1. Mathematical Operational Model
The Waterfall executes a lossless one-dimensional cascade decomposition from gross revenue to net profit. The settlement reconciliation engine strictly enforces the following energy conservation identity:
Kept (Net Profit) = Earned − Taken − Spent − Lost
Every dollar that enters the system exits through exactly one of these five terminal states. No leakage, no double-counting. The waterfall is a visual proof of this invariant.
2. Five-Bucket Vector Anatomy
The system atomically classifies TikTok’s 140+ granular fee line items into five discrete settlement-decoupled buckets via a config-driven UDF (fee_category_mapping):
Earned — Net Positive Asset Inflow
V_earned = [ subtotal_before_discount, cod_service_fee, platform_discount_subsidy ]
- Theoretical maximum revenue ceiling before platform extraction begins
- Includes seller-facing subsidy credits: platform coupon reimbursements, COD collection fees passed to buyer
Taken — Platform Pre-Payout Extraction (All TikTok Deductions Before Settlement)
V_taken = [ platform_commission, referral_fee, transaction_fee, refund_admin_fee, affiliate_commission, actual_shipping, fbt_fulfillment, return_shipping, shipping_subsidy_clawback ]
- All deductions TikTok executes before funds reach the seller’s bank account
- Includes platform infrastructure fees (commission, referral, transaction), logistics (FBT, shipping net of buyer-paid), and affiliate payouts
- Optimization surface: category reclassification (commission), fulfillment model switch (logistics), creator commission negotiation (affiliate)
Spent — Post-Payout Seller-Controlled Expenditure
V_spent = [ cogs, smart_promotion, gmv_max_ad_fee, tap_ads_commission ]
- Expenses paid from the seller’s bank account after settlement arrives
- COGS: physical product cost per unit — structural and per-SKU
- Ad spend: discretionary traffic acquisition — controllable on a daily basis
Lost — Reverse Flow Bad Debt & Uncontrollable Write-offs
V_lost = [ customer_refund, platform_discount_refund, fbt_shrinkage, exchange_rate_loss ]
- Revenue counted in Earned but subsequently clawed back or evaporated
- Refunds are the dominant driver; shrinkage and FX loss are secondary
Kept — Net Terminal Value to Bank
V_kept = V_earned − V_taken − V_spent − V_lost
- The residual that hits the seller’s bank account
- This is the only number that matters for business viability
3. Telemetry Deflection Thresholds
When any deduction bucket ratio breaches the following critical thresholds, the system flags a margin bleed event and triggers high-severity highlighting on the dashboard:
Any single bucket exceeding its threshold indicates a structural margin compression that will compound if unaddressed.
| Metric | Critical Threshold | Root Cause Audit Path |
|---|---|---|
| Taken / Earned | > 14.0% | Decompose into sub-vectors: commission (category mismatch?), logistics (dimensional weight?), affiliate (creator matrix overload?). Each sub-component has its own diagnostic. |
| Spent / Earned | > 15.0% | Ad spend efficiency declining or COGS creep. Audit per-campaign marginal ROAS and supplier pricing trajectory. |
| Lost / Earned | > 5.0% | Batch quality defect causing scaled returns, or geo-specific regulatory surcharge escalation. Isolate by SKU refund rate and shipping destination. |
4. Drilldown
Every waterfall bar is a click target. Click any bucket to see the SKUs contributing the most to that cost, which creators or content drove it, and how it compares to the prior period.
The drilldown answers: “This bucket took $12.8K — but from which SKUs, driven by which creators, and is it accelerating?“
5. Weekly Ledger Inspection Protocol
Execution cadence: every Monday. Total time budget: ≤ 180 seconds.
Step 1 — Lock the temporal axis
Select the 7D window covering the last complete week (Mon–Sun). This eliminates partial-week noise and aligns with TikTok’s settlement cycle boundaries.
Step 2 — Compute deviation slopes
For each bucket, compare week-over-week growth rate against revenue growth rate:
Margin Compression Signal = ΔCost_bucket / ΔRevenue > 1.0
If any bucket’s cost growth slope exceeds revenue growth slope, margin is compressing in that dimension — even if absolute profit increased.
Step 3 — Reverse-flow reconciliation
For the Lost bucket specifically: identify platform-coupon-induced write-offs (where TikTok issued a buyer coupon but the refund is charged to the seller). These require second-pass reconciliation against the platform subsidy credits in Earned.
Step 4 — Document and act
If a threshold breach or compression signal is detected:
- Flag the bucket
- Drill down to isolating SKU/creator
- Determine if the cause is structural (category rate, product weight) or tactical (campaign overspend, quality defect)
- Structural → escalate to strategy review. Tactical → fix this week.