How we audit our suppliers

Why we publish the process

Almost every page on this site that says "real, active accounts" or "30-day refill if engagement drops" is making a claim that depends on something most SMM services keep behind closed doors: how we actually audit our supply pool. This page is that audit, written down. The five-component scoring rubric, the weekly review cycle, the random-sample protocol, the demotion thresholds — all of it.

The companion page at /methodology/real-account-definition defines the BAR every account in our supply pool must clear (the conjunctive five-point criterion: profile completeness, account age, follow ratio, recent engagement, cross-account interaction). This page documents the PROCESS — how we measure compliance against that bar at supplier scale, what triggers demotion, and what the routing layer does once a supplier fails.

The five-component scoring rubric

Each supplier in our routing pool receives a weekly composite score from 0–100 across these five components. Weights add to 100%; thresholds for probation and immediate demotion are published per-component below. A supplier's composite score determines what fraction of the coming week's orders they receive.

The weekly review cycle

The audit cycle runs every Monday-through-Friday and is fully scripted in our internal pipeline — supplier scores, probation decisions, and demotion actions are applied automatically based on the rubric thresholds above, without weekly negotiation or operator override. Operations reviews exception cases (small-supplier buckets, missing data, scoring ties) on Friday before the next week's routing weights take effect.

The random-sample audit

The QA sample pass rate (component 3 above, weighted at 25%) is our highest-fidelity quality signal — drop rates lag by 30 days, but a sample audit catches the same pattern in 48 hours. The protocol:

1. 01Sampling rate: 1 in 50 orders. Higher rate (1 in 25) on suppliers in active probation. Lower rate (1 in 100) on suppliers in the top decile of scores for 4+ consecutive review cycles, freeing operations capacity for the suppliers under closer scrutiny.
2. 02Per-order sample: 10-20 of the delivered accounts, drawn uniformly from the order's recipient list. We don't oversample the first or last accounts (a known gaming pattern where bad suppliers ship clean accounts at the start to defeat sequential sampling).
3. 03Per-account scoring: each sampled account is scored against the published five-point real-account criterion. Each point is independent — an account fails the audit if it fails any one point, even if the other four pass.
4. 04Pass-rate computation: percent of sampled accounts that pass all five points. Below 80% on a single audit triggers an immediate operations escalation; below 90% on two consecutive weeks triggers probation; below 95% on four consecutive weeks demotes the component-3 weight to its threshold floor.
5. 05Audit operator rotation: no single operator audits the same supplier twice in a row. This eliminates the risk of a relationship-based scoring drift between an operator and a supplier they've evaluated repeatedly.

How the router selects a supplier

Once a supplier's weekly score is set, the router uses these rules to assign individual orders. The rules apply in order — earlier rules override later ones when they conflict.

• Pin no single supplier above 40% of weekly orders — even our best-scoring supplier maxes at 40% of routing share, so a quality regression at any one source affects no more than 40% of orders before the next weekly review catches it.
• Match supplier to package tier — Premium-tier orders (real-account criterion strict) only route to suppliers whose QA sample pass rate is ≥ 95%; Standard-tier orders route across the broader pool.
• Scope by platform — TikTok orders route to TikTok-specialised suppliers, not Instagram suppliers serving TikTok deliveries. Cross-platform supplier capacity is a known quality-degrading pattern that the per-platform scoping eliminates.
• Scope by geography — orders flagged with a geographic preference (e.g. "USA accounts") route only to suppliers whose pool has verified geographic distribution matching the preference.
• Backoff after failure — if any supplier returns a delivery error or partial fulfilment within 48 hours of a placement, their routing weight is automatically halved for the next 7 days regardless of their weekly score. Two error events in 14 days = probation.