ATTENTION, WITH PROOF.Raidz Score and Rank
Raidz Score V0 is a deterministic, explainable score from 0 to 1000. It summarizes account maturity, audience quality, real reach, content quality, topic fit, and verified marketplace performance.
The current model version is raidz-score-v0. It is implemented and tested, but it has not yet been validated against a substantial set of real Raidz campaigns.
Initial component weights
All component inputs are normalized to 0–100 before weighting.
| Component | Weight | Meaning |
|---|---|---|
| Account maturity | 12% | Age, continuity, and history of the connected account |
| Audience quality | 15% | Estimated authentic, relevant audience quality |
| Real reach | 15% | Typical observed distribution rather than headline followers |
| Content quality | 15% | Originality, clarity, usefulness, and format quality |
| Topic relevance | 10% | Fit with the campaign narrative or audience |
| Social scale | 8% | Account/community scale, deliberately below quality and reach |
| Campaign performance | 10% | Verified Raidz delivery outcomes |
| Conversion performance | 8% | Qualified downstream outcomes under consented attribution |
| Reliability | 4% | Deadlines, retention, and operational dependability |
| Marketplace ratings | 2% | Campaign-bound buyer feedback |
| Verification signals | 1% | Additional evidence and integrity signals |
The weighted onboarding value is:
onboarding = Σ(clamp(component, 0, 100) × component_weight)
Marketplace performance transition
As a creator completes verified campaigns, native Raidz performance progressively replaces external onboarding evidence.
| Completed campaigns | Performance share |
|---|---|
| 0–4 | 0% |
| 5–19 | 28% |
| 20–49 | 48% |
| 50+ | 70% |
The performance subtotal is:
performance = campaignPerformance × 35%
+ conversionPerformance × 35%
+ reliability × 20%
+ marketplaceRatings × 10%
raw = onboarding × (1 − performanceShare)
+ performance × performanceShare
Raidz Score = round(raw × 10)
This prevents an early creator from being penalized for having no platform history, while ensuring that real marketplace behavior becomes the moat over time.
Tiers
| Score | Tier |
|---|---|
| 0–249 | SCOUT |
| 250–449 | OPERATOR |
| 450–649 | SIGNAL |
| 650–799 | PRIME |
| 800–899 | APEX |
| 900–1000 | ELITE |
Tier boundaries are deterministic and belong to the score model version. A project may use a minimum score as one eligibility input, but should also consider narrative, language, region, service, and demonstrated fit.
Evidence contract
Every one of the eleven components must include:
sourceTimestamp— when the evidence was observed;confidence— a0–1value;explanationReason— a human-readable basis;- model version — attached to the scored output.
The preview endpoint rejects missing evidence instead of creating an unexplained score. Provider data collection, normalization, score snapshots, change explanations, appeals, and production recalculation schedules remain integration work.
What the score is not
Raidz Score is not creditworthiness, investment advice, identity proof by itself, a guarantee of reach, or a purchasable status. Wallet balance and $RAIDZ holdings are not score components. A model version should never be compared silently with another version; UIs and exports must show the version and freshness.
Improvement and challenge flow
Creators should be able to improve through sustained original work, better campaign fit, reliable completion, preserved deliverables, and qualified outcomes—not by manufacturing engagement. Production requires a way to inspect components, flag inaccurate source data, request recalculation, and distinguish an appeal from an economic dispute.