Ecommerce unit economics: four numbers, one quantity.

When an ecommerce brand stalls, everyone looks at the ad account. In our experience the account is the crime scene, not the criminal. The actual failure usually happened earlier, in the unit economics: a CAC computed on the wrong denominator, an LTV inflated by revenue instead of margin, a bidding target that was never derived from either. Paid media executes the math you give it. If the math is wrong, excellent execution just gets you to the wrong place faster.

The deeper problem is that these get treated as four separate metrics, usually owned by different people. Finance owns margin, growth owns CAC, the founder owns LTV, and somebody produces a MER number monthly because a board deck asks for one. They then disagree, and the disagreement gets resolved socially rather than arithmetically. They should not be able to disagree. CAC, LTV, payback and MER are four questions about a single quantity — the contribution margin one customer produces — and every one of them is answerable from the same waterfall.

• →Contribution margin — how much money a customer actually leaves behind after the costs of serving them. Everything else on this list is a question about this number.
• →CAC — what you paid to acquire the customer who produces it.
• →LTV — how much of it arrives, and over how long.
• →Payback — when enough of it has arrived to cover what you paid.
• →MER — what share of the whole pool you are handing to the platforms.

Start with the waterfall, because everything derives from it

Take the order economics above and write them out per order. This is the entire foundation, and it takes four lines:

PER ORDER
  average order value          85.00
  - cost of goods              32.00
  - shipping and fulfillment     8.00
  - payment fees                3.00
  = contribution margin        42.00   (49.4% of AOV)

WHAT THAT IMPLIES IMMEDIATELY
  breakeven CPA                42.00
  breakeven ROAS                2.02x
  target ROAS for 10 profit     2.66x  (allowable CAC 32.00)
  breakeven MER                 49.4%
  MER ceiling at 10 points
  of contribution after ads     39.4%

Every number in the second block is the first block rearranged. There is no additional information and no judgment in the derivation — which is why a team that has genuinely agreed on the first four lines cannot then hold four incompatible opinions about targets. Most teams have never written the first block down, so they argue about the second one forever.

CAC: the denominator problem

Customer acquisition cost sounds simple — spend divided by customers — and is quietly ruined by both terms. The numerator should be fully loaded: platform spend plus agency or team cost plus creative production. The denominator should be new customers only. Blending returning customers into the denominator is the most common flattery we see, and it does not shade the number, it transforms it:

ONE MONTH
  fully-loaded marketing spend   28,000
  total orders                    1,000
  of which new customers            400
  of which repeat                   600

CAC, TWO WAYS
  spend / all orders              28.00   <- looks 14.00 inside breakeven
  spend / new customers           70.00   <- actually 28.00 over breakeven

  breakeven CPA (from waterfall)  42.00

The same month reads as comfortably profitable acquisition or as losing twenty-eight pounds on every new customer, depending entirely on a denominator choice nobody discussed. And the error is self-reinforcing: as repeat share grows — which is what success looks like — the blended figure improves automatically while genuine acquisition efficiency can be deteriorating underneath it. A brand can watch its CAC "improve" for three consecutive quarters while the cost of acquiring a stranger doubles.

LTV: margin or it is fiction

Lifetime value computed on revenue is a fiction that flatters. The only LTV that can justify an acquisition cost is contribution margin: revenue minus product cost, shipping, payment fees, and returns, accumulated per cohort over time. Two further disciplines keep it honest. First, use realized cohort curves — what the 6- and 12-month value of past cohorts actually was — rather than projections from your best year. Second, segment: LTV by first product purchased and by acquisition channel routinely varies two-to-three-fold, which means a single blended LTV target is wrong for every campaign you run.

Payback window: the cash constraint

LTV justifies CAC eventually; cash flow decides whether you survive until eventually. This is the part that reads as accounting pedantry right up until it is the only thing that matters, so here is the arithmetic that shows a business with a respectable ratio walking into a wall. Take the seventy-pound real CAC above and a cohort that pays back over six months:

ONE COHORT OF 400 CUSTOMERS, CAC 70.00
  cash out at acquisition        28,000
  contribution returned:
    month 0 (first order)        16,800   (42.00 each)
    by month 3 cumulative        24,400
    by month 6 cumulative        31,200   <- payback here
    by month 12 cumulative       41,600

  LTV-to-CAC at 12 months          1.49x
  working capital held per
  customer until payback           28.00

NOW SCALE IT
  acquiring 400/month, flat      -11,200 per month of new float
  growing new customers 20%/mo    float compounds with the ramp
  six cohorts in flight at once   all pre-payback simultaneously

Every cohort is cash-positive by month six and the business is still starving, because at any moment six cohorts are in flight and only the oldest has paid for itself. Growth makes it worse rather than better: each month’s larger cohort demands more float than the one recovering. This is how brands die at ratios their board deck describes as healthy — not because the unit economics were wrong, but because nobody separated the question "is this profitable" from the question "can we fund it."

So set an explicit payback ceiling that matches your balance sheet rather than your ambition. Bootstrapped brands usually need first-order or three-month payback and should bid as if later revenue does not exist. Funded brands can stretch, and the stretch is a real strategic advantage — it lets them outbid you for the same customer while both of you are being rational. Let that ceiling, not the LTV ratio alone, cap how aggressively you bid.

MER: the blended guardrail

MER is total ad spend as a share of total revenue — a percentage where lower is better, so a 39% MER means thirty-nine pence of every pound goes to advertising. It also circulates as a revenue-over-spend multiple where higher is better, in which form 39% is 2.54x; the two are the same fact read in opposite directions, and the only real hazard is setting a target without saying which convention you mean. We use the percentage because it composes with a P&L, where every other line is also a share of revenue.

Its role here is structural rather than merely additional. Note what happened with CAC: it had to be computed on new customers only, or it flattered. MER is the opposite — it must be computed across everything, new and repeat, because it is the number that asks what share of the total contribution pool the platforms are taking. Applying the same denominator philosophy to both is precisely the error. CAC narrows deliberately; MER refuses to narrow at all, which is what makes it impossible to game by attribution.

And because contribution margin is roughly constant per order, breakeven MER simply equals your contribution margin percentage — 49.4% in the worked example. Spend that share of revenue on advertising and the last penny of contribution goes to the platforms. Subtract the contribution you intend to keep to get the ceiling. The mechanics of choosing which revenue figure feeds it, and why platform-reported revenue will inflate it, are in MER vs ROAS. Expect MER to worsen as you scale and prospecting share grows: that is arithmetic, not failure. Crossing the floor is failure.

Turning the numbers into targets

1. Compute contribution margin per order band, and set campaign-level ROAS targets from margin, not from habit — a 30%-margin catalog needs roughly 3.3x just to break even on first order. The free breakeven ROAS calculator does this per margin band in seconds.
2. Decide how much future-cohort value you are licensed to spend against (your payback ceiling answers this), and adjust acquisition targets accordingly.
3. Give every campaign a target derived from the margin and repeat behavior of what it sells — one blended account-wide target is the spreadsheet error that launches a thousand bad months.
4. Revisit quarterly with finance in the room. Margins move, shipping moves, product mix moves; targets that do not move with them rot silently.

Where this model stops being reliable

• →A blended waterfall hides a mixed catalog. One contribution margin across a 60%-margin hero product and a 20%-margin accessory range describes neither. If a promotion shifts volume between them, your breakeven moves without a single ad metric changing. Build the waterfall per margin band before you trust a single account-wide target.
• →Realized cohort curves are backward-looking by construction. They are still the honest input, but a brand whose product, price or customer mix changed a year ago is forecasting from a business it no longer runs. Re-cut the curves whenever something structural moves, and distrust any curve built on a single exceptional cohort.
• →Returns provisions are usually guessed and frequently wrong. In categories with high return rates the provision is one of the largest lines in the waterfall and the least carefully estimated. If you have never reconciled provisioned returns against actual, that line is decorative and so is your breakeven.
• →LTV cannot license spend your bank cannot fund. The payback ceiling is a hard constraint and the LTV ratio is a soft one. When they conflict, the ceiling wins every time, and treating a strong ratio as permission is exactly the failure worked through above.
• →None of this identifies the constraint on growth. Sound unit economics tell you whether to scale, not what is stopping you. If the binding constraint is creative, or supply, or the site, better arithmetic will not move it — see how to scale an ecommerce business for the diagnosis that comes first.
• →Below roughly ten thousand a month in spend, this is more precision than the data supports. Monthly cohorts are too small to produce stable curves, and the honest version is a rough contribution margin and a first-order payback rule until volume makes the rest legible.

None of this is glamorous, and all of it is upstream of every scaling decision you will make this year. Get the numbers agreeing — honest CAC on new customers only, LTV on margin from realized cohorts, an explicit payback ceiling from the balance sheet, a MER floor from the waterfall — and the ad account becomes a machine for executing a strategy instead of a place to argue about one. The disagreements do not survive the arithmetic, which is the actual reason to do it. This math is the entire operating system behind our DTC growth practice, and for how these numbers behave once you push spend, see what changes past $100k a month.