How the Ecommerce Profitability Flywheel Compounds

Every example and amount on this page is illustrative. It shows the method and is not a benchmark, a quote or advice.

What the Flywheel Is Really Measuring

The ecommerce profitability flywheel measures whether contribution profit can be converted into more profitable demand. The loop has four connected parts: contribution profit, reinvestment, customer acquisition and repeat demand. A sale creates contribution only after the costs required to deliver and acquire it are accounted for. Some of that contribution can then be reinvested in acquisition. Acquisition brings in customers; repeat purchases create additional contribution without treating the original order as the whole economic result.

This is not a measure of brand momentum, audience size or top-line revenue. Revenue can rise while the flywheel weakens if discounts, product costs, returns or advertising consume the increase. A £100 order with £40 of product cost, £20 of advertising and £15 of refunds or other variable leakage does not provide £100 to reinvest. The relevant question is how much contribution remains, which customers or products generated it, and whether subsequent demand adds more contribution than the acquisition and fulfilment costs required.

The unit of analysis matters. At product or collection level, compare contribution after advertising rather than sales alone; product profitability analysis explains the cost and return detail behind that view. At customer level, connect the first order with later orders and their profitability, not just cumulative revenue. Customer profitability analytics supports that distinction. For acquisition, keep total advertising efficiency separate from channel-attributed credit: MER and ROAS analysis helps show whether reinvestment is supported by store-wide contribution or only by an attribution view.

A healthy flywheel therefore means the measured contribution from acquired and returning customers is sufficient to support deliberate reinvestment. It does not mean every customer or campaign must be profitable immediately, nor does it claim that reinvestment will produce a particular result. The measurement is a financial relationship between realised contribution and the demand that follows.

The Formula, Without the Fog

Use this formula at the customer or cohort level:

First-order contribution after acquisition = first-order contribution profit − CAC

Lifetime contribution after acquisition = first-order contribution profit + (repeat purchase rate × repeat orders per repeating customer × repeat-order contribution profit) − CAC

Reinvestment capacity = max(0, lifetime contribution after acquisition) × reinvestment rate

The next-period acquisition capacity is then reinvestment capacity ÷ expected CAC. This is an arithmetic relationship, not a forecast that reinvesting will produce a specific result.

First-order or repeat-order contribution profit is the order’s realized revenue minus variable costs required to sell and fulfil it. Include product cost, payment fees, fulfilment costs and expected return costs where those costs apply. Keep advertising separate: CAC is advertising spend divided by the number of newly acquired customers in the same period. A 40% contribution margin on a $100 delivered order means $40 of contribution profit before acquisition cost; a $25 CAC leaves $15 after acquisition.

Repeat purchase rate is the proportion of newly acquired customers who place at least one additional order during the measurement window. Because a repeating customer may order more than once, multiply that rate by the average number of repeat orders per repeating customer. If your repeat purchase rate is r and repeaters place n additional orders on average, the model uses r × n repeat orders per acquired customer.

The reinvestment rate is the share of contribution after CAC that the business puts back into acquisition. At a 60% rate, $15 of contribution after CAC creates $9 of reinvestment capacity. Accurate inputs require delivered revenue rather than cancelled or returned sales, consistent cost treatment, and one trading currency. Product-level contribution can be checked in product profitability analysis, while customer repeat behaviour belongs in customer profitability analytics and acquisition cost in advertising profitability reporting.

A Worked Example From First Order to Repeat Revenue

Take a cohort of 100 new customers buying a £120 product bundle. All 100 orders are delivered, so realized revenue is £12,000. Product cost is £36 per order, fulfilment is £8, and payment fees are £4. Variable cost is therefore £48 per order, or £4,800 for the cohort. Contribution before acquisition is £72 per order, or £7,200 total.

The merchant spends £4,500 on the ads that acquired these customers. First-order contribution after advertising is £7,200 − £4,500 = £2,700, or £27 per customer. That is the amount available to cover fixed costs and fund the next decision, not £12,000 of sales. A product profitability analysis should keep this calculation at product or collection level where the economics differ.

Over the next period, 40 customers place a second £90 order. Each repeat order has £27 of product cost, £8 of fulfilment and £3 of payment fees, leaving £52 contribution. Repeat revenue is 40 × £90 = £3,600, and repeat contribution is 40 × £52 = £2,080. The cohort now produces £9,280 of cumulative contribution after ads: £2,700 + £2,080. On £4,500 of acquisition spend, that is £47.80 of cumulative contribution per acquired customer.

• Next investment decision: the observed cohort supports a controlled increase in acquisition spend only if the merchant can preserve these economics and fund the delay before repeat orders arrive. A £500 test would buy roughly 11 customers at the current £45 acquisition cost; it should be evaluated against delivered revenue, variable cost, repeat rate and contribution, not first-order revenue alone. Customer profitability analytics can separate those repeat behaviours, while advertising profitability reporting keeps total ad efficiency distinct from attributed channel results.

The Measurement Errors That Break the Loop

Blended figures can hide a loss inside a profitable-looking total. Suppose Channel A produces £8,000 of realized revenue from £1,500 of ad spend, while Channel B produces £2,000 from £1,800. Combined ROAS is £10,000 ÷ £3,300 = 3.03, but Channel B is close to consuming its revenue before product and fulfilment costs. A channel or campaign decision based only on the blended number can shift budget toward demand that does not create contribution profit.

Revenue also changes as an order moves through the period. A £100 order with a £10 discount, £20 refund and £8 delivery cost does not contribute £100. Before product cost and advertising, the usable amount is £100 − £10 − £20 − £8 = £62. Counting cancelled orders, pending returns or gross sales instead of delivered-order revenue overstates the base. The same issue applies when refund value is recorded weeks after the original sale. The product profitability calculation needs returns, discounts and fulfilment costs assigned to the order that generated them.

Inventory creates a separate timing trap. Buying 1,000 units at £12 creates a £12,000 cash outflow, but the period’s product cost depends on the units sold, not simply the purchase date. Conversely, leaving old stock out of the analysis makes a product appear profitable while cash remains tied up in dead stock. Product-level and collection-level profitability should be read alongside inventory position and remaining stock value.

Attribution can misstate both sides of the loop. Attributed ROAS assigns revenue to advertising rules; MER divides total store revenue by total advertising spend. They answer different questions, and neither replaces contribution after advertising. ORVX keeps MER and attributed ROAS separate, reports return rate by units and refund value against revenue, and calculates profitability from the store’s own orders, product cost model and connected ad spend. It also keeps trading currencies separate, avoiding another error: adding figures that were never comparable.

How to Read the Result

A positive flywheel result means the cohort generates more contribution profit than it requires to acquire and serve. For illustration only, if £1,000 of acquisition spend produces £1,300 of contribution after product, fulfilment and returns, the loop has £300 available for reinvestment. That supports scaling, but only when the result holds for the relevant product, channel and customer cohort rather than being caused by one blended total. Check product profitability and repeat behaviour before increasing spend.

A flat result means the business is replacing acquisition cash, not compounding it. £1,000 in and £1,000 out may be operationally sustainable, but it leaves no contribution to absorb volatility, fund growth or cover costs outside the calculation. Do not treat flat economics as a signal to scale. First identify whether conversion, average order value, repeat purchase or contribution margin is limiting the loop.

A negative result means each cycle destroys contribution. If £1,000 of acquisition spend produces only £750 after variable costs, the business loses £250 before considering fixed overhead. Reduce or pause the affected acquisition spend while isolating the cause. Low conversion usually points to improving the offer, landing page or checkout before buying more traffic. Adequate conversion with weak contribution points to pricing, product cost, fulfilment, returns or basket mix; raise contribution margin there rather than simply pursuing more orders.

Read the result at the same level at which you would act. A profitable collection can conceal an unprofitable product, and a profitable channel can conceal loss-making customers. Use advertising profitability to separate MER from attributed ROAS, then compare the resulting contribution by channel and cohort. Scale only the segments with positive, repeatable economics; fix, reduce or stop the segments that do not.

Turn the Answer Into an Operating System

Make the review cadence match the speed of the signal. Check daily for data quality, delivered orders, refunds, ad spend, stock position and unusual changes; use the weekly review for decisions; use a monthly cohort review for structural performance. A single bad day should trigger investigation, not a budget change.

The dashboard should show realized revenue, gross profit, contribution after advertising, contribution margin, ad spend, MER, attributed ROAS, return rate by units, refund value, repeat-order rate, customer acquisition cost and remaining stock value. Keep product and collection views beside customer-level profitability so a strong total cannot hide an unprofitable SKU or acquisition source. Product profitability analysis and customer profitability analytics provide the relevant lenses.

Use cohorts by first-order month, acquisition channel, product or collection, and customer status. Compare each cohort's first-order contribution with its 30-, 60- and 90-day realized contribution, using only completed periods. Label a change structural only when the same direction appears across at least two completed cohorts or review periods; otherwise mark it as volatile or unconfirmed.

• Scale: contribution after advertising is positive, margin clears the store's minimum, and stock can support demand.
• Fix economics: revenue is growing but contribution margin or repeat contribution is below the approved floor.
• Reduce or stop: a channel, product or cohort remains below its contribution floor after the agreed review window.
• Test: the economics are acceptable but the evidence is thin; record the baseline, decision threshold and review date before changing spend.

The profitability flywheel is useful only when each turn is measured with the same definitions: contribution profit after variable costs, acquisition spend tied to the right customers, and repeat revenue tracked by cohort. Calculate it consistently, test which input is limiting growth, and reinvest only when the next dollar has a credible path to more profitable demand.