The Margin Stack30 DAYS FREE + Free Analytics Session $500

Problems we solve / The money / Where margin leaks

Problem 02 of 16 · The money

Where exactly is margin made, and where is it leaking?

Every store has a margin number, and almost every store reads it as one figure. But margin is made and lost in the breakdown: a product, a channel, a cohort, a shipping rule. Margin analysis means cutting the number until the line that loses money on every order has nowhere left to hide.

The short answer

Break contribution margin down by product, channel and cohort at the same time, then step each cut from revenue down through every cost line so you can see which cost did the damage. Blufire's Unit Economics section is that breakdown: a pivotable Profitability Cube, a CM1 Waterfall, and a queue of leaks ranked by the dollars you can recover.

5.0 on Google · 100+ businesses · $153M revenue influenced

Why it happens

Averages are built to hide the thing you are looking for.

A store-wide margin is a weighted average of every order. Strong lines lift it, weak lines drag it down, and the result looks like a healthy middle. A product that loses A$2 on every order sits comfortably inside a 38% average, because the hero products carry it.

Leaks also form at the joins between dimensions, which is why cutting by one dimension at a time rarely finds them. A product can be profitable overall and loss-making through one channel. A channel can be profitable overall and loss-making on first orders. The leak is in the cell, product × channel × customer type, not in the row or the column.

And most leaks are not in COGS. Landed COGS is usually known and stable. The costs that drift are the ones that move with the order after the product is picked: free shipping on a small basket, pick and pack on a low-price item, refunds on a size that runs small, a discount stacked on a bundle. That is why margin analysis has to go past CM1 to CM2 and beyond (the full step-down is on The Math).

What changes

The decision you walk away with.

TodayWith Blufire

One store-wide margin, checked when the accountant sends it.

Margin by product, channel and cohort, pivoted in one cube.

Leaks are found by accident, usually after a bad quarter.

Every leak sits in a queue, ranked by the dollars you can recover.

Free shipping and refunds disappear inside "operating costs".

The waterfall gives each its own line, between revenue and CM2.

A fix goes live and nobody checks whether it held.

The Recovery Tracker reads what the fix actually returned.

The maths

A 38% average with a loss-making line inside it.

A skincare store, one month, read at CM2 (after landed COGS, pick and pack, shipping and payment fees, before any ad spend).

Worked example / demonstrative numbers
Store-wide: A$300,000 revenue, A$114,000 CM238.0%
Hero serum: A$150,000 revenue, A$75,000 CM250.0%
Gift sets: A$90,000 revenue, A$36,000 CM240.0%
Travel minis: A$60,000 revenue, A$3,000 CM25.0%
Minis to repeat buyers through email: A$20,000 revenue, A$7,000 CM235.0%
Minis as first orders through Meta, shipped free: A$40,000 revenue, −A$4,000 CM2−10.0%
One A$20 Meta minis order: less A$8.00 landed COGS, A$11.00 free shipping, A$2.24 pick and pack, A$0.76 payment fee−A$2.00
The leak: 2,000 orders × −A$2.00, before a dollar of ad spend−A$4,000

The store-wide 38% looks fine. Cut by product, the minis look weak but positive at 5%. Only when minis are cut by channel and customer type does the leak appear: first orders shipped free through Meta lose A$2 each, and that is before the ads that produced them are counted.

The fix is not to drop the minis, which earn 35% as a repeat add-on. It is to stop shipping them free as a standalone first order, with a threshold or a bundle. The free-shipping threshold calculator prices that change, and the contribution margin calculator checks any line on its own.

How Blufire answers it

Pivot the margin, find the leak, prove the fix.

Section S2, Unit Economics, is one pivotable Profitability Cube: contribution margin by product, channel or cohort, and by any combination of them, so the cell where margin leaks is one pivot away. The CM1 Waterfall steps list revenue down through product discounts, refunds, COGS, fulfilment, payment fees, the free-shipping give and ad spend, so you can see which cost took the margin on the way down.

A cost feed that is not connected shows as an honest zero with a note, never a guessed figure, and lines costed above their selling price are surfaced, not smoothed away. Then the Ranked Leak Queue orders every leak by the dollars recoverable, and the Recovery Tracker keeps checking after the fix, because a leak you found and never re-checked is a leak you still have.

  • Profitability CubeContribution margin pivoted by product, channel or cohort, down to the cell where it leaks.
  • CM1 Waterfall & bridgeWhat each cost line took on the way from revenue to CM1, CM2 and CM3, and what moved any period's margin.
  • Ranked Leak QueueEvery margin leak ordered by the dollars recoverable, so the biggest fix comes first.
  • Recovery TrackerHolds each fix accountable by tracking what was actually recovered after it shipped.
See section S2, Unit Economics→
S2 Unit Economics · List revenue → CM1 → CM2
Blufire CM1 waterfall stepping gross list revenue down through product discount, refunds, COGS, fulfilment, payment fees, free shipping and ad spend to CM1, CM2 and CM3

Real product screen, shown on sample data.

Proof

The team behind the numbers.

RainCoA$5k to A$170ka month in eight months, with the finish carrying 85% of revenue de-riskedRead the case study →
“They took the time to understand our business and goals, and delivered a clear, customised strategy that actually worked.”
JPJodie PatchGoogle review
Google review
5.0on Google
100+businesses served
$153Mrevenue influenced
AFR Fast 100APAC Search Awards 2025 WinnerGlobal Search Awards 2025 Finalist
PanasonicRainCoCheapest LiquorKing CoolingAuto ComfortiHeat & CoolAACAEInsider Experience SportsInterosPeter JacksonLa TrobeToy World
How to do it

Margin analysis that actually finds leaks.

  • Cut two dimensions at once.Product alone and channel alone both averaged the example leak away. Product × channel × new or repeat found it. The contribution margin ratio should be read per cell, not per row.
  • Step down past CM1.Shipping, fees and refunds cause most leaks, and all of them sit below CM1. Include the variable costs that move with each order.
  • Rank by dollars, not percentages.A −40% line on A$2,000 of revenue matters less than a 5% line on A$60,000 that should be 30%.
  • Check returns by line.A product with a high return rate can look profitable at order time and leak weeks later. See what returns are really costing.
  • Re-check after the fix.Leaks come back when a carrier rate or supplier price changes. A fix is done when the margin is still there next month.
FAQ

Questions operators ask.

Margin analysis is breaking contribution margin down by product, channel, customer cohort and cost line to see where profit is made and where it is lost. The store-wide figure is only the starting point. The useful answers sit in the cells, such as one product through one channel to first-time buyers.
Step revenue down through every variable cost, from landed COGS through shipping, fees, refunds and discounts to ad spend, then cut the result by two dimensions at once. Rank the loss-making cells by dollars, not percentage. The biggest recoverable dollars go first, and each fix gets re-checked the following month.
Contribution margin. Gross margin stops at the product cost, and most leaks sit below it in free shipping, pick and pack, payment fees, refunds and stacked discounts. A line can show a healthy gross margin and still lose money on every order once those costs are counted.
Because an average lets strong lines carry weak ones. A hero product at 50% can hide a loss-making product, channel or shipping rule inside a respectable store-wide figure. Cut the margin by product and channel together and the lines dragging the average down become visible.

Ready to see what you are actually keeping?

Free for 30 days. Free to install.