitter activity.
Conversely, the Warrior Nun Substack Twitter account is very new, established in January of this year. Its audience is also far less loyal to Twitter; often, you’ll note in people’s bios sayings to the effect of “I’m only here to save Warrior Nun”; they have no great loyalty to the platform itself. As a result, other options may be more appealing.
The key takeaway here is not to look at any one result, but to run the same tests yourself, on your own data. Your audience might be more engaged or less engaged, and your results likely will differ from even industry peers. Check your own data, then set your strategy appropriately.
Need help with your marketing data and analytics?
You might also enjoy:
Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, INBOX INSIGHTS. Subscribe now for free; new issues every Wednesday!
Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new 10-minute or less episodes every week.