About IG Audit
IG Audit was built on the belief that authentic brands and influencers should have a way to stand out. For any public profile, we look at a random sample of followers, and in ~15 seconds, estimate their real follower ratio using an internal classification algorithm.
This is useful for marketers who need to verify a potential influencer/partner's following, as well as the curious Instagram user.
How does it work?
- We fetch up to 200 random followers for the input user. For each follower, our algorithm looks at things like number of posts, follower/following counts, username, and whether the account is private/public, and decides whether it thinks the user is real/fake.`
Why do I see different results when I audit the same profile twice?
- IG Audit is designed to be fast first. For each audit, we examine a random sample of up to 200 followers, because examining all followers would be slow. The output percentage is an estimate, not an exact number.
How reliable is this sampling method?
- Statistically speaking, given a 200 follower sample, the output percentage will be within ~8% of the true value 99% of the time, regardless of the user's total follower count.
How does the like+comment analysis work?
- IG Audit averages likes+comments from up to 12 of the user's most recent posts. The expected like/comment values are the product of the user's follower count and the industry average like/comment rate for followings of that size. In the near future, I will also be introducing fake like/comment detection.
If I audit a user's profile, will they know?
- No, profile audits are done anonymously.
Why can't I audit private users?
- Currently, there's no way to fetch followers for a private user unless they've accepted your follow request.
Can I audit all of a user's followers?
- This is in the works. Since it takes more time and computing resources, I'll be building it out as a paid feature within the next month. DM me if you're interested in this.
Why build this?
- Many industries (music/modeling/etc) currently search for new talent through social media, but audit profiles manually. This tool intends to make their lives easier and bring more transparency to social media.
How does the follower classifier work?
- IG Audit uses machine learning to determine whether followers are real or fake. When tested on an initial dataset, the classifier correctly identified above 90% of real followers, and 80% of fake followers. In the near future, I'll be applying more advanced machine learning methods for identifying fake followers. I'll post updates here as they happen. }
Can you be more specific about the follower classifier?
- Unfortunately, giving away more specific details would compromise the integrity of the system. Think of it this way - if fake follower providers knew how my classifier worked, they'd much more easily be able to create followers that would fool the classifier.
Why not just use a site like SocialBlade to validate followers?
- SocialBlade is great for catching people who buy large amounts of followers at once. They track follower count daily, so if there is a huge jump in follower count, this is usually a good indication of inauthentic activity. However, fake follower providers have become more savvy, and are now providing \`drip followers\` - meaning you buy them in bulk, but they come in at a smooth rate over time as opposed to all at once. On SocialBlade, this method slips through the cracks because it looks just like organic growth - a steady increase in followers over a long period of time. In contrast, examining a random follower sample (as IG Audit does) will catch fraudulent followers even in this case. To have the most certainty, we recommend using both services.
Our founder, Andrew Hogue, is a Caltech class of '15 graduate who majored in Computer Science. He has previously worked as an engineer at NASA, Facebook, Snapchat, and Hooked. He founded Authentíque in 2018 on the idea that an influencer or brand's value should be based on the work they've put in to get there. We believe that real influencers and brands should be able to identify themselves, and we build tech to help them do so.
Want to support us?
Created by Authentíque, Inc. in March 2018.