Amazon Ads 推出对话式智能体广告格式与品牌控制工具

The agentification of Amazon Ads is underway.

At its unBoxed ads industry event in San Francisco on Tuesday, Amazon announced a major ad tech overhaul, with its DSP turning into a default agentic chatbot experience and new AI optimization tools for branding and performance campaigns.

To mark the point, the Amazon Ads console itself is being rebranded to the Amazon Ads Agent.

But Amazon’s AI makeover also extends to ad formats and campaign types inspired by the conversational AI consumers are already using.

Joining the chat

One of those new formats is called “Branded Conversations,” an in-chatbot unit that debuted at unBoxed.

Amazon’s first chatbot ads, Sponsored Prompts, are akin to Sponsored Listings. They suggest a product in response to a search or recent purchase. Branded Conversations, by contrast, is an agentic version of the ads on product and purchase pages that promote items as “Frequently bought together” or that “Customers also bought.”

The idea is to create a natural bundle or an upsell opportunity.

Branded Conversations turn prompt-based shopper searches into “multi-turn shopping conversations to help shoppers find the right products informed by brand-provided inputs,” said Amjad Jbara, the director of applied science for sponsored products at Amazon Ads, which is corporate-speak for leading the charge on conversational ads products.

With the new Branded Conversations, advertisers can shuttle Amazon shoppers from that initial sponsored prompt to an immersive experience that features the brand’s colors, fonts, imagery and product descriptions. Branded Conversations will also suggest new prompts and follow-ups.

In Amazon’s unboxed demo, a shopper searching for the generic term “coffee maker” sees a sponsored prompt that says “Help me decide.” Although the prompt is labeled as sponsored Amazon generates the wording. Clicking on the initial prompt opens a Branded Conversations where the shopper can choose from series of suggested follow-up prompts, such as “maximum automation,” “works with capsules” and “fresh ground beans.”

A Branded Conversation could help shoppers zero in on the features that matter to them. A laptop’s super-high resolution and computing power might draw avid gamers, for example, while its patented cooling tech could appeal to an engineer running AI simulations. (Apparently patented cooling matters to engineers. This was Jbara’s hypothetical and presumably Amazon’s “applied science” guy would know).

These new formats are designed to move shoppers toward a decision, not just show them a list of options.

In Amazon’s unboxed demo, the coffee-maker search starts with a sponsored “Help me decide” prompt, not a buffet of options with one sponsored item at the top. The responses are whittled down earlier in the shopping search process, often even quickly identifying one particular product, which is sponsored.

Another variation of a sponsored prompt could be something like “Find the right fit for me,” Jbara suggested, which could easily lead to a Branded Conversation on behalf of a fashion or accessory brand.

A say in the conversation

But how much say do brands get in those conversations? Amazon has to strike a delicate balance.

Its answer is Conversational Briefs.

These are not underpants that talk to you, but rather a new way for advertisers to inform the AI recommendations that Amazon Ads makes about their products and how they want them presented to shoppers.

There are clear guardrails, however.

Say a brand wants Amazon’s AI to surface positive reviews that highlight some key product differentiator. Can it make those reviews more visible within AI search results or use them to inform a Branded Conversation?

In short, according to Jbara: “No.”

But although they can’t pick which reviews the AI surfaces, they can shape the conversation in other ways. For example, advertisers can use Conversational Briefs to tell Amazon which products are often purchased together. And advertisers can also use a free text field for their product descriptions, features and use cases to inform how Amazon’s multi-step Branded Conversations AI will proceed.

The full guidelines are available in the documentation, Jbara said. If an advertiser doesn’t abide by them or provides false or improper info via the Conversational Brief, he said Amazon “will give feedback” to that buyer through the console.

No stars guaranteed

Conversational Briefs give advertisers a way to tell Amazon’s AI about their products. But Amazon is also announcing another new ad product that focuses on another input: reviews.

Although advertisers can’t cherry pick the ones they want, reviews do inform Amazon’s agentic recommendations.

“Request Reviews for Sponsored Ads,” also announced on Tuesday, does exactly what it says. For products that have fewer than 1,000 reviews behind them – and especially for brands or products that are only just launching – the new ads will nudge people who’ve bought something to leave a star rating and review. There are no incentives allowed, such as samples or discounts. The ads will appear on the homepage, the order page and in email confirmations after a purchase.

Requested reviews are capped at 1,000, and run on a cost-per-click model, which means an advertiser might end up paying for someone who clicks and then doesn’t leave a review. Then again, advertisers have to pay even if they end up gathering lots of poor reviews.

So, the best advice here is to focus on product quality first. Just because you ask for a review, doesn’t mean it’s going to be good.