Max connects every ad dollar to actual profit after COGS, shipping, returns and fees, then names which products earn and which lose money on every sale.
Ad platforms optimise to revenue because revenue is the number they can see. COGS, shipping, payment fees and returns all land after the pixel fires, so the campaign that looks best in the dashboard is often the one funding the discount.
A campaign is scaled on week-two performance and the return rate lands in week six. By the time the refunds are booked, the budget decision is two cycles old and nobody re-runs the math on it.
A category average of 58% can carry a hero SKU at 71% and a bundle at 12%. Optimising against the average scales both, and the one that pays for the other stays invisible until inventory turns.
A 20% code is approved because last year's worked. Pull-forward, margin hit and the customers who would have paid full price are not in the model, because there is no model.
Each capability answers a specific question: what is the margin, which SKU earns, is this discount worth running, when does the customer pay back, and what does the rest of the workforce need to know.
Revenue less COGS, shipping, payment fees and returns, per order rather than per report, so margin is a fact about the basket instead of a quarterly average.
Example. Example. £84 order, £31 COGS, £6 ship, £2.40 fees: £44.60 contribution before ad cost.
Every SKU ranked by what it actually contributes, so the hero product and the one quietly subsidised by it stop being averaged together.
Example. Example. The best-selling variant ranks fourth on contribution once returns are netted.
A discount modelled before it runs and scored after it ends, including the pull-forward it borrows from the weeks on either side.
Example. Example. A 20% code that lifts units 40% can still cut contribution if half the buyers were coming anyway.
How long a cohort takes to pay back what it cost to acquire, in contribution rather than revenue, split by new and returning.
Example. Example. Blended payback of 2.1 months hides a new-customer figure closer to 4.
Reads Parker's de-biased attribution so margin is charged against real incremental revenue, and hands Sam the constraint that makes a scenario honest.
Example. Example. Sam tests a 30% Meta shift; Max prices it against contribution, not claimed ROAS.
Timings are the shape of the cycle. Exact figures depend on catalogue, return rate and payment mix.
Pulls the weekend's orders and nets refunds, shipping and payment fees into contribution per order.
Re-ranks the catalogue on contribution. Two SKUs that led on revenue drop out of the top ten.
Scores last week's promo against its pre-promo baseline and reports the pull-forward separately from the lift.
Flags a bundle whose contribution has gone negative since the shipping rate changed.
Prices Sam's proposed budget shift against margin rather than ROAS and returns the constraint.
Updates cohort payback as the week's second orders land, and hands Dex the figures for the recap.
Profit, not proxies
Max reads your orders, costs and returns, charges spend against Parker's incremental revenue, and tells you which products and which campaigns pay for themselves.
Looking for something specific? Parker de-biases the revenue Max charges against →, Sam tests the moves Max prices → or Dana reconciles the orders underneath →.
