Discount Dependency: How Promotions Become The Business

The short version

Discounting reliably lifts conversion. That reliability is the trap: because it always works, it gets used whenever numbers need to improve, and each use trains customers that waiting is rewarded.

The end state is a business where the discounted price is the real price, full-price periods produce almost nothing, and margin has been permanently transferred to customers who would have paid more.

Getting there does not feel like a decision. It feels like a series of reasonable months.

The mechanics of the trap

Each promotion pulls demand forward rather than creating it. Post-promotion periods underperform because you already sold to those buyers.

The underperformance justifies the next promotion. The quiet period looks like weak demand, and the fix that worked before is available.

Customers learn the cadence. Predictable promotions are the most damaging kind. If your sale falls at the same time each month, nobody buys in between.

Your acquisition mix shifts. Discount-acquired customers repeat differently, so your customer base gradually becomes people who only buy on discount.

What it does to your advertising numbers

Reported ROAS improves. Higher conversion at the same traffic cost. The metric says you are doing better.

Contribution margin falls. The discount comes straight out of it, and margin is what actually pays for the advertising.

Returns rise. Discounted purchases are returned more often, so net revenue falls further than the discount alone.

Incrementality falls. Much of the lift is people who would have bought anyway, now buying cheaper. You paid to reduce your own prices.

The combination is why discount-heavy periods can show the best ROAS of the year and the worst profit.

Discounting that does not create dependency

Discounting is not always wrong. Structure decides.

Conditional rather than universal. A threshold that raises basket size, a bundle, first-order only. The customer does something to earn it.

Unpredictable rather than scheduled. Irregular timing prevents waiting behavior.

Segmented rather than sitewide. Offered to customers who need it, not to everyone including those about to pay full price.

Non-price value instead. Free shipping, extended returns, a gift. Costs less margin and does not reset the reference price.

That last point is the important one: a discount changes what customers believe the product is worth. Added value does not.

Measuring whether you are dependent

Share of revenue from discounted orders, tracked over time. Rising is the signal.
Full-price conversion rate between promotions. Falling means customers are waiting.
Repeat rate split by first-order discount depth. If deep-discount customers barely repeat, acquisition through discount is not building anything.
Contribution margin per order, not revenue per order.

Practical rule

Discount conditionally, unpredictably, and to segments — never as the routine answer to a slow month. And judge promotions on contribution margin net of returns, because reported ROAS improves in exactly the cases where the business does not.