Understand the 7 AI scoring agents that evaluate creator quality — content risk, authenticity, brand safety, audience quality, sentiment, community trust, and ROI prediction.
When brands evaluate influencers, they need more than a single number. CreatorScore’s 7 AI scoring agents each evaluate a distinct dimension of creator quality, producing a transparent, explainable score from 1-100. Here’s what each agent measures, why it matters, and how it affects the final score.
The Content Risk Agent is the most heavily weighted agent, reflecting the reality that a single piece of harmful content can destroy a brand partnership overnight.
This agent uses a 5-component model analyzing every piece of creator content:
Hate speech scores above 90% cap the overall CreatorScore at 35. NSFW scores above 95% cap at 35. These are non-negotiable safety thresholds.
Creators with 0 analyzed posts are capped at 50/100 (insufficient data), fewer than 5 posts cap at 70, and fewer than 10 cap at 85. This prevents new or low-content creators from receiving artificially high scores.
The second most important dimension. Authenticity determines whether a creator’s audience and engagement are real.
Bot rate above 60% caps score at 20/100. Engagement pod rate above 80% caps at 30/100. Read more about detecting fake followers.
Beyond content, this agent evaluates the creator’s broader reputation and partnership track record.
Not all audiences are created equal. This agent evaluates whether a creator’s followers are the right match for brand campaigns.
An audience that writes thoughtful comments and asks genuine questions is exponentially more valuable for brand campaigns than one that leaves emoji-only responses.
How does the public feel about this creator? Sentiment analysis provides a pulse check on audience reception.
CreatorScore uses Claude AI (Anthropic) to reclassify borderline comments that automated NLP models get wrong, improving accuracy on sarcasm, cultural context, and nuanced language.
Trust is built over time through consistent behavior. This agent evaluates the creator’s conduct and compliance track record.
Disclosure compliance below 10% (with verified brand ad data) caps the overall score at 35/100.
The only forward-looking agent. While other agents evaluate historical data, the ROI Prediction Agent projects future campaign performance.
This is a unique differentiator—no other scoring platform includes predictive ROI modeling as part of the core scoring system.
Each agent normalizes its raw signals to a 0-100 scale, then the weighted average produces the final CreatorScore:
CreatorScore = (Content Risk × 0.20) + (Authenticity × 0.20) + (Brand Safety × 0.15) + (Audience Quality × 0.15) + (Sentiment × 0.10) + (Community Trust × 0.10) + (ROI Prediction × 0.10)
After the weighted average, knockout factors are applied. If any knockout threshold is breached, the score is capped at the knockout level regardless of the weighted average.
Every score comes with SHAP explainability—transparent drivers showing exactly which factors pushed the score up or down. No black boxes.
For the full technical methodology, see our Scoring Methodology page.
What is creator scoring? Learn how AI-powered platforms evaluate influencers on a numerical scale using multi-agent scoring, weighted dimensions, and explainable algorithms.
What is a brand safety score? Learn how influencer risk is quantified, what factors go into brand safety scoring, and how AI-powered platforms calculate creator risk on a 1-100 scale.
Learn how to vet influencers before brand partnerships. Step-by-step process covering audience authenticity, content safety, FTC compliance, and ROI prediction using AI-powered tools.