Skip to content
Home / About / ugc creator fit scoring
UGC Agent

Scoring methodology

An explainable first-pass rubric, not a prediction of creator performance.

Reviewed by VicProduct-backed claimsCorrections: vic@ugcagent.app
01

Baseline public signals

The implemented rubric awards points for UGC language in the username, display name, or bio; an available email; portfolio language; known portfolio hosts; and structured niche information. The result includes the reasons that contributed to the score.

02

Default exclusions

The default micro-UGC workflow excludes private profiles, accounts outside the configured follower range, strongly non-English bios, missing emails, scores below the keep threshold, and profiles that fail the campaign-relevance gate. User preferences can add further constraints.

03

Email classification

Available addresses are classified as personal, creator-owned domain, likely agency, or unknown using domain patterns and profile context. Unknown addresses remain uncertain. Users can opt to exclude likely agency and management emails for direct-outreach searches.

04

Brand-collaboration mode

When the task is to find people who worked with a specific company, verified recent brand evidence replaces the normal UGC-signal score gate. The workflow carries public proof-post URLs, excludes the brand's own accounts, and uses a wider default follower band.

05

Limitations

The score cannot determine availability, rate, deliverable quality, legal compliance, audience authenticity, brand safety, or campaign performance. Public profiles change, classifications can be wrong, and every retained lead requires human review before outreach.

06

Human review and overrides

Operators should open the profile, inspect recent relevant content, verify the contact source, and compare the creator with the current brief. A reviewer can retain or remove a profile after that inspection; the useful record is the evidence and reason, not blind agreement with the numeric score.

07

Method changes

Material changes to scoring inputs, exclusion rules, email classification, or brand-evidence requirements should update this page and its reviewed date. Historical scores should be interpreted under the product version that produced them rather than silently compared across changed rubrics.

Next step

Turn the brief into a creator list.

Try a restricted sourcing run free, then upgrade only when you need more visible or monthly creator leads.