How the Koinon Fair Score Is Calculated (Third-Party Creator Scoring Methodology)

Updated 2026-08-07

Every creator in our index gets a Koinon Fair Score (0–100). Once a creator verifies they own the account, they can claim their own certificate and use it in a media kit or a pitch. This page opens up the whole method, because a score you cannot audit does not deserve to be called fair.

In one line: it measures whether the numbers are real, not whether they are big

Follower count carries zero weight. Followers can be bought, and the number does not answer the question a brand actually has: are those people my customers? Size enters the calculation in exactly one place — building the peer cohort so like is compared with like.

The four components

ComponentWeightWhat it measures
Authenticity35%Whether the follower/engagement structure looks natural: large followings with near-zero engagement, follow-farming ratios, and high-follower accounts that barely post all lose points
Engagement quality25%Real engagement relative to the account’s own size
Audience value30%Whether the audience is decision-makers and practitioners rather than a general consumer crowd (from audience profiling)
Focus & activity10%Whether the account consistently serves one defined niche and is still publishing

The three rules that make it fair

1. Percentile first, weight second. Raw values are not comparable across platforms or sizes: LinkedIn does not publish follower counts, and engagement rates on short-video platforms are structurally higher than on text platforms. So each component becomes a position inside a platform × size × industry cohort before it is weighted. Your score answers "where do you sit among your peers", not "what is your absolute number".

2. Missing data is not a low score. If a component has no raw material — for example the audience profile has not been built yet — that component is excluded and its weight is redistributed, rather than being scored as zero. Scoring it zero would punish creators for what we failed to collect. The certificate states whether it is the basic or the full version.

3. Too few samples, no claim. A component needs at least 8 comparable peers in the cohort to be scored at all, and a cohort needs at least 30 members before the certificate shows a ranking. We would rather say less than say something unfounded.

The three bands

Upgrades apply immediately; downgrades require 14 consecutive days below the line. Without that hysteresis, borderline accounts would flip bands weekly and nobody would display the badge.

How the fair partnership range is derived

The range on the certificate uses exactly the same formula as our public rate estimator:

range = average real engagements per post × cost per engagement × (0.7 – 1.4)

Two things have to be stated plainly. First, it is the average post, not the best one — those differ by multiples, and using the peak as if it were typical overstates a creator systematically. Second, the cost-per-engagement input is the estimator’s default, not a market rate we measured, which is why it is printed directly on the certificate for anyone to recompute with their own assumption. The range only appears when we genuinely hold average-basis engagement data; when we do not, the field stays empty rather than guessing.

What we refuse to do

How brands should read the score

The Fair Score answers "is this account real and is its audience valuable". It does not answer "is this creator right for you" — that depends on whether their audience and your customers are the same people, which is what audience matching is for. A 90-scoring creator whose followers are not your buyers is still a zero for you. Do not collapse the two questions into one.

Find creators in your industry in a minute

Sign up with email to start

Open Koinon Link →