GMV Max dashboards can make a campaign look profitable before refunds, reversals and creator costs have settled. The early sales number is useful for traffic response. It is a weak input for a budget increase.
Use a maturity rule. Count only orders old enough to support the decision. Build a conservative spending cap from their contribution. Increase spend in a small step and wait for the next complete cohort before approving another step.
This model is for operators who can join ad spend to order cohorts. It does not prove incremental lift. It prevents immature orders from receiving the same weight as settled orders.
Define the settlement window
Choose the number of days L that an order cohort must age before its contribution can enter the decision.
The window should cover:
- the store's refund and cancellation pattern;
- delayed creator commission or affiliate adjustments;
- the time needed to reconcile platform and payment fees;
- any known settlement corrections that change contribution.
For a store with late refunds, 14 days may be too short. Use the observed lag distribution. A common operating rule is the age at which at least 90% of refunds for comparable cohorts have arrived.
The budget rule treats younger orders as UNMATURED. It does not set their contribution to zero forever. It keeps them outside the current spending decision.
Calculate the mature contribution floor
For every cohort that has reached age L, calculate:
mature orders = orders with refund treatment complete contribution per matured order = ( net sales after seller discount and mature refunds - cost of goods sold - fulfilment - platform and payment fees - creator and affiliate commission ) / mature orders pre-ad contribution floor = mature orders x contribution per matured order Ad spend stays outside contribution until this point. A fixed creative or production cost should also stay separate so the same cost is not deducted twice.
The word "floor" matters. The model assigns no contribution to immature orders. That can understate eventual contribution. The tradeoff is deliberate: a budget step must survive without relying on orders that are still changing.
Set a spending cap
Define a minimum contribution coverage ratio:
coverage ratio = pre-ad contribution floor / ad spend spending cap = pre-ad contribution floor / minimum coverage ratio If the team requires $1.35 of pre-ad contribution for every $1.00 of ad spend, set the minimum coverage ratio to 1.35.
This is not a universal target. A low-margin product can require a higher coverage ratio. A product with verified repeat purchases may use a different measure, but only when repeat contribution is evidenced and included consistently.
Keep the minimum coverage ratio fixed for the test period. Changing it after seeing the result turns the cap into a post-hoc explanation.
Increase in small, delayed steps
Use this sequence:
- Build a cohort that has reached age
L. - Calculate the pre-ad contribution floor and current coverage ratio.
- Calculate
headroom = spending cap - current spend. - Cap one increase at both
10% of current spendand the available headroom. - Apply the smaller number.
- Wait until a new cohort reaches age
L. - Recalculate before approving another increase.
The formula is:
headroom = max(0, spending cap - current spend) proposed step = min(0.10 x current spend, headroom) next spend = current spend + proposed step Stop increasing when the contribution floor falls below the minimum coverage requirement. Do not count a campaign's immature revenue increase as evidence that the new spend has passed the rule.
Worked example
The numbers below are synthetic and use a 14-day maturity window.
| Input | Current cohort |
|---|---|
| Attributed orders | 1,000 |
| Mature orders after 14 days | 820 |
| Mature maturity rate | 82.0% |
| Net sales after mature refunds | $45,920 |
| COGS | $20,500 |
| Fulfilment | $4,920 |
| Platform and payment fees | $2,296 |
| Creator commission | $2,296 |
| Pre-ad contribution floor | $15,908 |
| Current ad spend | $9,000 |
| Contribution after ads | $6,908 |
| Current coverage ratio | 1.768 |
| Minimum coverage ratio | 1.350 |
| Spending cap | $11,783.70 |
The cap has $2,783.70 of headroom. A 10% step is $900, which is smaller than the headroom. The proposed test spend is $9,900.
Do not approve the next increase from the same cohort. Wait for the next 14-day cohort and repeat the calculation.
Reproducible SQLite query
The query below groups orders by acquisition day, joins cohort-level ad spend and returns only cohorts old enough for the decision.
WITH mature_cohorts AS ( SELECT o.cohort_day, COUNT(*) AS attributed_orders, SUM(CASE WHEN o.order_status = 'mature' THEN 1 ELSE 0 END) AS mature_orders, SUM( CASE WHEN o.order_status = 'mature' THEN o.net_sales - o.refunded_amount - o.cogs - o.fulfilment - o.platform_fee - o.creator_commission ELSE 0 END ) AS pre_ad_contribution_floor FROM orders AS o WHERE julianday(:decision_day) - julianday(o.cohort_day) >= :maturity_days GROUP BY o.cohort_day ), cohort_economics AS ( SELECT m.cohort_day, m.attributed_orders, m.mature_orders, m.pre_ad_contribution_floor, s.spend, m.pre_ad_contribution_floor / NULLIF(s.spend, 0) AS coverage_ratio FROM mature_cohorts AS m JOIN cohort_spend AS s USING (cohort_day) ) SELECT SUM(pre_ad_contribution_floor) AS contribution_floor, SUM(spend) AS ad_spend, SUM(pre_ad_contribution_floor) / NULLIF(SUM(spend), 0) AS coverage_ratio, SUM(pre_ad_contribution_floor) / :minimum_coverage_ratio AS spending_cap FROM cohort_economics; This query returns a portfolio total for all eligible cohorts in the chosen window. Run it by store, currency and market. Do not combine currencies into one spending cap.
A reproducibility check
The following SQLite example creates two orders and two spend rows. One cohort is mature and one is younger than the decision window.
CREATE TABLE orders ( order_id TEXT PRIMARY KEY, cohort_day TEXT NOT NULL, order_status TEXT NOT NULL, net_sales REAL NOT NULL, refunded_amount REAL NOT NULL, cogs REAL NOT NULL, fulfilment REAL NOT NULL, platform_fee REAL NOT NULL, creator_commission REAL NOT NULL ); CREATE TABLE cohort_spend ( cohort_day TEXT PRIMARY KEY, spend REAL NOT NULL ); INSERT INTO orders VALUES ('mature-1', '2026-11-01', 'mature', 100.00, 5.00, 30.00, 8.00, 4.00, 3.00), ('young-1', '2026-11-14', 'unmatured', 200.00, 0.00, 60.00, 16.00, 8.00, 6.00); INSERT INTO cohort_spend VALUES ('2026-11-01', 40.00), ('2026-11-14', 80.00); At a decision day of 2026-11-20 and a 14-day maturity window, the mature cohort contributes $100 - $5 - $30 - $8 - $4 - $3 = $50. Its coverage ratio is $50 / $40 = 1.25. A 1.35 minimum produces a spending cap of $37.04, below the current $40, so the rule blocks a larger step.
The younger cohort remains outside the decision. Its revenue cannot be used to justify the increase.
Failure modes
Refund policy changed
Run separate maturity curves before and after the change. A store-wide curve can hide a new policy that increases late refunds.
Spend and orders use different dates
Join spend to the same attribution day as the order cohort. A weekly spend total can make several daily cohorts look more profitable than they are.
Creator cost arrives late
Record creator commission as an event with its own settlement date. If the platform export omits unpaid commission, increase the maturity window or hold the budget instead of filling the gap with an estimate presented as actual.
Multiple currencies
Use one currency and one exchange-rate policy per analysis. A total that mixes currencies cannot produce a valid spending cap.
Low mature order count
A high contribution per order from nine mature orders is not a reliable basis for scaling. Require a minimum mature order count or extend the observation window.
Operating checklist
- Freeze the maturity window and minimum coverage ratio before the test.
- Count only complete cohorts.
- Calculate contribution after refunds, fulfilment, fees and creator costs.
- Keep ad spend separate and assign it to the same cohort.
- Cap each step at 10% or the available headroom, whichever is lower.
- Wait for the next complete cohort before another increase.
- Record the date, rule version, inputs, spend cap and decision.
The rule is conservative. That is the point. It makes the budget decision reproducible while the newest orders are still changing.
For the complete planning framework, see the TikTok Shop GMV Max guide. The free break-even ROAS calculator can test the contribution threshold before applying this cohort rule.
The example uses synthetic data. It does not prove incremental lift or guarantee ad performance. Maintained by Starway Global, publisher of the linked GMV Max guide and break-even ROAS calculator.

