Amazon just enrolled your campaigns into agentic shopping. You did not opt in.

On 11 June 2026 Amazon Ads told advertisers their existing campaigns are already running inside the AI conversation — nobody had to opt in, and most sellers do not know it happened. In a first-party post titled "What agentic shopping means for advertisers," Amazon confirmed that ordinary Sponsored Products and Sponsored Brands campaigns are automatically eligible to serve inside Alexa for Shopping, with no additional setup. That is good news for reach and a quiet problem for anyone who judges spend on a keyword report. Here is what actually changed, and the discipline it demands.

What Amazon actually announced

The backdrop first. In May 2026 Amazon merged its Rufus shopping assistant — used by over 300 million customers in 2025, by Amazon's own count — into Alexa for Shopping, a single AI layer that now answers questions in the main search bar, builds comparisons, and can complete a purchase inside the conversation. We covered the discovery side of that shift in how Alexa for Shopping is rewriting Amazon PPC. The June announcement is the advertising sequel: Amazon describing how paid placements live inside that conversation. Three formats matter for sellers:

• Sponsored ads inside Alexa for Shopping. Your existing Sponsored Products and Sponsored Brands campaigns are, in Amazon's words, automatically eligible to serve within the assistant's conversations — no new campaign, no new opt-in.
• Sponsored Products and Sponsored Brands prompts. AI-powered conversational ads that appear in search results and on product pages and open a dialogue with the assistant about your product. Existing campaigns are eligible to be optimised as a prompt.
• Alexa+ Conversational Ads and Prime Video Sponsored Tiles. Ad formats on Alexa-enabled devices where, again, advertisers with sponsored ads campaigns are automatically enrolled.

Notice the repeated phrase. The connective tissue across all three is automatic eligibility. Amazon opened new ad surfaces and enrolled the campaigns you already run, rather than asking you to build for them.

The contrarian read: a new surface is a new place to leak

The industry reaction has been to treat this as pure upside — more impressions, a fresh placement, an AI that hands the customer the answer. It is upside. It is also a structural shift in where your money goes and how visible that spend is to you.

A keyword report tells you which typed query spent your budget. An agentic surface does not. When the assistant recommends one of three competing products inside a natural-language conversation, the impression and the spend can sit outside the exact-match keyword logic you have spent years pruning. The placement is new; the discipline that governs it is not. Break-even per click still decides whether the click was worth buying.

This is the same blind spot that makes broad match and auto campaigns leak — spend lands on intent you did not explicitly price. We mapped that mechanism in where 30 to 40 percent of ad spend goes and in auto vs manual campaigns. The agentic surface adds a third doorway into the same room: relevant-looking traffic that an aggregate dashboard reports as a win, with no per-query line for you to negate or cap.

Why measurement is the real question

Amazon frames the agentic experience around "closed-loop measurement" — impressions tracked all the way to conversions. That is genuinely valuable, and it is also Amazon's loop, scored by Amazon, on surfaces Amazon controls. The same incentive we have written about elsewhere applies: the platform earns when you spend, so a tool or a placement that grows spend is not a neutral party to whether that spend cleared profit. We made the version of this argument about Amazon's own campaign assistant in what Ads Agent does and does not do — it is useful, and it is not profit-aware.

The point is not to distrust the surface. It is to keep an independent profit ledger that does not move when Amazon adds a new place to bid. Whether a click came from a typed keyword, a broad-match guess, or an assistant's recommendation, the question is identical: did the contribution margin on the sale exceed the cost of the click? A surface cannot answer that. Your margin model has to.

What does not change

It is easy to over-rotate on a platform announcement. Most of the fundamentals are untouched. Sponsored Products still wins high-intent placements. Break-even CPC still decides whether a click is worth it — the CPC benchmarks hold. Inventory still constrains how hard you should push, especially as Prime Day demand spikes; the logic in inventory-aware bidding applies on the conversational surface exactly as it does on search. And profit, not a proxy, is still the only honest target — see ACoS tells you efficiency, TACoS tells you the truth. The conversation is a new container. The economics inside it are the same.

What EU sellers should do about it

1. Confirm you are enrolled, then decide if you want to be. Because eligibility is automatic, the first job is awareness: check that your campaigns are serving on the new surfaces and treat that as a live decision, not a default you sleepwalked into.
2. Anchor every campaign to break-even per ASIN. An agentic impression you cannot see in a keyword report is only safe if the bid behind it was already priced against margin after COGS, fees, and returns. Set the ceiling once and it governs every surface.
3. Write listings that read as a clear answer. The assistant surfaces the product that best matches a stated need. Use-case clarity, complete information, and tight category relevance are now budget lines, not content niceties — and machine-translated copy reads as machine-translated, which compounds the cross-border penalty in EU-native is not a translation layer.
4. Keep an independent decision trace. If the platform owns the measurement loop, you need your own record of why each bid was set and what it was worth. That is the only way to defend spend on a surface you cannot fully inspect.

The one-line version

Amazon did not ask whether you wanted to advertise inside an AI conversation — your campaigns are already eligible, and the new surface adds reach the same way it adds a place for spend to leak unseen. The defence is unchanged and unglamorous: price every bid against margin, judge the account on profit rather than the loop Amazon scores for you, and keep a record of why each decision was made — so a placement you cannot watch is still one you can account for.

A profit model that scores break-even and semantic relevance on every bid — across search and the agentic surface alike — and writes the reason into an exportable trace is built for exactly this kind of shift. See what a single decision trace contains, or read why we will never charge a percentage of your ad spend.