Unilever's chief marketing officer Keith Weed called for “urgent action” to clean up the influencer marketing ecosystem, immediately committing to stop working with any influencer found to be buying followers and prioritise partners who are actively trying to improve transparency.

“The key to improving the situation is three-fold: cleaning up the influencer ecosystem by removing misleading engagement; making brands and influencers more aware of the use of dishonest practices; and improving transparency from social platforms to help brands measure impact,” Weed said.

“We need to take urgent action now to rebuild trust before it’s gone forever.”

Brands including L’Oreal, eBay, Samsung and Diageo were all in agreement – it’s a problem and one they have been quietly trying to figure out their own solutions to.

Many media and social agencies The Drum spoke to have also let out a collective sigh of relief that an advertiser of Unilever's might ($7bn in annual ad spend) had publicly called for change.

“It’s about time,” said one agency chief executive.

“It was a no brainer this would happen,” said another.

How did the industry get to this point?
The first challenge in combating the problem is pinning down a figure on real versus fake accounts on social media – a virtually impossible task. According to some estimates, as many as 15% of Twitter’s 'users' may be fake while up to 60m Facebook accounts could be automated, or bots.

Despite this opacity, the influencer marketing sector has been growing at a rapid rate, with one company, Mediakix, estimating it to be a business worth in excess of $1bn in 2017 and one which is predicted to continue growing exponentially over the coming years.

Marketers, meanwhile, are willing to splash out as much as £75,000 for a single social media post which mentions their brand by someone with over one million followers and they would pay 'micro-influencers' – those with under 10,000 followers – an average of £1,500 for a mention.

In a survey of its members, the Association of National Advertisers found that 75% of marketers currently work with influencers and of that 43% plan to increase spending in the next year. And of those who are not currently using influencer marketing, 27% indicated they plan to do so in the next 12 months.

Yet only 36% said they judged their influencer marketing efforts as effective and 19% admitted they thought that the money spent was ineffective.

All of these stats can be boiled down to the following: the costs of influencer marketing is increasing, as are the budgets to fund it, but there is no standardised measurement to judge effectiveness, giving way to a market where less-scrupulous influencers can earn some quick cash with few repercussions.

Though there’s little research available into how each of the main platforms used by influencers (namely YouTube, Instagram, Snapchat, Pinterest) compare in terms of fraudulent behaviour, it's Instagram – which 40% of the ANA respondents ranked as the single most important influencer channel – which has emerged as the one of most concern.

Anti-fraud company Sway Ops found that a single day’s worth of posts tagged using the recommended hashtags #sponsored or #ad (the avoidance of which is another issue in itself) contained over 50% fake engagements and that bots are responsible for over 40% of total comments for over 500 of 2,000 sponsored posts made each day.

Among the most-duped brands, per research from Points North Group, was Ritz-Carlton which has seen up to 72% of the ‘people’ reached through its influencer programme later outed as fake. FMCG brands including Pampers, Magnum Ice Cream and Olay had fake follower rates of 32%, 20%, and 19% respectively.

Nik Speller is a senior influencer manager at Socialyse, a division of Havas which works with clients like the BBC, Topshop and O2 on their social media strategies. He told The Drum that his “crass assessments” on bot interaction would indicate that anywhere between 10% and 20% of engagement an average Instagram post is fake.

“I was looking at some Instagram demographic stats the other day for one UK influencer and over a third of their audience was based in Brazil,” he said. “Whether they are cheating the system or not, I don’t know – but either way it’s irrelevant. Does the client want to spend money advertising to people in Brazil?”

And therein lies the problem – figuring out who might be “cheating the system”, as Speller puts it, is as difficult as deducing how many fake accounts are out there in the first place.

Who is cheating the system?
One solution had been through APIs like Social Blade, which essentially monitors the follow and unfollow activity of any individual on social media. A surge in followers can indicate that they have bought engagement.

However, one outcome of the Facebook/Cambridge Analytica scandal has been a crackdown on open APIs and Social Blade, like many others, lost access to Instagram, though it still has historic data which marketers can use as indication of previous nefarious activity.

For its research into fake followers, Points North created its own algorithm to analyse influencer content – raising red flags on things such as posts with comments in languages that are out of context; multiple posts with the exact same comments or if an influencer’s follower count is high but then they have low-to-no engagement (likes and comments).

But even with these sort of tools, rooting out influencer fraud seems to be about relying on gut instinct rather than hard data.

“Most [agencies] tell you they have a secure vetting process of influencers but until an industry standard is developed, how can we tell?," said Mark Wright, chief revenue officer at an influencer marketing agency backed by YouTube star Caspar Lee. The better ones out there do capture some data, but most only capture it from the influencers page therefore ‘likes’ and ‘comments’ are as clever as it gets.

“In a world where agencies and their clients have been used to seeing the full marketing cycle played out before their very eyes, while being securely vetted, what we have currently in the influencer space is simply not good enough.”

Though Socialyse has several processes in place to weed out bad actors that come as part of the due diligence course for clients, Speller admitted that “unfortunately, there’s no real way of knowing”.

Social Circle, another influencer marketing agency, said it uses a “a weighted engagement algorithm” to work out the potential reach of an influencer campaign which can “make it far easier to stamp out fraudulent activity and assess creator audiences fairly impact measure".

“It is, however, still a problem that needs to be addressed more fully,” said its chief executive Matt Donegan. “Incorrect, faulty or even fraudulent data comes as a result not just of bad practices, but from brands and agencies not knowing the right questions to ask – and often from a lack of education on the part of the influencers.”

Time for the platforms to step up
But should it be down the clients and/or agencies to be creating their own algorithms and “asking the right questions” for this problem to be solved? As part of his action plan, Keith Weed has called on the tech giants of Silicon Valley to sort it out – saying he wants “more transparency” from them “to help brands measure impact”.

Socialyse’s Speller was less diplomatic, slamming Instagram for “burying its head in the sand” over the issue, suggesting that any admission of fraudulent activity on influencer posts may bring down the price of ads across the entire Facebook ecosystem.

“I suspect that’s why the platforms aren’t keen on tackling it – a large volume of interactions it what [Facebook] bases its market price on. If it were to turn around and admit to advertisers [some of it is fake], then it could have a serious problem on its hands. My hope is that [Weed] forces the platforms into action – they have untold wealth and brainpower to figure it out.”

In response to Unilever’s announcement, Instagram said it was “committed to upholding the integrity of our platform and to helping content creators and businesses create the best experiences for the Instagram community”.

But in lieu of any immediate action, Weed used the draw of the Cannes Lions festival to convene a group of industry representatives, including the tech platforms, to push forward the agenda of how to better detect and deal with bot accounts and influencer fraud.

Until a more robust solution emerges, many of the marketers The Drum has since spoken to have committed to invest more time into understanding and managing the relationships with these social media stars.

On an industry wide-level, the Internet Advertising Bureau (IAB) updated its Good Practice Guidance on Disclosure earlier this year to include influencer marketing to help both advertisers and influencers understand what the rules are and how to comply with them.

“Beyond that, we’re working on a blueprint to help advertisers be more confident in asking the right questions of their suppliers to ensure genuine transparency in the supply chain,” said IAB chief executive John Mew.

While the future of influencer marketing might be in question, marketers are upbeat that Weed has drawn a line in the sand and what will emerge over the coming year is a more mature sector where those investing can have better assurances over what they’ll get in return.

“We strongly believe that this announcement will inspire a renewed emphasis on transparency, that will help to shape this still young industry into one that rewards professionalism from creators and practitioners,” surmised Wright. “Unilever’s step to increase trust and transparency is not only welcome, it’s overdue.”