With influencer fraud becoming more and more prevalent, and as the industry continues to search for ways to address fraudulent activity, we’re leading the charge and releasing new, automated tools to help detect fraudulent behavior and manage influencer risk.
Our new Follower Fraud and Engagement Fraud indicators are powered by machine learning algorithms that assess a statistically significant sample of an influencer’s followers, as well as accounts that engage with their posts, to determine whether a given Instagram account is from a real person or if the account is operated by a bot. Based on that analysis, an influencer is flagged as being a high, medium, or low risk of having purchased followers or engagements:
- Follower Fraud: Indicates abnormal follower-to-engagement rate ratio, and/or patterns of bot activity such as rapid increases in followers over time or that a large percentage of followers from developing countries.
- Engagement Fraud: Indicates abnormal engagement rate-to-follower ratios, bot activity among accounts that engage with the influencer, and/or patterns of inauthentic or abnormal engagement.
“Thanks to the sophistication of influencer marketing platforms, the number of influencers being activated by marketers has grown exponentially over the last two years–there were more than 21 million #sponsored posts created in 2017 alone,” said Lyle Stevens, Co-Founder and CEO of Mavrck. “Unfortunately, the methods by which some bad actors try to game the system with fraudulent followers and engagements have also grown in sophistication to the point where visual inspection is no longer good enough to filter out fraud, especially if you’re activating thousands per month like some of our Fortune 100 customers. To help address this growing problem, Mavrck has developed an automated way to flag an individual as potentially engaging in fraudulent activity so that a marketer can filter out bad actors.”
Download Product Overview
Investigating a sample of 3,992 Instagram influencers with at least 5,000 followers, we found that 394 (9.8%) engaged in fraudulent behavior by buying followers or engagements. To address this issue, we developed a machine learning algorithm that assesses a statistically significant sample of an influencer’s followers, as well as accounts that like or comment on that influencer’s posts, to determine if an Instagram account is from a real person or if the account is operated by a bot for the purposes of creating artificial followers, likes, and/or comments.
Based on our analysis, an account can be flagged as being High, Medium, or Low risk of having purchased followers or engagements. This helps our customers to ensure that their campaigns are being executed by influencers who best fit their brand needs and objectives. With beta testing now complete, we are rolling the capability out to select customers as a premium feature and will make it available to more customers in the coming months.
Learn How to Asssess Influencer Fraud and Manage Brand Risk
Many brands, agencies, and platforms continue to rely on superficial metrics like reach, impressions, and engagements to quantify influence and influencers’ performance (and therefore, influencers’ value), which continues to fund fraud culture.
To help address this growing problem, we’ve also created a webinar discussing how marketers can better determine whether or not an influencer has engaged in fraudulent behavior, as well as introduced a new framework for measuring influencers’ value. Based on our best practices gleaned from years spent identifying patterns of behavior, in our latest webinar, Learn How to Detect Influencer Fraud & Assess Brand Risk (in 7 Minutes or Less), you’ll learn:
- What is influencer fraud?
- What signals indicate fraudulent influencer behavior
- How to assess your risk for influencer fraud
View this webinar on-demand to get started.