Social LiveDocumentationPLATFORM DOCS

Creator Program -Structure Design

Proposed layered monetization structure -tiers, agencies, quotas, PK economy, payout rails.

Last updated Thu Jul 16 2026 00:00:00 GMT+0000 (Coordinated Universal Time)

TL;DR -This is a proposed, not-yet-built design for how Social Live could reward its live hosts beyond basic gifting: ranks that raise your pay, schedule bonuses, competitive battle prizes, and talent agencies. Every reward is paid in diamonds through the existing money plumbing, and the whole thing is capped so it can never cost more margin than the platform can afford.

Who this is for -Executive & Business: read the first three sections for the strategy and the cost ceiling. Product & Design: the layer overview and "what we do differently." Engineering: everything, including the ledger rules and worst-case math.

Status: PROPOSED DESIGN -not implemented. Nothing in this section exists in the codebase yet. It is the target structure for Social Live's creator monetization program, modeled on the industry-standard layered economy used by TikTok Live, Bigo, Poppo, Tango, and Chamet -adapted to our existing two-currency economy and its no-loss invariants. When a layer ships, its constants move into economy.constants.ts and its page here becomes normative.

What this is, in plain terms

Today Social Live already has a working economy: viewers buy coins (think of them as arcade tokens), spend those coins on gifts for hosts, hosts receive diamonds (the tickets a performer wins), and diamonds cash out to real money. That loop works. What it doesn't do is give a host any reason to show up on a schedule, get better, or compete.

The Creator Program is the layer that adds those reasons. It borrows the proven playbook from the big live-streaming platforms and stacks four extra systems on top of the coins-and-diamonds core:

  • Host tiers -a monthly rank (Rising → Legend) that raises how much of each gift the host keeps.
  • Quotas & bonuses -stream enough hours on enough days and earn a percentage bonus on top of what you already made.
  • PK battle economy -head-to-head gift battles between two hosts, with a prize for the winner and a season-long ladder.
  • Agencies -talent organizations that recruit and coach hosts, earning a platform-paid rebate and, optionally, a capped cut of their hosts' earnings.

Two promises hold the whole thing together. First, every reward is paid in diamonds, never in coins or cash directly -so it rides the same settlement, withdrawal, refund, and fraud machinery that already exists. Second, the platform funds these incentives out of its own margin, never out of another user's balance, and the total is capped by design.

Why a layered structure

The core economy (coins → gifts → diamonds → cash) already works. What it lacks is everything that makes hosts stay and stream on a schedule: progression, guaranteed income paths, competitive formats, and a recruiting channel. Industry platforms solve this with four layers stacked on top of the virtual-currency core.

The diagram below shows the core economy at the bottom (already shipped) and the four proposed layers stacked above it, each building on the one below.

Rendering diagram…

Each layer has its own page. The table maps them to the real-world platforms that inspired each one.

LayerPageIndustry analogue
Host tiersHost TiersBigo "official host" ranks, Poppo host levels
Quotas & bonusesQuotas & Bonuses30h/15-day valid-hour quotas
PK economyPK Battle EconomyTikTok/Bigo PK win pools & ladders
AgenciesAgenciesMCN / talent-agency programs
Payout railsPayout RailsPayoneer / bank / weekly clearing
Schema & rolloutData Model & Rollout-

Design principles (non-negotiable)

These are the rules every layer is designed within. If you read nothing else technical, read these -they are why the program can't blow up the economy. Every layer below inherits the core economy's rules.

  1. The no-loss invariants are untouchable. Bonus coins never earn diamonds; conversion stays one-way lossy; self-gifting stays blocked. No program layer may introduce a coin→diamond amplification loop.
  2. All program payouts are diamonds, never coins or fiat directly. Tier multipliers, quota bonuses, PK rewards, and agency rebates all credit diamonds (pending, 72h hold like everything else) so they flow through the existing settlement, withdrawal, refund-clawback, and fraud machinery for free.
  3. Every movement is a ledger row with an idempotency key, in the same $transaction as the balance change -program payouts get new LedgerTransaction types (QUOTA_BONUS, PK_BONUS, AGENCY_REBATE, AGENCY_SHARE).
  4. Rate snapshots. Any multiplier or split in effect at credit time is written into the ledger row's metadata, so changing program constants never rewrites historical financials -same pattern gifting already uses.
  5. Platform-funded, margin-capped. Program incentives come out of the platform's margin, never out of another user's balance. The total is capped by design (worst-case math below).
  6. Paid-coin gross only. Tier progress, quota milestones, PK scores, and agency rebates count only paid-coin gift value. Bonus-coin spend is invisible to the program layer -otherwise promos become farmable.

Worst-case payout math

Here's the promise that the program can never overspend, shown as arithmetic. The baseline is that a host keeps 25% of the money viewers spend on their gifts. Even if every incentive layer stacks up at its maximum for one host in one month, the total payout still lands under half of gross.

Baseline: host payout is 25% of gross gift value (2 coins = 1 diamond, 1 diamond = $0.005). The stacked program maxes out at:

Baseline payout                      25.0% of paid gross
Tier multiplier ceiling (×1.40)     +10.0%   (25% × 0.40)
Quota milestone bonus (+10% of
  earned diamonds, quota met)        +3.5%   (35% × 0.10)
PK winner bonus (5% of battle
  gross, assume 100% of gifts
  arrive inside battles -extreme)   +5.0%
Agency rebate ceiling (10% of
  roster diamond earnings)           +3.9%   (38.5% × 0.10)
────────────────────────────────────────────
Absolute worst case                 ≈47.4% of paid gross

After the 30% store fee the platform's worst-case net is ≈ 22.6% of gross -and the worst case requires a Legend-tier host who meets quota, receives every gift inside a winning PK battle, and belongs to a top-band agency. Realistic blended payout is 28–35%, in line with the industry's 40–70% platform-retention range. Every percentage above is a constant in one place (creatorProgramConfig, planned alongside economyConfig), so the margin envelope is tunable without touching flow code.

What we deliberately do differently

Most of the industry's problems with creator programs come from a few design choices we're choosing not to copy. Each item below trades a flashier promise for a system that's harder to defraud and cheaper to run.

  • No salary contracts at launch. Industry "official host" base salaries are a large fixed liability and the #1 fraud magnet. Our quota layer pays percentage bonuses on real earned diamonds instead -a host who earns nothing costs nothing. Salaries can be layered on later as agency-funded (not platform-funded) commitments.
  • Missing quota never slashes earned gift revenue. Industry platforms cut a host's bonus/salary by 50% for missing an hour. We simply don't pay the milestone bonus. Gift earnings are the host's, full stop.
  • Agency cuts are on-platform and capped at 20%. Industry agencies take 15–40% off-platform, invisibly. We enforce the split in the ledger so hosts see exactly what their agency takes, and cap it.
  • No fake-traffic tolerance. Agency "traffic boost" services are grounds for termination; detection is specced in each page's anti-abuse section.