RaidzATTENTION, WITH PROOF.
Open app

Marketplace model

Raidz connects project demand with creators who have relevant reach, skills, and verified delivery history. The marketplace is designed around explicit work orders and evidence—not rewards for likes, reposts, follows, views, or coordinated buying.

Marketplace surfaces

Radar: open missions

Radar lists open creator work orders called Signals. Each Signal should expose:

  • project and ownership-verification status;
  • objective, channel, category, and original deliverable;
  • gross budget, creator pool, per-creator reward, and currency;
  • minimum Raidz Score and other eligibility conditions;
  • number of positions and remaining capacity;
  • start, deadline, and retention timestamps;
  • disclosure, content, proof, refund, and jurisdiction rules;
  • current lifecycle state.

The current UI uses obvious demo fixtures. The live API deliberately returns an empty mission array until persistent, funded campaigns exist.

Market: direct services

Creators can define a service type, price, lead time, supported jurisdictions, and availability. Examples include original content, research, reviews, memes, video, Telegram features, AMAs, Spaces, community support, and product testing.

The database schema exists. Authenticated listing management, booking, service escrow, change requests, cancellation, delivery, disputes, and reviews remain planned.

Performance: qualified outcomes

The later performance market links creator activity to predeclared outcomes. A project must define the activation or onchain event before launch. Attribution events carry consent, source category, pseudonymous subject reference, deterministic deduplication key, and expiry.

Performance work must not collapse into last-click claims or compensate artificial metrics. It remains gated until the measurement system, fraud review, and causal limitations are clear to buyers and creators.

Campaign economics

For gross campaign funding G:

marketplace fee = floor(G × 800 / 10,000)
creator pool    = G − marketplace fee

For each fixed creator reward R:

immediate release = floor(R × 7,000 / 10,000)
retained release  = R − immediate release

The prototype therefore uses an 8% marketplace fee and a 70/30 creator release. Integer division rounds the first component down; the remainder stays in the creator pool or retained component so accounting balances exactly.

The factory rejects campaigns whose fixedReward × creatorLimit exceeds the creator pool. A creator may claim at most one position in a prototype campaign.

Campaign and claim state

The domain campaign lifecycle and onchain claim lifecycle are related but distinct:

  • The campaign describes project review, funding, live operation, pause, close, expiry, refund, and settlement.
  • A claim describes one creator position: claimed, submitted, verified/rejected, retention, and settled.

Closing or refunding a campaign must never orphan an open creator obligation. See Campaign lifecycle and Proof and settlement.

Eligibility and policy preflight

Every compensated campaign is evaluated before funding. The implemented baseline blocks:

  • language promising guaranteed profit, guaranteed price movement, or risk-free returns;
  • compensated likes, reposts, replies, views, or follows;
  • coordinated buying, artificial volume, fake scarcity, or fake metrics;
  • compensated X crypto campaigns in jurisdictions covered by the current versioned restriction.

Rules are versioned with source, effective dates, channel, service, jurisdiction, compensation condition, decision, and reason code. This is a campaign control, not legal advice or a complete jurisdiction engine.

Reputation on both sides

Creators accumulate verified delivery, retention, conversion, reliability, and buyer-review history. Projects accumulate payment, brief quality, change-request, dispute, and creator-review history. Reviews are campaign-bound, one per reviewer, and can target either a creator or a project.

Neither token ownership nor payment size directly increases Raidz Score. A high score signals evidence under a particular model version; it is not a guarantee of future performance.