Twitter's moderation efforts and the brand impact of the bot cull

Zenith's Sam Bueno De Mesquita shares a unique analysis of the july Twitter bot purge and what this means for brands - particularly those who work closely with influencers.

Over the last few weeks, Twitter CEO Jack Dorsey has repeatedly referred to his company as a ‘public square.’ He personally defended the decision to allow conspiracy theorist and supplement salesman Alex Jones to remain active on the service back in August – but then a month later, Twitter followed Facebook, Apple, Spotify and YouTube in banning both Jones’s personal account and that of his Infowars platform.

Twitter is a for-profit enterprise, with no obligation to allow hateful rhetoric in the name of free speech, and that’s why their inconsistent and muddled approach to moderation has felt so jarring. Dorsey admitted to the Senate Intelligence hearing this week that the company has been “unprepared and ill-equipped for the immensity of the problems we’ve acknowledged.”

But they’ve at least made some progress in tackling fake and bot accounts, particularly with the bot purge of July this year. Sam Bueno De Mesquita from Publicis Media Content analysed its impact on brands, paying particular attention to those working with influencers.

Publicis Media’s SocialTools platform tracks the top 37,000 brands on Twitter. Sam had a look at the drop to see what we could learn:

1) The average brand didn’t lose many followers

Twitter claims to have deleted 6% of accounts in this purge. But if your brand’s social media team is sending a self-congratulatory round robin email about how they lost ‘only’ 3% of followers, someone should have a word. Brand accounts on average lost just 1.42% of their followers.

2) Where there’s money, there’s bots

In general, we see lots of bots working in categories where there is money to be made by Tweeting brand content - either those areas where influencers are being paid based on their follower accounts (Apparel, Beauty), or where there are trackable affiliate and voucher programs (consumer electronics, personal goods).

Source: Publicis Media

3) Some influencers have not been influencing very many actual human beings

We are shocked (Shocked!) to discover that ‘Influencers’ have been buying fake followers.

Of the accounts tracked by SocialTools that lost more than half of their followers, 45% were ‘influencers’. That included one self-described ‘Social media guru’ from Pakistan who lost over 330,000 followers.

4) South-East Asia and the Mediterranean are top spots for bots

Looking at countries where we track over 100 branded Twitter accounts, South-East Asian countries appear 3 times in the top 5, while Italy leads the table with 4% of brand followers disappearing in the cull.

Source: Publicis Media

What can we learn from this? A lesson in simplicity

Although Twitter has the reputation for being riddled with bots, it has the advantage of being a great deal more transparent than the other digital networks. We simply don’t know how many bots there are on Facebook, Instagram, Google and the rest.

This simple fact ensures that digital metrics are a work in progress, and marketers should aim to reduce complexity wherever possible. When even something as simple as a follower count has multiple definitions, there are limits to what you can learn from the (literally) hundreds of metrics offered by Facebook, Twitter and Google.

Track your digital spend on the simplest metrics possible, because everything else changes all the time.

(Oh, and if you’re working with influencers, do it through a respectable agency or MCN. It’s a jungle out there).

Sam Bueno De Mesquita is senior content strategy director at Zenith

A note on the data: The accounts deleted are known bots, most of which Twitter suspended years ago, but didn’t delete from users’ follower counts. SocialTools tracks publically available data from 180,000 brand and influencer social media accounts, obtained via APIs from Twitter, Facebook, Google and LinkedIn. The data is from the period 1st-15th July, and we have removed from the figures brands that seemed to be rapidly gaining followers over this period, in order to prevent this growth skewing the data.