KNOW WHAT
THE NUMBER MEANS.

1. BUILD THE COST STACK.

Production and run costs are spread over good units. Dead stock is the share of good inventory that never sells. Financing uses simple interest on landed good-unit cost for the entered period, separate from the dead-stock adjustment. It is an estimate, not a loan schedule.

Per-order costs are spread across units per order. Return handling is per returned unit. Resellable returns receive an inventory recovery credit, not a cash payment. The model assumes full product refunds at the discounted charge and retention of original payment fees.

Quick entry separates postage, packaging, picking and packing, per-order app fees, customer support, and customer-paid shipping. Older saved drafts may still use a combined total; it stays visibly active until you choose the separate breakdown. Unknown inputs count as zero only within a visibly incomplete estimate. They remain missing in saved records and exports.

MONTHLY APPS ARE A SHARED STORE BILL.

Add your monthly website and software subscriptions, then divide by every item your whole store ships in that same month. A $150 bill shared across 300 items costs $0.50 per item. The number of different products you list is not the number to divide by.

Per-order app charges are entered with order costs. Fixed subscriptions and optional other overhead are shown in the amount left after monthly bills, separately from product contribution and the price ladder. Those fixed bills stay due even if sales slow down. Do not count the same bill in both places. Only the shared monthly bills you enter are counted.

2. SEPARATE PRICE FROM REVENUE.

Contribution is net retained product revenue less modeled inventory, delivery, return, and payment costs. It is shown before ads and overhead. Margin uses net product revenue as its denominator. The reveal compares both naive and modeled costs against the same sticker price.

Payment fees include the product charge before expected refunds and the shipping charge you enter. Add fees on tax, duties charged at checkout, payment mix, reshipments, or additional charges to your per-order cost estimate where applicable. V1 supports the listed two-decimal currencies and prices entered excluding tax; it does not convert currencies.

3. ROAS HAS A REVENUE BASIS.

This excludes tax and shipping from its numerator and applies expected refunds. An ad dashboard may use a different purchase-value basis. Do not compare the two without reconciling what each counts. Zero or negative contribution provides no positive budget for first-order acquisition.

4. TARGETS ARE NOT PROMISES.

The price ladder solves mathematical targets, then rounds executable prices upward and recalculates. Its names describe scenarios, not universal apparel benchmarks. A margin can be unreachable because payment and return rates consume too much revenue regardless of price.

Acquisition rungs require a confirmed homogeneous-basket forecast. Every modeled unit must earn the selected product's contribution. Storewide repeat purchases of different SKUs do not automatically qualify. 30-day orders include the first order in a customer's first 30 days; lifetime orders are your declared observation or forecast, not a claim that short history predicts lifetime behavior.

At a new price, allowable lost unit volume holds contribution and the rest of the entered assumptions constant. It does not forecast demand, revenue, or final net profit. Keep raising until the extra money no longer covers the lost sales. A higher price needs more value behind it.

5. KEEP PRICE TESTS HONEST.

Total contribution matters, but a price may generate more total contribution simply because it received more visitors. The log also shows contribution per eligible visitor and distinguishes observational results from comparable tests. Traffic, promotions, observation periods, basket composition, and return maturity all affect the comparison.

A comparable label records your declared protocol. The tool does not manufacture statistical confidence or prove a price caused the difference. The test decides, not a multiplier.