Discounting is the most-used and least-modelled tool in ecommerce. It works, which is the problem, because working and being profitable are different things and only one of them shows up in a revenue report.
Here is the arithmetic most brands never run.
The example
A product sells at €100. Contribution margin is 40%, so €40 per order after cost of goods, shipping, processing and returns.
Apply a 20% discount. The price becomes €80. Your variable costs do not change, because the product still costs the same to buy and ship.
New contribution: €80 − €60 = €20.
A 20% price reduction removed 50% of contribution margin.
The volume you now need
To generate the same total contribution as before, you now need twice the orders.
Not 20% more. Twice.
And those additional orders carry additional variable costs of their own: more picking, more packing, more customer service, more returns. The true requirement is higher than double.
Run this at different margins and the pattern is stark.
| Contribution margin | 20% discount removes | Orders needed to break even |
|---|---|---|
| 60% | 33% of contribution | +50% |
| 50% | 40% of contribution | +67% |
| 40% | 50% of contribution | +100% |
| 30% | 67% of contribution | +200% |
| 25% | 80% of contribution | +400% |
At 25% margin, a 20% discount requires five times the order volume to stand still. No promotion achieves that.
What the report shows you instead
During a discount period, revenue rises. Order volume rises. Conversion rate rises. ROAS often improves, because discounted products convert better on the same ad spend.
Every visible metric improves. Contribution margin falls, and it is the one number that typically is not on the dashboard.
This is why discounting becomes habitual. It reliably produces the appearance of success, and the cost is recorded somewhere nobody is looking until the annual accounts arrive.
The second-order damage
The direct margin loss is the smaller problem.
Customers learn the schedule. If you discount predictably, buyers wait. Your full-price weeks progressively weaken, which creates pressure to discount again, which teaches the lesson more firmly.
You acquire the wrong cohort. Discount-acquired customers have measurably lower repeat rates and lower lifetime value. You are paying acquisition costs for your least valuable segment, then using their volume to justify the promotion.
Reference price resets. Sustained discounting moves what customers believe the product is worth. Recovering from that takes considerably longer than establishing it.
When discounting is correct
There are genuinely good reasons.
Clearing end-of-season stock where the alternative is writing it off. Loss-leading a first purchase in a category with high, evidenced repeat rates. Matching competitors during periods when consumers are actively comparing, such as Black Friday, where absence is more costly than participation.
What these share is that the decision was modelled before it was made.
What to do instead
Model every promotion before it runs. Required volume uplift at your actual contribution margin. If the number is implausible, do not run it.
Discount narrowly rather than sitewide. Categories with genuine margin headroom, or specific stock that needs to move. Sitewide is the most expensive possible version.
Break the pattern. Irregular timing prevents customers from learning the schedule.
Track cohort value by acquisition source. Compare repeat rates for customers acquired on promotion against full price. The gap usually settles the argument.
Try non-price alternatives first. Free shipping thresholds, bundles, loyalty benefits and gift-with-purchase all reduce contribution less than an equivalent headline discount, because they either raise basket size or cost less than they appear to.
The pattern we see most
When we take on an account where discounting has become the growth strategy, the first phase is always the same: establish contribution margin by category, then model what the promotional calendar actually costs.
With one fashion brand, that analysis changed the entire media plan. Replacing flash sales with always-on activity, alongside restructuring the paid mix, produced 32% revenue growth with first-order contribution margin up 21%.
Revenue growth that arrives with lower contribution margin is not growth. It is turnover, and turnover does not pay salaries.
