In a previous article, we talked about how a platform such as HypeAuditor helps your influencer marketing campaign planning. There are many similar tools out there and over 800 companies offering influencer services. The most important point is to get one because you most certainly need it.
If you’re somewhat like us, you tend to question everything. So in this article, we decided to dig deeper into how influencer marketing tools work for identifying fake influencers.
Disclaimer. We used HypeAuditor for this experiment but we have no vested interest to promote any platform. We bought our credits on HypeAuditor with our own hard-earned $$$. We genuinely like the platform and what they’ve done to add transparency to our industry.
In order to set up this experiment, at first, we ran over 200 Southeast Asian influencers through HypeAuditor. We also had a list of 5 whom we called “suspicious”. This group came from our fellow marketers who had first-hand experience working with those influencers and were seriously doubting their stats.
Additionally, we used the top 5 most-followed Instagram accounts like Christiano Ronaldo, Selena Gomez, Kylie Jenner, Leo Messi, and Kim Kardashian West as benchmarks. We were hereby assuming that celebrities have no reasonable incentive to buy followers or engagements.
The objective of all of this was to see if we can spot any key data differences between the suspicious influencers vs. the rest. After analyzing, comparing and discussing all the stats, we came to an understanding that there seem to be 3 key indicators of an influencer who likely buys engagement and/or followers:
- Fake influencers seem to have more than 10% ‘Mass Followers’
- Fake influencers rapidly follow and unfollow other Instagram profiles
- Fake influencers have a low Engagement Rate
Even though breaking down the Audience Type by HypeAuditor seemed useful at the start, we realized every influencer has a certain level of suspicious accounts. However, without a meaningful comparison, this breakdown remains only interesting, and not actionable.
Fake profiles have more than 10% ‘Mass Followers’
On the contrary, the definition of ‘Mass Followers’ on HypeAuditor is an account that follows more than 1,500 accounts. And most other influencers we analyzed had less than 10% of Mass Followers vs. the suspicious ones.
In the sample above, this “influencer” has almost 40% ‘Mass Followers’ accounts. Additionally, if you look closely at the audience’s location, they are mostly located in the United States rather than the origin country of this “influencer” which is in Malaysia. And if it is broken down by cities, the profile becomes even more random.
Fake profiles rapidly follow and unfollow other Instagram profiles
One of the oldest trick in the book to increase your Instagram following is to follow other accounts and hope for them to follow you back.
Once Instagram started picking up as a platform, some dubious companies started building on the same idea. However, automated scripts help to make this process incredibly fast and build in rules like “unfollow if the profile doesn’t follow me back in 24h”. This is the good old follow/unfollow game on steroids.
In the above graphs, we can clearly see significant spikes above a normal mean. The tool also warns us of “Mass following patterns”, which we assume is based on growth velocity. The rapid growth of followers is great when it happens, but unlikely in the world of Social Media. Almost if… this has been engineered.
In addition, the amount of followers seems to spike on certain days and remain stagnant from around 18k to over 30k followers.
Fake profiles have a low Engagement Rate
The bots will turn ON when payment is made. Otherwise, they don’t. Hence these irregular patterns are created that leave us bits and pieces of evidence we can work with.
Fake influencers typically tend to have low Engagement Rates, because that’s why they turn to click farms, to begin with. Many of the comments underneath the posts of our “suspicious” group were dubious at best (eg. “OK”). And their ER also varies drastically from one post to the next, again likely because of what was “paid for” and whatnot.
And by the way, unfortunately, HypeAuditor has increased their pricing from (1.99$ per report to 30$ per report; a staggering 175% increase!). We still think it deserves a 4/5 rating.
This can be steep for many brands. Their new Discovery platform seems to include most of the audience analytics we spoke about here and new management features (starts from $3,588 per year; $299 per month). Maybe worth considering instead. We would love to hear from you if you’re tried it.
Even though we take deliberate steps when designing the methodology of our experiments, none of this is scientific and doesn’t pretend to be. While working on influencer campaigns every day, we seek to share insights that can help the data-driven influencer marketer already today. Sometimes, the quick and scrappy way is better than slow and rigorous.
So what did we learn here…
- Get an influencer identification tool, you will most certainly need one if you plan to run more than one campaign a year.
- 3 metrics that seem to reveal an influencer is likely buying followers and/or engagement
- Fake profiles have more than 10% ‘Mass Followers’
- Fake profiles rapidly follow and unfollow other Instagram profiles
- Fake profiles have a low Engagement Rate