Favikon Methodology | Algorithm
The Favikon Algorithm is the engine behind how influencers are ranked and scored on the platform. It evaluates multiple factors, including engagement, consistency, reach, and content quality across social media networks. In this article, we'll explain how the algorithm works and what metrics it focuses on.
This article is one of the 4 part series explaining Favikon's scoring & ranking algorithm in detail. See all articles here:
1. Favikon Methodology | Tiers
2. Favikon Methodology | Algorithm
3. Favikon Methodology | Authority Score
4. Favikon Methodology | Influence Scores
Favikon Methodology
The Algorithm🧮
Our scoring system uses the centile method to evaluate and rank a creator’s influence on social media. Here’s how it works in simple terms:
What Are Centiles?
Centiles divide a dataset into 100 equal parts, where each part represents 1% of the total. In other words, a centile shows the value below which a certain percentage of observations fall. For example, the 20th centile means that 20% of the data is below this value.
How We Use Centiles to Rank Creators
We follow a straightforward process to rank data and assign scores out of 100:
1. Data Collection: We gather data on specific metrics like monthly views, engagement, or follower growth.
2. Sorting the Data: The data is arranged in ascending order, from the lowest to the highest.
3. Calculating Centiles: The sorted data is then divided into 100 centiles.
4. Assigning Scores: Each centile corresponds to a score from 1 to 100. For instance, if a creator’s views place them in the 75th centile, they receive a score of 75.
Illustration:
How Monthly Views Are Scored
Let’s break down how monthly views might be scored using the centile method:
• Imagine we have a list of users with monthly views like this: [100, 150, 200, …, 1050].
• We sort the data from lowest to highest.
• Then, we divide the data into centiles.
For example:
• The user with 100 views falls in the 1st centile and gets a score of 1.
• The user with 1050 views falls in the 100th centile and gets a score of 100.
• This approach allows us to rank each user based on their performance relative to others.
Weighted Scoring: Fine-Tuning with Evolutionary Algorithms
To refine our scoring system, we use an evolutionary algorithm. This helps us assign the right importance (weight) to different metrics like growth, audience size, and engagement.
Here’s how it works:
1. Initialization: We start with a diverse set of possible solutions—different ways to weigh each metric.
2. Evaluation: Each set of weights is tested to see how well it ranks influencers.
3. Selection: The best-performing weights are kept and used to influence the next round of testing.
4. Crossover and Mutation: We mix and slightly alter the best weights to create new sets, ensuring continuous improvement.
This method helps us optimize the scoring system, ensuring that the most important factors have the most influence on a creator’s overall score.
Visual Representation
In the visual dashboards (like the one you provided), you’ll see:
• Colored Bars: These show how a creator’s performance stacks up. Green indicates top performance, orange is moderate, and red is below average.
• Trophy Icons: These represent the importance of each criterion in the overall score.
• Overall Score: This is calculated by combining the scores from different criteria, each weighted according to its importance.
This approach gives a clear, fair, and easy-to-understand measure of a creator’s social media influence.
Illustration
Let’s break down the social media scoring methodology using the detailed stats from Gael Monfils’ “X” (formerly Twitter) profile, as illustrated in the provided image.
Step 1️⃣: Data Collection
The image shows key metrics such as total followers, engagement per post, views per post, and follower growth. This data is collected to assess the overall influence and activity on the platform.
Step 2️⃣: Sorting the Data
Each metric (like total followers, engagement, and views) is ranked against other users. For instance, if we were evaluating “Engagement per Post,” we’d sort the engagement numbers from all users in ascending order.
Step 3️⃣: Calculating Centiles
Once the data is sorted, we divide it into 100 centiles. For example:
• Gael Monfils’ engagement per post is 4.2K, which is placed in the 91st centile. This means his engagement per post is higher than 91% of users on the platform.
• His views per post of 134.2K place him in the 90th centile, showing strong content visibility.
Step 4️⃣: Assigning Scores
Each centile directly corresponds to a score out of 100:
• Engagement per post in the 91st centile earns a score of 91.
• Views per post in the 90th centile earn a score of 90.
These scores are then combined to reflect overall performance. In Gael Monfils’ case, the weighted scores for each category contribute to his overall Influence Score of 88.8/100.
Step 5️⃣: Weighting the Metrics
Different metrics carry different importance, which is reflected in their weights. For example:
• Engagement per post carries a weight of 22.99%, indicating it has a significant impact on the overall score.
• Views per post has a weight of 15.41%.
• Growth is weighted at 8.75%, showing that while it’s important, it doesn’t have as much influence as engagement or views.
This weighting ensures that the most critical factors in social media influence, like engagement and views, have the most impact on the final score.
Step 6️⃣: Calculating the Final Score
The overall score of 88.8/100 for Gael Monfils on X is calculated by multiplying the score for each metric by its weight, then summing them up. The result is a balanced score that reflects his overall influence on the platform, considering his strengths in engagement, views, and audience size, while also factoring in areas like growth.
Summary
Using the centile method and weighted scoring, the system provides a clear and fair assessment of a creator’s social media influence. In this example, Gael Monfils’ score of 88.8 reflects a strong presence on X, driven by excellent engagement and viewership, while also highlighting areas like growth that may have room for improvement.