Amazon did not raise your ad costs on 1 August — it took away the float.

Amazon did not raise the price of your ads on 1 August — it took away the float, and the sellers who confuse those two things are about to cut the wrong campaigns. From 1 August 2026 a directly-notified group of advertisers had their default payment method switched from a card to deduction from their seller or vendor account balance. Nothing about cost-per-click changed. What changed is when the money leaves — and if your bidding was quietly subsidised by thirty days of card float, you are about to discover that it was never profit in the first place.

What Amazon actually announced

The precise wording matters here, because the retelling has been sloppier than the source. Amazon Ads published the update on 14 April 2026: a small number of advertisers had been told their available payment methods were being changed to their seller or vendor account balance, or Pay by Invoice. After feedback, Amazon deferred the change to 1 August 2026 to give that group more time to prepare. The original seller emails, circulating from early April, had named 15 April as the date.

Three details from the announcement that most summaries drop. First, Pay by Invoice remains available — Amazon issues an invoice at the end of each month and payment is due 30 days later. Second, if an advertiser did not select a preference before the change took effect, the default became deduction from the available account balance automatically. Third, the existing card is retained as a backup method to cover insufficient funds, not removed. And Amazon was explicit that this applies only to the group of advertisers contacted directly — via email and via banners in the campaign manager and billing pages of the Ads Console.

So the honest scope is narrow and the honest mechanism is boring: for some accounts, ad spend now nets against proceeds before disbursement instead of hitting a card. That is a treasury change, not a pricing change.

The contrarian part: the float was never margin

The loudest objection during the April backlash was about card rewards and payment terms — sellers describing the gap between spending on ads and paying for them as part of how the business worked. Set aside whether any particular rewards rate is real; the structural point stands on its own, and it is uncomfortable.

Float is a financing benefit. It is not a marketing return. If a campaign only cleared its bar because you paid for the clicks a month after the sales landed, the campaign was not profitable — it was borrowing, and the loan just got called. Removing float does not make a good campaign bad. It makes a bad campaign visible.

Break-even ACoS = unit margin ÷ price. Notice what is not in that equation: payment terms, card rewards, disbursement timing. Moving ad spend from a card to proceeds deduction changes your cash conversion cycle. It does not move your break-even ACoS by a single basis point.

That is the whole arithmetic, and it is why the correct response to this change is not an account-wide bid cut. Your profitable ceiling on 2 August is identical to your profitable ceiling on 31 July. Anyone telling you to reflexively pull spend because of a billing change is confusing a liquidity constraint with an efficiency one. They are different problems and they have different fixes — a point we made the opposite way round in the EU €3 parcel duty post, where the cost really did land on unit margin and bids really did have to move.

Where it does bite: working capital, not efficiency

None of which means this is harmless. A liquidity constraint is a genuine constraint, and it binds hardest exactly where EU sellers already feel squeezed: inventory. Proceeds deduction means your disbursement arrives net of ad spend, so the cash available to place the next purchase order shrinks by whatever the ads cost. Scale spend aggressively in September and the reorder you needed for Q4 gets funded out of a smaller pot.

The trap is the feedback loop. Ads drive velocity, velocity drains inventory, and now the same ads also drain the cash that replaces that inventory. Run out of stock and the account does not merely stop selling — it loses rank it will pay to rebuild. Bidding as though inventory were infinite was always a mistake; it is now a more expensive one, which is the argument in inventory-aware bidding made in cash terms rather than stock terms.

There is a peak-season wrinkle too. Q4 is when spend is highest and when the gap between paying for ads and being paid for the resulting sales matters most. Whatever you decide about payment method, decide it before the ramp, not during it — the budget framework is the right place to model it.

The per-marketplace reading

Amazon's notice is account-level and does not carve out European marketplaces, and the announcement page is published for the European stores among others. But the effect is not evenly distributed across a multi-marketplace account, for a reason that has nothing to do with advertising: your proceeds and your ad spend are not in the same proportion in every store.

A seller with mature volume in Germany and a young, ad-heavy campaign in Italy has very different net disbursement dynamics in each. Netting ad spend against proceeds is comfortable where sales are deep and punishing where you are buying your way into a new market — which is precisely the phase where spend runs ahead of revenue by design. A blended, account-wide view of "can we afford this" will mislead you here in the same way a blended ACoS target does in France, Italy and Spain.

How to handle it

1. Check whether you are actually in scope before doing anything. Amazon said this applies only to directly-notified advertisers; the FAQ pointed sellers to Billing → Payment settings in the Ads Console to see which methods their account has. Most accounts were already on balance deduction.
2. If you are in scope, make the payment-method choice deliberately. Pay by Invoice preserves an end-of-month invoice with payment due 30 days later. Balance deduction is automatic and is what you now have by default if you never chose.
3. Do not touch your ACoS targets on account of this. Your break-even is a function of unit margin and price. If you want to revisit targets, revisit them for margin reasons — start from what a good ACoS actually is, not from a billing email.
4. Re-forecast cash, not efficiency. Model your next two disbursements net of ad spend and check the result against your reorder schedule. That is the number this change actually moved.
5. Put a floor under the interaction between spend and stock. If a burst of spend can starve a reorder, the throttle belongs in the bidding system, not in a monthly spreadsheet review.
6. Audit which campaigns were only ever solvent on float. Any campaign that needed thirty days of payment terms to look acceptable is a campaign whose real economics you have not seen yet.

Where a profit-first system fits

A change like this is a good test of whether your tooling reasons about the business or only about the dashboard. A bidder optimising to a fixed ACoS target has nothing to say about disbursement timing — it will keep bidding to the same number whether the cash arrives today or in sixty days. What you want instead is a system that already treats inventory and risk as first-class inputs to a bid rather than as context a human is expected to hold in their head.

That is how Mirox is built: sixteen specialised agents against one objective, with a Sentinel agent whose job is inventory- and risk-aware throttling, and seven safety layers — including a daily spend anomaly trip and per-marketplace thresholds — sitting between the AI and the budget. Thresholds and spend caps are tuned per marketplace rather than blended, which is the level at which a netting change like this is actually felt. And every bid carries its trace: the Bayesian conversion estimate, the alternatives ruled out, which gates were checked, exportable as CSV. If you want to see what the reasoning looks like against your own account before any money moves, Simulation Mode runs 30 days free on the real account with no card — the AI proposes, nothing executes until you flip Live Mode.

The one-line version

Proceeds deduction changes your cash conversion cycle, not your break-even ACoS — so re-forecast your disbursements and your reorder cover, protect the inventory loop, and leave your targets alone. The only campaigns that genuinely got worse on 1 August are the ones that were never profitable without the float.

See the reasoning behind a single bid, or read ACoS tells you efficiency, TACoS tells you the truth for the profit lens this all sits under.