Discounting and ROAS are joined by one number that never appears in an ad report: gross margin. A discount does not cost you its own percentage in profit. It usually costs most of the profit, because the cost of the goods does not move when the price does. Take a $100 order at 40% margin. A quarter off cuts contribution from $40 to $15. Break-even return jumps from 2.5x to 5.0x. Nothing in Meta or Google will mention any of that. Your campaign keeps reporting a 3.0x, keeps looking healthy, and quietly loses money on every order it books.
This is for ecommerce operators who run promotions and then judge them on the platform’s return figure.
How discounting and ROAS actually connect
Through margin, and only through margin.
Return on ad spend is revenue divided by spend. A discount lowers revenue slightly and contribution enormously. Cost of goods was fixed when you bought the stock.
So the metric barely reacts while the economics collapse underneath it. Sell 100 units at full price or 120 at a quarter off. The platform may report a similar multiple for both. Only one of those weeks paid your wages.
Effective margin is the number to track. It is contribution divided by the price actually charged, after the discount has been applied. Every break-even calculation should run on that figure, not on the list margin in your product sheet.
Most teams never rebuild the target when the offer changes. The offer changes weekly, and the target changes never.
The arithmetic on a $100 order at 40% margin
Full price first, so the comparison is clean.
A $100 order at 40% gross margin carries $60 of goods and leaves $40 of contribution. Break-even return on ad spend is one divided by 0.40, which is 2.5x. Anything above that multiple makes money on the first order.
Now apply 25% off. Revenue becomes $75, the goods still cost $60, and contribution falls to $15. Effective margin is $15 over $75, which is 20%. Break-even return is now 5.0x.
Read those two numbers together. Discount depth of a quarter doubled the multiple you have to hit. Reporting still shows the same campaign with the same creative.
A 3.0x return was comfortable before the promotion. During it, that same 3.0x loses money on every order.
Discounting and ROAS at 20% margin, where it breaks
Thin margin businesses cannot discount at all, and many of them do it anyway.
Across my ecommerce clients gross margin runs from 20% to 50%. The low end of that range is a real business, not a hypothetical one. Take the same $100 order at 20% margin. Goods cost $80 and contribution is $20, putting break-even at 5.0x before any promotion.
Apply 25% off and revenue is $75 against $80 of goods. Contribution is minus $5. Every order loses money before a single penny of advertising is counted.
The 50% margin business survives the same offer. Goods cost $50, discounted revenue is $75, contribution is $25, and effective margin is 33%. Break-even moves from 2.0x to 3.0x, which is painful and survivable.
Identical offer, identical creative, three completely different outcomes. Discounting and ROAS cannot be discussed sensibly without the margin attached.
What Adobe measured about discount depth in holiday 2025
Somebody does publish credible discount data, with a disclosed method.
Adobe Analytics reported peak apparel discounts of 25.1% and electronics at 30.9%, in holiday analysis published in January 2026. Both cover the 2025 US online season, where spending reached $257.8B, up 6.8% year on year. Mobile carried 56.4% of transactions.
Sample size is the reason to trust it. Adobe’s figures come from over a trillion visits to US retail sites and 100 million SKUs. No other source in this subject discloses as much.
Hold the apparel figure next to the arithmetic above. A category discounting at roughly a quarter needs margins near the top of the 20% to 50% band. Anything lower will not survive its busiest weeks.
Peak rates are also not season averages. Read 25.1% as the deepest point rather than the running level. Check your own promotional pricing against the same distinction.
Why the platform never sees the discounted margin
Your ad account records whatever your purchase event tells it, and that is often wrong twice over.
Some stores fire the event with list price rather than the amount paid. Others include tax and shipping in the value. Both inflate reported revenue, so the return figure sitting in your dashboard was never the number you thought.
I run one US automotive equipment ecommerce account with entirely ordinary delivery metrics. Those read $8.28 CPM, $0.28 CPC, a 2.92% click-through rate and 43,415 link clicks. Its conversion side reports a return on ad spend of 44.37, which no ecommerce business on earth achieves.
That is a counting fault rather than a result. If an event can be wrong by that margin, treating its revenue figure as discount-adjusted is optimistic.
Place a discounted test order and read what value actually arrives. Fix the event before you argue about targets.
Discounting and ROAS on first orders versus repeat buyers
A welcome code is acquisition spend that never reaches the ad account.
Give 15% off a first order worth $100 and you have spent $15 buying that customer. Finance sees it. Meta does not. Your true new customer acquisition cost is media cost plus discount. Only one of those two appears in the interface.
Repeat buyers are the more expensive mistake. A discount sent to somebody who was already going to reorder is pure margin donation. It also trains the list to wait for the next code.
Segmentation fixes most of this. Reserve depth for people who have never bought, keep full price for buyers with a proven repeat pattern, and measure the two cohorts separately.
Do that and discounting and ROAS stop fighting each other, because the discount is buying something you were not going to get for free.
Codes leak beyond the campaign that issued them
A public discount code stops belonging to the campaign roughly an hour after it launches.
Coupon sites and browser extensions collect them, and organic, direct and email traffic all start redeeming a code that was written to buy new customers. Margin leaves through channels that were never going to need it.
I run Meta, Google Ads and Klaviyo for one US flag and banner ecommerce client. Three channels can each claim the same discounted order, and the discount is deducted from none of their reported revenue.
Two practical fixes work. Issue unique single-use codes per campaign, or apply an automatic discount tied to a specific landing page rather than a typed word.
Search matters here as well. When coupon sites rank for your brand name plus the word discount, you are paying for demand you already own, and the fix is an organic one rather than a bidding one.
Discounting and ROAS in Q4, when CPMs rise as well
Both sides of the equation move against you in the same fortnight.
Gupta Media’s social CPM tracker put Meta’s Cyber Monday cost per thousand at $17.70, which is 138% above the 2024 annualised average. Thanksgiving week averaged $13.42, against a 2025 Meta platform average of $8.19. Those figures come from a tracker published in October 2025 running tens of billions of impressions on a 14-day trailing average.
Stack that on the margin arithmetic. Media costs roughly double while your effective margin halves, and the break-even ROAS you set in September is meaningless by the last Monday in November.
Plan two targets. One for full-price trading and one for promotional weeks, each built from the effective margin that applies during it.
Brands that skip this step usually discover the problem in January, when the finance team reconciles a record revenue month against a disappointing profit line.
What I reach for before cutting price
Price is the last lever, not the first, because it is the only one that cannot be quietly reversed.
Raising average order value comes first. Bundles, a free shipping threshold and a sensible upsell lift contribution without touching price integrity or the auction.
Cutting waste comes second, and it is free. On that flag and banner account, removing one failing geographic test moved the period from a 1.58 return across 64 purchases to a 1.99 return across 61. The test had spent about $1,027 for three purchases at roughly $341 each, on a 0.25 return.
Note what it looked like beforehand. That ad set had the highest reach in the account, 17,086, and the most link clicks, 542. Traffic metrics called it the winner and revenue called it the only loser.
Honest caveat on that account: the client has never disclosed gross margin to me, so I cannot tell you whether 1.99 was profitable. I can tell you it beat 1.58 for free.
What to change this week
Three things, and the first one takes ten minutes.
Recalculate break-even on your effective margin after your standard promotion, then compare it with the target currently sitting in your bidding. Most accounts are running a full-price target through a discounted quarter.
Then place a test order with a code applied and check what revenue value your purchase event sends. List price, or the amount actually paid, changes every number downstream.
Finally, split your promotional and full-price periods in reporting, so you can see which of them is carrying the year.
Working out what an order is genuinely worth is the valuation step underneath all of this. Two dated figures for what a purchase costs gives my own volumes and periods. The cost gap between a first order and a repeat one explains why cohort discounting beats blanket discounting, and the real target hiding inside your blended figure covers the denominator problem. Coupon sites ranking on your own brand terms are an organic problem, which is what my SEO work deals with. To have your promotional maths checked before the next sale, book a teardown.