Your Discounts Are an Ad Budget You're Not Measuring

Your Discounts Are an Ad Budget You're Not Measuring

A store owner will obsess over a $2,000 ad campaign. They'll check ROAS daily, pause weak ad sets, and argue about attribution windows.

Then they'll run a 20% off sitewide sale that gives away $50,000 in margin and check… nothing. Not incrementality. Not margin. Just "revenue went up."

Discounts are the largest unmeasured marketing line in ecommerce. And unlike ads, they're usually not even tracked as spend.

A discount is a media buy

Think about what a discount actually is: money you spend to change behavior. You give up margin in exchange for something — a sale you wouldn't have had, a bigger basket, a returning customer.

That's exactly what an ad is. Money spent for a behavior change.

So why do we measure ad spend to the dollar and discount spend to the vibe? Because ads have a dashboard and discounts don't. The ad platform reports ROAS. Your store just reports that revenue went up.

The problem: revenue counts the sales you already had

When you run a discount and revenue rises, you have no idea how much of that rise you caused.

The research is not kind here:

• Classic promotion research (Ailawadi & Neslin) found that 53% of promotions were negative-profit — they cost more margin than they generated in incremental contribution.
• Studies of European grocery promotions found 60–80% were not ROI-positive.
• Decompositions of promo "lift" typically find only about 30% is genuinely incremental — the rest would have happened anyway.

If that's even roughly true for you, most of your discount budget is a transfer, not an investment. You're paying customers to do what they were already going to do.

The metric: incremental ROAS for discounts

Ads have ROAS: revenue ÷ spend. Discounts need the same thing, corrected for incrementality.

Discount iROAS = incremental contribution margin ÷ discount cost

• Incremental contribution = margin from the orders the discount caused, not all orders it touched.
• Discount cost = total margin given away.

Worked example. A campaign reports 1,000 conversions and $50,000 revenue on $4,000 of discount cost. Reported "ROAS" = 12.5. Looks incredible.

Now run a holdout: 50,000 sessions with the discount, 50,000 without. Control converts at 2.00%, treatment at 2.40%.

• Incremental orders = (0.024 − 0.020) × 50,000 = 200
• Incremental revenue = 200 × $50 = $10,000
• Incrementality rate = 200 ÷ 1,000 = 20%

So 80% of those conversions were going to happen anyway. Corrected ROAS = 12.5 × 0.20 = 2.5. At a 40% margin, incremental contribution = $4,000 — exactly the discount cost. Break-even, before the discount handed to the other 800 orders that didn't need it.

The campaign looked like a 12.5x winner and was margin-negative.

Why nobody does this

1. No dashboard. Incrementality isn't a number your platform shows. You have to design a test.
2. It's uncomfortable. The honest answer is often "most of that discount did nothing."
3. It needs a control group. A 5% holdout is invisible in your revenue and tells you everything about the other 95%. The resistance is psychological, not practical.

What to do

1. Track discount spend like ad spend — put a dollar figure on margin given away, by campaign.
2. Run a holdout on every significant promo — 5% of traffic sees no offer; compare conversion and contribution.
3. Compute discount iROAS — incremental contribution ÷ discount cost. Below 1 means the promo lost money, whatever revenue did.
4. Fix or kill the losers — some promos are genuinely incremental (new-customer offers, real events); others are margin leaks dressed as growth.

The uncomfortable conclusion

You're running a marketing channel — discounts — with the biggest budget and the worst measurement in the business. You'd never accept that from your ad platforms.

The fix isn't to stop discounting. It's to treat discount spend with the same rigor as every other dollar you spend to grow.

Your discounts are a media buy. Measure them like one.