Sign in


Frequently Asked Questions

Last updated: May 2019

Work-in-progress. Hopefully I will find time to write about the site's data collection, analysis methodology and machine learning model. For now, these are questions I get asked the most. You can send yours to [email protected]

What are fake followers

Fake followers are followers with behaviors that fit a certain set of criterias of being ran by non-human(bots).

We use the following labels to train our statistical learning model [7]:

  • Lack of profile photo, generic description and/or originated from countries that speak different languages from influencer
  • Auto-likes based on tags
  • Auto-likes within seconds of parent post
  • Auto-follows and constant follow / unfollow pattern
  • Sudden spikes and growth in likes and followers trend

In addition bots and bought followers, it also considers engagement groups with real users trading likes and follows as fake followers (podding).

Fake followers ratio is a prediction from machine learning model. It is not 100% accurate.

Youtubers do not have fake subscriber calculation (yet).

What are power followers

Followers with more than 10k followers themselves.

How accurate is your demographic data

The demographic data is fairly accurate when viewed in aggregation, and less so when viewed individually. For example, the age and gender data are very accurate for fashion influencers as a group. Inaccuracy does appear for specific influencer ~5% of the time. Mostly due to influencers having more than 2 major niches or languages (an influencer that posts about travel and tech in 2 or 3 languages). That affects her followers distribution, which in turns affect the demographic calculation.

To be more technical, demographic data are calculated using different ML image and NLP models:

Gender: Gender is identified by analyzing followers real name [1] and profile image [2]. Androgynous label is given to names like Sam.

Age: Age is identified by analyzing followers profile photo and their 5 most recent instagram photos (think HOW OLD DO I LOOK photo app) [2] [3]

Country: By far the most inaccurate one for now. The model is still shit. We determine influencers language and country using their profile description and post text. [4]

Niche: We use ResNet50 [5] to extract features from influencers posts and calculate their word distance to a predefined classes of niche. For example, a food influencer will most likely have features like sardines, apple and steak extracted from their posts and these features are closer to 'food' than 'fitness'. [6]

How many influencers are you tracking at the moment?

As of May 2019, this site is tracking 2,234,099 Instagram influencers and 87,614 Youtubers.

I am an influencer. Why am I not on your site?

We only track influencers with more than 10k followers at the moment. If you are a developer, you can submit influencer(s) for tracking by making a POST request.

curl -X POST \ \ -H 'Content-Type: application/json' \ -d '{"usernames":["kimkardashian", "user2", "user3"], "type": "instagram"}'

How can I submit profiles for analysis?

Developer can make a POST request to submit instagram and youtube usernames for tracking.

curl -X POST \ \ -H 'Content-Type: application/json' \ -d '{"usernames":["kimkardashian", "user2", "user3"], "type": "instagram"}'

How do you calculate engagement rate

$$Engagement Rate = {Likes + Comments \over Followers}.$$

Do you provide API access to your data

Yes. Just append json=1 to the end of URL to retrieve JSON-formatted result.

Is this site free?

Yes. We do have a few sponsors with better rate-limit, historical data and access to analytic tools such as Sentiment Analysis and Audience Overlap.

Full historical data is available with our pro plans. The most recent 3 months data are available for everyone.

What are sponsors?

Sponsors are companies who pay monthly subscription plan for historical data and advance analysis tools. In addition, subscribers get preview access to upcoming features (youtube, twitch) before they are released to everyone. You can sign up for the plans at

Why are you doing this? started out as a weekend project to gather annotated data for training purposes. Social influence is increasingly being traded commercially (fake Google Map reviews, Amazon reviews, Instagram followers, Youtube views etc) so I am trying to train a model to predict the authenticity of this growing influence.

Is this AI-powered™?


Exclusion of Liability

The published information has been collated carefully, but no guarantee is offered of its completeness, correctness or up-to-date nature. No liability is accepted for damage or loss incurred from the use of this site or the information drawn from it. This exclusion of liability also applies to third-party content that is accessible via this offer.

[1] Zed Gecko gender-verification by forenameZed Gecko
[2] IEEE Workshop on Analysis and Modeling of Faces and Gestures (AMFG), at the IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Boston, 2015Gil Levi and Tal Hassner | Department of Mathematics and Computer Science | The Open University of Israel
[3] Rude Carnie: Age and Gender Deep Learning with TensorFlowDaniel Pressel
[4] Compressing text classification modelsA. Joulin, E. Grave, P. Bojanowski, M. Douze, H. J├ęgou, T. Mikolov
[5] ResNet-50 Pre-trained Model for KerasKaiming He Xiangyu Zhang Shaoqing Ren Jian Sun | Microsoft Research | {kahe, v-xiangz, v-shren, jiansun}
[6] Efficient Estimation of Word Representations in Vector SpaceMikolov, Tomas; et al.
[7] Identifying a Large Number of Fake Followers on InstagramSRF Data, Jennifer Victoria Scurrell, Timo Grossenbacher ([email protected])