Greats was about to pay an influencer with 10.5 million followers a midsix-figure sum for a long-term partnership. But the deal never went through.
That’s because Ryan Babenzien, founder and chief executive officer of Greats, discovered that the majority of the influencer’s followers were fake. In doing “due diligence,” to get a better understanding of this person’s metrics — largely to make sure her followers aligned with the brand’s target audience — Babenzien was supplied with screenshots containing details about her Facebook and Instagram followings.
“When we mapped that over the world, we found that it was mathematically impossible for her to be as popular as she was and not have any of the 10 major cities in the 10 major countries be in her top 10 follower cities. It was impossible,” Babenzien said. “I don’t know if they paid for followers or if they are bots, but there wasn’t the alignment she had in followers from the cities that she was allegedly popular in.”
He declined to name the influencer, but told WWD that negotiations were well under way with the person, who would design a capsule collection of women’s footwear for the direct-to-consumer sneaker brand, which to date has mainly focused on the men’s sector.
The above instance, which could have cost the business hundreds of thousands of dollars, wasn’t Babenzien’s first brush with dishonest content creators.
“Many lie, from the biggest celebrities in the world to a micro-influencer with 10,000 followers,” said Babenzien. “It’s just become part of the norm.”
Influencer marketing has become a multimillion-dollar investment for fashion, beauty and retail brands, with sizable portions of their advertising budgets now dedicated to elaborate activations and paying content creators. But the closer the industry looks at influencers and their followers, the more it doesn’t look like what it seems: in many cases, inflated follower counts that result in campaigns yield little return on investment.
The fashion and beauty industries — as well marketing firms, public relations agencies and influencers who have come by their followings honestly — are up in arms about the matter, which they claim is tantamount to stealing from the brands paying them based on false information.
Unfortunately, many firms discover this too late and only after shelling out tens or even hundreds of thousands of dollars. Or even worse, many are worried that dishonest content creators and bloggers will ruin it for the rest of the group by giving influencer marketing a bad rap. This means that brands, disappointed when campaigns underperform because influencers fail to generate the traction that someone with a following of their size is expected to, will begin to tighten their belts.
“If a brand works with five influencers and three have large fake followings, then the brand isn’t going to get the ROI and they’ll think influencer marketing doesn’t work. For influencers, that’s such a dangerous thing. The idea that influencer marketing isn’t valuable is probably the single biggest threat to the industry,” said James Nord, cofounder and chief executive officer of Fohr Card, a digital agency that works with influencers and brands predominantly in the fashion, beauty and lifestyle spaces.
He acknowledged that the foundational economics of the entire industry are based on followers and engagement — both of which have become easy to purchase. And the fact that content creators are paid based on certain thresholds makes it all the more tempting to buy a following. If someone tacks on 10,000 or 50,000 more followers it means they can make more money tomorrow. Additionally, an “everyone-else-is-doing-it-too” attitude pervades influencers, many of whom think it’s become the norm to add a little padding to their follower counts.
At Fohr Card, Nord’s team built a Follower Health tool to monitor the situation and help brands partner with influencers with authentic followings. Participating influencers get a Follower Health score that determines the “health” of users following them, inclusive of bots, accounts that “look more like real people” and non-bought followers who follow so many people that even if they do follow the influencer, there’s a small likelihood they’ll ever see that content because of Instagram’s algorithms.
“That’s why we call it Follower Health, because without bank statements, without actual proof, we can’t prove that they’ve actually purchased these followers. We just know their following is made up of accounts not very likely to see their content, so it’s not a very healthy following,” he noted.
Today the majority of Fohr Card’s clients no longer work with influencers with low scores. But on the rare occasion that a fashion or beauty company knows someone bought followers and still wants to work with them, Nord will just take the information and use it to negotiate better terms. An example: if Fohr Card has proof that 30 percent of a content creator’s following isn’t real, then the brand will offer to pay 30 percent less than that individual’s standard fee.
Not everyone is guilty of augmenting their followings, though. Nord cited Wendy Nguyen of Wendy’s Lookbook and Marianna Hewitt as two influencers with “nearly perfect” Follower Health scores.
Hewitt, even though she’s considered a top beauty influencer, still has yet to reach one million followers. She has 782,000 — and she’s very comfortable with that.
“I’m so confident in the platform I built and the audience I have but I can see how someone would easily fall into this trap…A while ago there was the pressure of having to hit certain numbers, but it’s definitely something I don’t feel pressured about — which is another reason I’m fine growing at the rate I grow,” said Hewitt, who insisted that she has never bought any followers.
She admitted that, to an extent, follower count matters when it comes to inking the top echelon of brand deals, but not as much as it once did. With the emergence of the brand builder versus sales converter distinction, many brands want to work with a top influencer (whether they’ve purchased followers or not, or in Hewitt’s case, even if she doesn’t have the million-plus follower count her contemporaries do) “because of that person,” she said. Sometimes it’s less about the sales or other metrics and more that a brand just wants to associate with a particular influencer.
“There are people really good for brand awareness that maybe have a ton of followers but the exchange isn’t high, and there are smaller influencers who push a lot of sales,” Hewitt said. “There are personalities people want to partner with and that’s OK. I don’t think there’s anything wrong with that. They just have to know their goals going into it.”
Benefit Cosmetics has made some of the most significant investments in influencers across the entire beauty industry — including spending a reported $10 million on influencer initiatives to support its BadGal Bang Volumizing Mascara launch this month.
Encouraged by positive ROIs from past influencer programs (as well as being armed with a host of new tools that help to sniff out frauds), the LVMH Moët Hennessy Louis Vuitton-owned brand has a no-tolerance policy when it comes to fakes.
“If Benefit gets credible info on an influencer with a lot of fake followers, we simply stop working with them,” Toto Haba, senior vice president global digital at Benefit, told WWD.
Asked how Benefit identifies fake followers, the executive pointed to a few paid tools that can estimate the percent of fake followers on an account, including InfluencerDB, SocialBakers, Social Audit Pro and SocialBlade.
“Every media channel has some level of fraud that marketers must manage [and] tolerate. Whether it’s inflated impression numbers from nonviewable ads or bots driving up search clicks, Benefit is constantly trying to optimize our spend versus what drives business results. With that greater context, Benefit continues to see good ROI from working with influencers,” Haba said.
He estimated that the brand only engages with roughly 10 percent of the people posting about Benefit so there are “more than enough” alternative influencers to choose from. Additionally, the brand tends to only partner with influencers who have a deep knowledge of beauty and not random “Instagram models” (which there appear to be no shortage of on the platform).
“Assessing their skill level is pretty easy. It’s hard to fake makeup knowledge,” he added.
It’s also easy to assess if an influencer has the real, engaged following they say they do, according to Haba. All one has to do is monitor how much engagement an influencer post gets, and as a result measure how much conversation your brands gets because of said post. If this content creator is able to get people liking, talking and engaging with your own brand channels (in addition to their own), it’s likely followers are authentic. If there’s no lift in these metrics, there’s a high chance that followers are fake or bots.
It’s not just the brands that are in an uproar, though. Marketing and public relations firms are also feeling the heat.
Managing influencer relationships and facilitating blogger brand deals have become key services that agencies today must offer — with many clients prioritizing digital over the traditional press placements once considered the holy grail. But the fake followers issue has added another layer of pressure. Agencies, now tasked with identifying which individuals have authentic followings, must take time away from strategizing and executing communications strategy to pore over data and metrics for what sometimes could amount to hundreds of potential influencer candidates. Publicists are now just as culpable as the brands they represent in sniffing out the fakes.
Alison Brod Marketing and Communications founder Alison Brod said the agency is taking action by working with talent that allows her team to use tools to evaluate their followings — no matter what size their followings are. Those with nothing to hide typically have no problem letting a potential client, brand or agency take a deeper dive into their follower metrics (Fohr Card’s Nord confirmed this), but it’s not always that easy.
“Being paid by a brand based on fakes is the equivalent of getting into medical school after cheating on every test. It is like doping to win the Olympics,” said Brod, who called it “cheating” and “stealing” to be paid for posts based on fake follower counts.
Unfortunately, detecting fakes on social media is hard, and even though there are tools, inauthentic content creators keep finding ways to game the system. Also, sometimes the only way to obtain comprehensive data is via the influencers themselves, who must give brands and data or marketing firms permission and access to analyze their metrics. If someone doesn’t want to give this information, it’s hard to definitively come to a conclusion about one’s following. Plus, the process is time consuming, considering the thousands of influencers out there today.
For a given account, an intrepid investigator may pluck a popular post, drill down into who liked or commented, and evaluate the users for authenticity or fakery. Now multiply that by billions of shares across hundreds of millions of accounts, and the task’s complexity starts to come into focus. Manual methods can’t cover the breadth, scope and velocity of a problem that has become epidemic.
That’s why, in the battle of the bots, social media companies are fusing artificial intelligence with human-powered efforts. Think of it as fighting fire with fire.
Like any tool, AI bots aren’t inherently bad. Businesses use autonomous software to great benefit, whether generating content or chatting with customers in messaging apps. Retailers, brands and others turn to bots for everything from customer service to product recommendations and styling services. But in the wild west of the Internet, where social influence can rake in millions of dollars and even sway elections, the scourge of fake followers can have companies blowing big bucks without the commensurate reach to show for it.
The effect on brands and public perception isn’t the only impact of the bot onslaught. The noise can bury rising stars with genuine fans.
“The only difference between a micro-influencer and a macro is reach,” said writer Lisa Jean-Francois, aka lisa a la mode, whose content on fashion, beauty and lifestyle has drawn 30,000 followers.
“The problem, of course, is that would-be influencers have learned that brands seem to be less interested in the quality of the content [than] the popularity of said influencer. So these would-be influencers have figured out a way to game the system: They buy bot accounts so they appear to have the reach of a macro influencer,” she added.
And there’s no shortage of purveyors selling fakes.
The New York Times brought the issue to the forefront in an exposé late last month that called out Devumi, a company that sells Twitter followers and engagement. Devumi allegedly took in millions of dollars for selling retweets and phony Twitter followers, some of whom pose as real people. Devumi bots have been found among followers of celebrities like model Kathy Ireland and even Twitter board member Martha Lane Fox. The company denies any knowledge of this, but investigations are underway, most recently by the attorney general in New York.
Twitter is at the heart of the Devumi matter, as well as congressional inquiries into social’s role in the 2016 presidential election. Now it seems to be taking a surgical approach. After the micro-messaging platform tweeted that it would target companies like Devumi, millions of presumably fake followers disappeared.
It also suspended accounts linked specifically to Russian propaganda efforts, like that of the Internet Research Agency, a group tied to the Russian government. And the company expanded e-mail notifications to Americans who engaged with such accounts during the election period, going from 677,775 to roughly 1.4 million people.
The sweat equity in policing networks like Twitter, which has about 328 million users worldwide, looks overwhelming. Research from the University of Southern California and Indiana University estimated that bots made up to 15 percent of its userbase last year. In other words, at least 30 million fake accounts may have been swamping the platform.
Following the election, Twitter introduced an Information Quality initiative.
“We detect and block approximately 523,000 suspicious logins daily for being generated through automation,” the company wrote on its blog. “In December 2017, our systems identified and challenged more than 6.4 million suspicious accounts globally per week — a 60 percent increase in our detection rate from October 2017.”
If the tactics seem vague, that’s by design. The company wants to be transparent, said a Twitter spokeswoman, but “we do have to be careful to not inadvertently reveal details of our antiautomation measures that would help bad actors evade our policies.” Its blog mentioned a few techniques, such as enhanced phone verification and authentication challenges, like reCAPTCHAs. Twitter also targets “malicious automation” signaled by nearly instant replies, patterns or non-random tweet timing, and coordinated engagement, the company wrote.
Facebook uses both automated and manual systems to block fraudulent accounts on its 1.4 billion-user network, as well as on Instagram. In earnings calls following Facebook’s congressional testimony, chief executive officer Mark Zuckerberg pledged more resources to battle fake news and protect the network. The company now has about 10,000 people working on safety and security, and it plans to grow the number to 20,000 over the next year.
“In many places, we’re doubling or more our engineering efforts focused on security,” said a Facebook representative. “And we’re also building new AI to detect bad content and bad actors — just like we’ve done with terrorist propaganda.”
According to the Facebook official, security systems run in the background millions of times per second to find and remove threats before they see the light of day.
In 2014, singer Justin Bieber shed 3.5 million Instagram followers when the Facebook-owned business booted phony accounts. The fine line between blocking fakes and impeding real users snapped into focus in January, when Insta influencers and marketers expressed concerns about the company limiting posts — initially to just 10 percent of a user’s followers — before distributing to the rest based on performance. An Instagram spokesperson denied it, saying that “no content posted by an account is ever hidden from the feeds of those who follow that account.” But such moves seem feasible in these bot-crazed times.
Snapchat sidesteps many of these issues, thanks to the ephemeral nature of its posts. In other words, the payoff of spamming, using fake accounts or circulating pics via bots isn’t worth the effort, not when all content — legitimate or not — self-destructs in a day by default. Still, its Abuse Engineering team focuses on “hardening” the app against malicious actors. The work is similar to that of financial apps and banks, making Snapchat “very difficult to reverse engineer,” a spokeswoman for Snap Inc. said.
For Jean-Francois, fixes for the fake-follower problem can’t come soon enough. But they may not change her decision to leave the business and turn her fashion and beauty focus into a hobby.
On one hand, she understands why brands flock to influencers with big audiences. But on the other, she also blames them for creating the situation. “The fact that quality of content comes second to popularity is a huge problem,” she said. “I have had some wonderful opportunities, and I hope to have more, but only if I can be respected as a professional, regardless of my following count.
“Unfortunately, I just don’t think this particular industry allows for that anymore,” she added. “There may be some exceptions to this rule, but soon it will become a standard, and I need to get out before that happens.”