Second Chance Deal Days is not your event — but for ten days your ACoS pays for it.

Second Chance Deal Days is not your event, you cannot enter it, and for ten days in September it will quietly raise your ACoS anyway — because what changes is not the auction, it is the shelf your clicks land on. On 28 August 2026 Amazon confirmed that Second Chance Deal Days runs from 1 to 10 September in the United Kingdom, Germany, France, Italy and Spain — the exact five marketplaces most EU sellers actually live in. Discounted returned and refurbished stock, priced against the RRP of the equivalent new product. Your bids do not change. Your competitive set does.

What Amazon actually announced

The announcement is written as a consumer shopping notice, which is why most sellers will scroll past it. The numbers inside it are not consumer numbers.

Amazon states it lists over 60 million returned and refurbished items across nearly all product categories in Europe. Outside any promotion, second-hand purchases already save European customers an average of 20% against the RRP of the equivalent new product. During Second Chance Deal Days that goes to savings of up to 50% against new-product RRP, on tens of thousands of items across personal tech, kitchen, gaming and home improvement. Named brands in the European window include Apple, Sony, Samsung, Beats and LEGO. For the first time, UK and German shoppers also get authenticated pre-owned bags and accessories from the Amazon Luxury seller Luxclusif at a further 15% off already-discounted prices.

The scale underneath: Amazon put European second-hand sales above €2 billion a year, €380 million saved by customers buying second-hand in the prior year, and a returns-operations investment in Europe of more than $4 billion. This is not a clearance bin. It is a parallel catalogue with its own supply chain, and Amazon has just spent ten days pointing shopper attention at it.

Two scheduling details matter and almost everyone will get them wrong. The European window is 1–10 September. The United States runs something different — Second Chance Savings folded into Labor Day Deals from 28 August to 7 September, at up to 30% off already-discounted second-hand prices, which is a completely different discount mechanic. Mexico and Brazil run 31 August to 6 September at their own ceilings. If your agency or your playbook is American, the dates and the depth it hands you are both wrong for your account.

The contrarian part: this is a conversion event, not an auction event

Every seller reflex around a deal event is an auction reflex: CPCs will spike, raise bids to hold placement, or cut bids to survive it. That reflex is misapplied here, because Amazon has not announced a traffic event on your keywords. It has announced a price event on the shelf those keywords lead to.

Think about what a shopper sees after clicking your Sponsored Products ad during that window. Same query, same placement, same ad. But the consideration set now contains a ‘good condition’ unit of a comparable product at up to half the price of new. Your impressions hold. Your click-through rate holds. Your conversion rate does not. And the metric you monitor is downstream of all three.

ACoS = CPC ÷ (CVR × price). Every term except CVR is under your control. Hold CPC and price constant and let conversion rate fall by 15%, and ACoS rises by 1 ÷ 0.85 — about 17.6% — with no change to your bids, your keywords, or your creative. The bid did not get worse. The shelf did.

That identity is the whole argument. A CVR-driven ACoS rise and a CPC-driven ACoS rise look identical in a weekly report and require opposite responses. If competitors outbid you, paying more may still be rational. If shoppers are converting elsewhere at half the price, paying more is buying clicks into a comparison you lose. Anyone who reacts to a September ACoS spike with a bid change before decomposing it is guessing — which is the same failure mode we described in the dynamic bidding post, where the visible setting gets blamed for what the layer underneath is doing.

Where the exposure actually sits

A 60-million-item catalogue spread across nearly all categories does not mean uniform exposure. Second- hand inventory is generated by returns, and returns concentrate. Amazon names electronics, home and kitchen as the top-selling second-hand categories, with vacuums, espresso makers, phones, cameras and tablets as the top-selling products. If you sell a €200 kitchen appliance in Germany, you are exposed. If you sell consumables or a private-label product with no meaningful refurbished equivalent, you are close to unexposed and should not touch anything.

The market split is uneven too, and in a way that punishes blended management. Luxclusif is UK and Germany only. Category depth in returned stock is not the same in Italy or Spain as in Germany, because the underlying retail volume is not the same. An account-level ACoS target across five marketplaces will average a real problem in one store against no problem in three others and produce a number that tells you to do nothing anywhere — the exact blending mistake we set out in France, Italy and Spain.

There is one more mechanism worth naming, because it sits directly in the path your ad spend travels. Amazon flags products with high return rates to shoppers, prompting them to check descriptions and reviews before buying. That flag operates between a paid click and a conversion, and it is indifferent to your bid, your budget and your creative. In a ten-day window built entirely around returned goods, the returns conversation is the one shoppers are already having.

What to do about it

1. Decide whether you are exposed before you decide anything else. Search your top ten ASINs’ main keywords in each marketplace and look for used, renewed or ‘good condition’ offers on comparable products. No refurbished shelf, no action — close the tab.
2. Take a clean pre-event baseline now, per marketplace: CPC, CTR, CVR and ACoS for the fortnight to 31 August. Without it you cannot tell in October whether September was a price problem or an auction problem, and you will spend the winter guessing.
3. Decompose daily during 1–10 September rather than reading ACoS. If CPC is flat and CVR is falling, this is the event, not your campaigns. If CPC is climbing too, you have a separate auction issue — check it against your own CPC benchmarks before attributing it here.
4. Do not cut bids account-wide. Cut where CVR fell, in the marketplace and the ad group where it fell. A blanket cut surrenders rank on the queries that were never affected, and you pay to rebuild that rank afterwards.
5. Where you are exposed, defend on the detail page rather than in the auction. Warranty, condition, full manufacturer support and bundle contents are the arguments a refurbished unit at half price cannot make. That is a listing fix, and it is cheaper than a bid war you cannot win on price.
6. Watch your branded and defensive placements hardest. Your own returned units can appear alongside your new ones, and paying top-of-search rates to send a shopper to a page where your cheaper used listing wins the sale is the most expensive version of this problem.
7. Ring-fence the window in your reporting. Ten days of depressed CVR will contaminate any 30-day lookback you use for Q4 planning, and Prime Big Deal Days sits close behind it. Annotate the dates before the data goes into your Q4 budget model.

Where a profit-first system fits

A short, sharp shift in conversion rate is the hardest thing for a naive bidder to handle well. Optimise to a fixed ACoS target and a ten-day CVR dip reads as failing keywords: the system cuts bids into the event, loses placement, and is still cutting when the shelf returns to normal on 11 September. Then it spends weeks paying to climb back.

Mirox is built so that the conversion estimate is an explicit, inspectable input rather than something buried in a target. Every bid carries its Bayesian conversion estimate, the relevance score, the agent that proposed it, the alternatives ruled out and which of the seven safety gates were checked — and the whole trace exports as CSV, so decomposing a September ACoS move into price and conversion is reading a record, not reconstructing one. Thresholds and spend caps are tuned per marketplace rather than blended, which is the level at which a UK-and-Germany-only wrinkle like Luxclusif is even visible. Sixteen specialised agents work one objective, with Sentinel handling inventory- and risk-aware throttling and seven safety layers — including a ROAS circuit breaker and a daily spend anomaly trip — between the AI and the budget. If you want to watch that reasoning against your own account before a cent moves, Simulation Mode runs 30 days free on the real Amazon Ads account, no card: the AI proposes, nothing executes until you flip Live Mode.

The one-line version

From 1 to 10 September, up to 50% off 60 million refurbished units changes your shoppers’ comparison set in five EU marketplaces, not your auction — so baseline your CVR now, decompose ACoS daily instead of reading it, defend on the detail page, and cut bids only where conversion actually fell.

Common questions

Can sellers take part in Second Chance Deal Days?

No. Amazon runs the event on its own returned and refurbished inventory, so third-party sellers cannot enrol products. There is no promotion to opt into and no deal budget to plan — only a change in what shoppers see beside your listing.

Should I lower my Amazon PPC bids during Second Chance Deal Days?

Only where conversion rate actually fell. An account-wide cut treats a ten-day conversion event as a keyword-quality problem: you lose placement going in, and you are still paying to climb back after the shelf returns to normal on 11 September.

Which marketplaces does Second Chance Deal Days affect?

The United Kingdom, Germany, France, Italy and Spain, from 1 to 10 September 2026. Read it per marketplace: a UK- and Germany-only effect disappears the moment you look at blended account numbers.

How do I tell an event conversion dip from a real bidding problem?

Decompose ACoS instead of reading it. ACoS moves when CPC rises or when conversion rate falls, so baseline both before 1 September and track them separately each day. Flat CPC with falling CVR means the auction is unchanged and the shelf is the cause.

Get a free waste audit of your own account to see where the September spend actually went, see the reasoning behind a single bid, or read the Prime Day 2026 PPC checklist for how to run the event-window discipline this all depends on.