Every creator receives a brand safety score from 1 to 100, calculated by 7 independent AI scoring agents. CreatorScore provides a transparent, explainable influencer vetting methodology — so brands know exactly why a creator scored the way they did.
Scores range from 1 (highest risk) to 100 (lowest risk). Higher is always better.
Top-tier creator. Consistently brand-safe, authentic audience, positive community. Ideal for premium brand partnerships.
Very low risk. Strong performance across all agents. Suitable for most brand campaigns with confidence.
Low risk overall. Minor areas for improvement but generally safe for brand partnerships.
Some caution needed. Review the score breakdown to understand specific concerns before partnering.
Significant concerns identified. High risk for brand reputation. Detailed review strongly advised before partnering.
Each creator is evaluated by 7 independent agents. Each scores 0–100 internally, then a weighted average produces the final 1–100 CreatorScore.
Evaluates content for hate speech, explicit/NSFW material, violence, extremist ideology, and profanity. This is the most heavily weighted agent because a single brand safety incident can cause lasting reputational damage.
A single offensive post can go viral and damage both the creator's and brand's reputation overnight.
Detects fake followers, bot commenters, spam engagement, engagement pods, and artificially inflated metrics. Ensures the audience is real people, not purchased bots.
Fake engagement means your ad spend reaches bots, not real customers. Authenticity directly impacts ROI.
Assesses overall brand association risk by combining content safety signals with FTC ad disclosure compliance. Proper disclosure protects both brands and creators legally.
Brands need partners who protect the relationship. Missing FTC disclosures expose brands to legal liability, and past controversies signal future risk.
Measures the quality and engagement of a creator's audience — combining community health signals with engagement performance metrics, normalized by tier and platform.
High-quality audiences mean real reach and genuine influence. Poor audience quality means wasted ad spend.
Measures whether a creator's voice is coherent and consistent enough for brands to predict what they'll say next. A universal, brand-agnostic signal: it rewards creators with a clear voice (even one that ranges across topics) and flags genuine drift — escalating controversy, sudden tonal volatility, or a 90-day trend turning negative.
Brands need predictable partners — but predictable doesn't mean monotone. Voice Stability distinguishes a creator whose tone wanders by topic (fine) from one whose voice is genuinely drifting toward conflict or burnout (real risk). The earlier variance-only approach unfairly penalized the former.
Evaluates creator conduct and consistency — how they engage with their community and whether they maintain a reliable posting cadence that brands can depend on.
A creator's community reflects their brand. Hostile audiences, disengaged creators, or inconsistent posting lead to negative brand association and unpredictable campaign delivery.
Projects likely campaign return by combining engagement performance, content consistency, and growth trajectory. Helps brands estimate the value of a partnership before committing budget.
High engagement and consistent growth mean the audience actively cares about the content. Declining engagement signals a creator whose influence — and your campaign ROI — is fading.
CreatorScore combines multiple layers of AI analysis, each specialized for a different type of content and risk detection.
Advanced AI language models analyze every caption, comment, and transcript for hate speech, sentiment, toxicity, and spam.
Purpose-built vision models scan thumbnails and video frames for explicit, violent, or inappropriate visual content.
Optical character recognition reads text overlaid on images and videos to catch hidden messages not in the caption or audio.
Audio transcription analyzes what creators actually say in videos, not just what they write in captions.
A large language model reviews all findings in context — understanding niche, tone, and intent to minimize false positives.
Every score comes with a transparent breakdown so brands and legal teams can see exactly which factors raised or lowered it.
A score is only as good as the data behind it. When a creator's profile is too new, too quiet, or too private to evaluate fairly, we say so out loud — instead of fabricating a number that would mislead brands or unfairly penalize creators.
We require at least 5 public posts per platform within our content window to produce a per-platform score. Below that threshold, we run a brand-safety risk scan on whatever content exists — so brands still get risk visibility — but we don't publish a FICO-style score for the platform, because the data isn't sufficient to be fair to either side.
Five is the floor where engagement, sentiment, and content-risk signals stabilize enough to produce a number that's defensible.
Our default scoring window looks back 180 days. For creators who post less frequently, we automatically widen the window to 720 days (two years) on the first attempt before concluding there isn't enough content to evaluate.
Dormant creators with older-but-real content still get scored. The window only widens once per scoring pass — we don't retry indefinitely or burn API tokens on creators we've already determined are inactive.
If every platform on a creator's profile is below the minimum after widening the window, their public profile shows a clear “Not Enough Data” banner — with platform-specific reasons (“Only 4 posts available on LinkedIn,” “No posts available on Facebook”) — instead of a misleading score.
Brands see honest “we don't know yet” instead of a low number that looks like a verdict. Creators see exactly what they need to do to become scoreable. We re-check every 24 hours and rescore automatically on the next post.
Even when we can't publish a score, we still scan every piece of available content for hate speech, NSFW material, violence, and other brand-safety risks. A creator with a single inflammatory post is still flagged for it — risk visibility doesn't depend on having enough data for a number.
Brand safety is a watchdog, not just a score input. Insufficient data for ranking is not insufficient data for screening.
A nano-influencer with 5,000 followers and a celebrity with 10 million followers operate in completely different realities. CreatorScore uses tier-based normalization so every creator is scored against benchmarks appropriate for their size and platform.
Engagement Rate
A 2% rate is excellent for a mega-influencer but below average for a nano-creator. Both are scored fairly.
Reply Rate
Mega-influencers can't reply to millions of comments. We weight reply quality over quantity at scale.
Niche Context
Comedy creators aren't penalized for casual language the same way family creators would be.
Get a full brand safety score breakdown for any creator in minutes, not days.