The CFO's Guide to Amazon Prime Day

Right now, every consumer tech site is running the same headline. The Verge has a Prime Day 2026 deals guide; so does every other outlet. They are written for the person holding the cart, the shopper trying to time a laptop purchase. If you own a brand that sells on Amazon, you read those guides as a shopper too, and that is exactly the problem. Because on the other side of that cart, Prime Day is not a deals event. It is one of the most financially dangerous weeks of your year, a margin-and-cash event wearing a sales-event costume.

Prime Day 2026 runs June 23 to 26. US shoppers spent roughly $14.2 billion during the 2024 event, up 11.7% year over year, and the gravitational pull is so strong that even brands running no formal deal saw about a 46% sales lift versus a normal week. That number cuts both ways: the traffic is real and worth competing for, but so is the cost of competing for it. This guide is the CFO's version of the deals roundup. Not what to buy, but how to run the four days so they build your business instead of quietly bleeding it.

The two Prime Days: the one shoppers see and the one that hits your P&L

There are two Prime Days happening at the same time. The shopper's Prime Day is a list of discounts. The seller's Prime Day is a contribution-margin and cash-flow event, and almost every founder I talk to is unconsciously running the shopper's version from the seller's chair. They see a big GMV number on the dashboard, feel good, and only discover in July that the month was less profitable than a quiet week in May.

The reason is that revenue is the most flattering and least informative number in the building. A 40% spike in units sold tells you nothing about whether you made money, because Prime Day attacks your economics from three directions at once: a deeper discount, a stack of Amazon fees, and an advertising auction that gets brutally expensive for exactly the four days you most want to be in it. GMV goes up; contribution margin can go to zero or below while it does. The job for the next four sections is to make all three of those forces visible and put a number on each, so Prime Day becomes a decision you make on purpose rather than a wave you get swept up in.

A Prime Day deal is a cost stack, not a discount

When founders think "25% off," they picture giving up 25% of the price. The real arithmetic is worse, because the discount lands on top of a fee stack that does not shrink with it. Walk one unit through the whole stack and the picture gets honest fast.

Start with a product that normally sells for $40 and carries a healthy contribution margin. On Prime Day you mark it 25% off, to $30. Amazon's referral fee (around 15% for most categories) is charged on the discounted price, so that is $4.50. FBA fulfillment takes its fixed cut regardless of your discount, call it $5.50 for a mid-size unit. Your COGS and inbound freight do not care that it is Prime Day, so that is another $13.50 or so. If you enrolled the SKU as a Best Deal or Lightning Deal, there is a deal fee of $100 plus 1.5% of sales sitting on top. And then there is the advertising, which is the lever most founders underestimate: Amazon CPCs spike 60 to 80% during Prime Day, reaching roughly $1.35 to $1.45 on average and well above that in competitive categories, because every seller bids into the same auction at the same moment.

That is the whole point in one row: a SKU that throws off $9 of contribution on a normal week can go slightly negative on Prime Day, and the founder running the shopper's Prime Day never sees it because the units-sold line looks fantastic. The discount did not cost you 25%. The discount, plus the fees that ride on every unit, plus the ad auction you were forced into, cost you all of it.

None of this means do not run Prime Day. It means run it with the stack in front of you. The brands that win on Prime Day are not the ones with the deepest discount; they are the ones who picked the two or three SKUs that can absorb the stack and still clear a contribution they chose on purpose. For the wider discipline on this, our discount and promo strategy guide walks through how to set markdowns that defend margin instead of training customers to wait.

The cash-flow trap nobody models: you spend in June, you get paid in July

Here is the part that turns a margin problem into a survival problem, and it is brand new for 2026. On March 12, 2026, Amazon's DD+7 payout policy went live. Your FBA sales proceeds are now held until seven days after the order is delivered, and only then do they join the next 14-day disbursement cycle. Net effect: cash from an FBA sale reaches your bank 14 to 27 days after the order, not after you ship it.

Lay that over the Prime Day calendar and the trap is obvious. The sale happens June 23 to 26. The cash does not land until roughly mid-July. But the costs of generating that sale are all due in June: the inflated ad spend is billed daily across the event, and if Prime Day clears your shelves, the restock purchase orders (often with deposits to your supplier) need to go out immediately so you are not stocked out for the rest of July. You spend in June. You get paid in July. The gap in between is real money you have to find somewhere.

A useful rule of thumb: the working capital cycling through the DD+7 reserve at any moment is roughly your daily revenue times seven. A brand doing $1M a year on Amazon (about $2,740 a day) has roughly $19,000 tied up in the reserve on a normal week; during a Prime Day spike that figure balloons with the volume. If you are eligible, Amazon's Express Payout (free for sellers transacting $1M or less) can shorten the tail by getting eligible funds to you within about 24 hours, and it is worth turning on before the event. But the structural answer is to forecast the gap, not to wish it away. This is exactly the muscle our cash-flow forecasting guide builds, and why a cash reserve and runway buffer earns its keep in the weeks around any peak event. If your processor and marketplace payout timing is something you have never mapped, payout timing and cash flow is the place to start.

The pre-event finance checklist (what to lock now)

For Prime Day 2026 specifically, a piece of news first: the deal-enrollment deadlines have passed. Best Deal and Lightning Deal submissions closed April 30, so for this year you cannot add new headline deals. That is fine. The three levers that actually protect the P&L are still fully in your hands, and they are the ones founders neglect anyway.

First, set a contribution-margin floor per SKU, not a discount percentage. Build the stack from the table above for each SKU you intend to promote or advertise hard, and decide the lowest contribution you will accept. Anything that breaches the floor either gets a shallower deal, gets excluded, or gets explicitly reclassified as capped customer-acquisition spend with a dollar limit attached. Second, confirm inventory cover. Stocking out mid-event is the worst outcome, because you pay the elevated CPCs to send traffic to a listing that cannot convert, and you lose the post-event ranking bump. Know your expected sell-through and your reorder lead time, and make sure you are covered through mid-July. Our demand-forecasting guide covers how to size that without over-ordering into dead stock. Third, set the ad guardrails: a daily budget and a daily ACOS ceiling per campaign, decided in advance, so you are not making bidding decisions emotionally at 11pm on June 23.

And if a chunk of your COGS is import duty, remember that the Prime Day discount is being applied to a unit whose landed cost may already be inflated by tariffs; our DTC tariff exposure index is worth a look before you decide how deep you can afford to go.

During the event: the three numbers a CFO watches

Once the event is live, the dashboard will try to seduce you with GMV. Ignore it. There are three numbers worth watching in close to real time, and none of them is revenue.

The first is running contribution margin, not sales. If you have built the per-unit stack, you can track contribution dollars as they accumulate and know whether the day is actually making money. The second is ACOS against the ceiling you set, by campaign. When the auction heats up, the temptation is to raise bids to "stay competitive"; the discipline is to let volume fall before you let ACOS blow through the line you drew, because chasing the auction is how a profitable event turns into a paid one. The third is sell-through against cover, so you can see a stockout coming hours before it happens and decide deliberately whether to pull ad spend on a SKU that is about to run dry rather than burning clicks on it. Revenue is the output of running those three well; it is not a number you steer by.

After: reconcile the true Prime Day P&L and carry it into BFCM

The work that separates operators from gamblers happens the week after. Pull the Amazon settlement report and build the true, fully-loaded Prime Day P&L: gross sales, minus every fee, minus the deal fees, minus the actual (not budgeted) ad spend, minus COGS and inbound, minus the returns that come in over the following two to three weeks. Returns matter here because they land after the event and quietly claw back contribution you thought you had banked. Only once that is reconciled do you know whether Prime Day made or lost money, and on which SKUs.

That reconciled number is the most valuable thing Prime Day gives you, because it is your dress rehearsal for BFCM. Black Friday and Cyber Monday are the same forces at a larger scale and a worse cash position (you are buying holiday inventory at the same time). The brand that knows its true Prime Day contribution by SKU, knows which deals worked and which were vanity, and has measured its own cash gap under DD+7 walks into Q4 with a model instead of a hope. The brand that only remembers "Prime Day was big" walks in blind. If you want this run against your own numbers, sized for your categories and your cash position, that is the work we do as a fractional CFO for ecommerce brands. The clean version of that reconciliation also makes your books materially easier at year end; if Amazon accounting is a mess for you, start with the best accounting software for Amazon sellers.

Sources and methodology

Prime Day 2026 dates (June 23 to 26) and the 2026 deal-fee schedule (Best Deal and Lightning Deal at $100 plus 1.5% of sales, Prime member coupons at $5 plus 2.5%, enrollment closing April 30) are drawn from Amazon seller-policy reporting via scaledon.com and amzprep.com. The $14.2 billion US Prime Day 2024 figure (up 11.7% year over year) and the roughly 46% non-deal halo lift come from smartscout.com and Acadia's event data. Amazon advertising CPC benchmarks (average near $0.98 in 2025, spiking 60 to 80% to roughly $1.35 to $1.45 during Prime Day and Q4) are from sellermetrics.app and Ad Badger. The DD+7 payout mechanics (effective March 12, 2026; FBA funds held seven days after delivery, then released on the 14-day cycle, for an order-to-bank window of 14 to 27 days; reserve working capital of roughly daily revenue times seven; Express Payout free for sellers at or below $1M) are from riverbendconsulting.com and onrampfunds.com.

The unit-economics and cash-timing tables are illustrative Eightx models built on those benchmarks to show the mechanics; they are not a forecast for any specific brand. Real figures vary by category, FBA size tier, discount depth, ACOS, and return rate, which is exactly why the recommendation is to build the stack for your own SKUs rather than trust a category average. Operator-voice observations are synthesized from anonymized founder-call patterns, with no client named.

Frequently asked questions