Prime Big Deal Days submissions close September 8. Black Friday and Cyber Monday close October 20. The $50 early-bird discount on PBDD expired on August 5, so if you’re deciding now you’re deciding at full freight.
Which means over the next three weeks, hundreds of brands will pick which ASINs go on deal, at what depth, using a calculation that is wrong in a specific and expensive way.
After managing hundreds of brands through Q4, the single most common analytical error we see isn’t in advertising, inventory or creative. It’s this: brands compute deal performance across every unit sold during the deal window, when the only units that matter are the ones that wouldn’t have sold otherwise. That distinction is not academic. On a lot of the deals we’re asked to review after the fact, it’s the entire result.
The Denominator Problem
Here’s the calculation almost everyone runs. Deal week: 2,400 units at $29.99 discounted from $39.99. Revenue $71,976. Compare to the prior week’s 900 units at full price, revenue $35,991. Revenue doubled. The deal worked.
Now here’s the calculation nobody runs.
Of those 2,400 units, some number were going to be purchased anyway. Your baseline was 900 units a week. Even with zero deal, deal-period traffic on Amazon is elevated — the whole platform is busier — so a fair baseline for that week might be 1,100 or 1,200 units at full price.
So the honest split is roughly 1,200 baseline units and 1,200 incremental units. And on the baseline units you didn’t gain anything. You handed each of those buyers $10 they were prepared not to receive.
Run it as contribution. Say landed cost is $9, referral is 15% and FBA is $6.
- At $39.99: revenue $39.99 − $9 COGS − $6.00 referral − $6 FBA = $18.99 contribution
- At $29.99: revenue $29.99 − $9 COGS − $4.50 referral − $6 FBA = $10.49 contribution
Baseline units, had there been no deal: 1,200 × $18.99 = $22,788.
Actual, all 2,400 units on deal: 2,400 × $10.49 = $25,176.
Add the deal fee — US promotions currently run $100 prepaid plus a 1.5% variable fee on deal sales, capped at $5,000 — and that’s $100 + $1,080 = $1,180. Now you’re at $23,996.
You doubled units, doubled revenue, ran a week of operational strain, and generated roughly $1,200 more contribution than doing nothing. Before you count a single dollar of the ad spend most brands add on top to hit deal velocity.
That is not automatically a bad outcome. There are real reasons to buy velocity at close to break-even. But it is a completely different decision from the one the revenue chart described, and most brands never see the second version of the number.
The Four Costs That Don’t Make It Into the Model
1. The discount applied to units you already had. Covered above, and it’s the big one. The deeper the discount and the higher your baseline volume, the more of your own money you’re spending on people who had already decided.
2. The promotion fees. $100 per promotion plus 1.5% of deal sales. Small relative to the discount, but they’re real and they’re per-deal, which means the brand that submits eleven ASINs to be safe is paying $1,100 in prepaid fees for a portfolio it never modelled at the SKU level.
3. The ad spend nobody attributes to the deal. Almost every brand raises bids or budgets during deal windows to make sure the deal gets seen. That’s often correct. It is also a deal cost, and it gets booked to advertising, where it disappears into a monthly ACOS number. If you increased spend 40% for a week to support a deal, that increase belongs in the deal’s P&L, not in the ad account’s.
4. The post-deal trough. This is the one that surprises people. When a visible discount ends, conversion on that ASIN routinely runs soft for two to four weeks. Some of it is demand pull-forward — buyers who would have purchased in week three bought in week one. Some of it is that shoppers who saw the lower price now have a reference point, and your regular price reads as expensive against the number they remember. Whatever the mix, the recovery weeks are part of the deal’s result, and a P&L that closes on Sunday night doesn’t contain them.
Measure the deal over six weeks, not one. The first week tells you what you sold. Weeks two through six tell you what it cost.
How to Actually Measure Incrementality
You don’t need a data science team. You need a baseline built before the deal runs, and almost nobody pulls one.
Build the baseline from two sources, not one. Take the trailing four weeks of units at full price, and take the same calendar window last year. The trailing four weeks tell you your current run rate; last year’s window tells you how much your category lifts during that event regardless of what you do. If last year’s PBDD week ran 20% above your then-baseline without a deal, apply that lift.
Then compute contribution on incremental units only. Incremental units × deal contribution per unit, minus the discount given away on baseline units, minus fees, minus incremental ad spend. That’s your number.
Check rank 30 days later. The strongest argument for running a deal at thin contribution is that velocity feeds ranking, so you’re buying organic position that keeps paying after the discount ends. We think that’s frequently true. We also think most brands assert it and never verify it. Screenshot organic rank on your top ten keywords the week before the deal and check the same terms 30 days after. If rank held, the deal bought you something the P&L can’t see. If it slid straight back, you bought a week of revenue at a discount and you should price the next one accordingly.
Do it at the SKU level. A blended read across eleven deal ASINs is an average that describes none of them, and the SKUs that behave worst are almost never the ones anyone is watching.
When a Deal Is Genuinely Worth Running
We’re not against deals. We’re against unmodelled ones. Four cases where the math usually works even at thin contribution:
Rank-building on a newer SKU. Low baseline volume means the discount-on-existing-demand cost is small — there isn’t much existing demand to subsidise — and the velocity is buying position you’ll monetise for months. This is the cleanest case in the catalog and it’s the one brands under-use, because deals get assigned to hero SKUs by habit.
Clearing aged inventory before October 15. The peak fulfilment surcharge window opens October 15 and aged-inventory surcharges compound on units sitting past their thresholds. A deal that moves stock ahead of both is being paid for by costs you avoid, not just margin you capture. Compare the deal contribution against the carrying cost, not against full-price contribution.
Genuinely high-margin SKUs. If a 25% discount still leaves double-digit contribution per unit, the deal is buying reach at a real profit. Most catalogs have two or three of these and they’re rarely the ones submitted.
Multi-SKU brands buying new-to-brand customers. If the deal SKU is an entry point into a range with real repeat purchase, thin contribution on the first order is an acquisition cost and should be judged on new-to-brand percentage, not on the deal’s own ACOS. Pull the new-to-brand split — if it’s low, you discounted to your existing customers and called it acquisition.
When It’s a Bad Deal
High-repeat replenishment SKUs with a Subscribe & Save base. These are the worst deal candidates in most catalogs and the most frequently submitted, because they’re the top of the revenue list. A large share of that volume is buyers who would have purchased regardless — subscribers, repeat purchasers, people arriving on branded search. You are running a promotion aimed at strangers and paying for it out of the pockets of your most loyal customers.
Thin-margin SKUs. If contribution at full price is $4, a 20% discount on a $25 item removes $5. There is no volume that fixes that, and “we’ll make it up on rank” is not a plan you can price.
Your only hero, with nothing to halo into. The acquisition argument depends on somewhere for the customer to go next. A single-SKU brand running a deep deal is buying revenue, not customers.
The Pre-Submission Math, in Four Steps
You have three weeks before September 8. This is a 90-minute job for a whole catalog.
FAQ
Is a deal that breaks even on contribution a failure?
No — but it needs a stated reason. Rank, aged inventory, or new-to-brand acquisition are all legitimate reasons to run a deal at break-even. “It’s Prime Big Deal Days” is not a reason, it’s a date.
How much of my deal-week volume is typically incremental?
It varies enormously by SKU and category, and anyone quoting you a single percentage for your catalog is guessing. What we can say confidently is that it is never 100%, it’s lower on high-repeat SKUs than on discovery SKUs, and the only way to know your number is to have a baseline written down before the deal runs.
Does the post-deal trough mean I should avoid deals entirely?
No. It means you should measure over six weeks. Some SKUs recover in ten days and some take a month. The ones that take a month are usually the ones where the discount was deep enough to reset the price the shopper thinks the product is worth.
Should I submit lots of ASINs to see what performs?
That’s a $100-per-promotion experiment with a discount attached to every arm. Submit the SKUs you’ve modelled. A wide submission isn’t a test, it’s a hedge, and hedges cost money in both directions.
We missed the August 5 early-bird deadline. Does that change anything?
It costs $50 per deal, which should not change your decision on any deal worth running. It’s a useful signal about your internal approval speed, though — if a $50-per-deal discount passed without anyone noticing, the September 8 date is worth putting on someone’s calendar by name this week.
—
If you’re looking for a team that manages every lever — creative, advertising, and operations — Velocity Sellers works with brands doing $100K+/month on Amazon. Contact us for a free account audit.
