Alexa+ Agentic Ads: Amazon Turns the conversation into Checkout

Amazon Alexa+ Agentic Ads debuted at Cannes Lions 2026, and the format does something no ad unit has done before: it lets a shopper go from seeing an ad to completing a purchase entirely inside a spoken conversation — no click, no QR code, no second screen. It is running in beta on Echo Show devices with a closed list of launch partners, and Amazon is framing it as the first agentic ad format live at scale. The pitch is bold; the caveats are bigger.

This is not a how-to for buying Alexa ads — you cannot yet. Pricing is undisclosed and the beta partner list is closed. What changed at Cannes is structural. The click, the unit digital advertising was built to measure, disappears. When the conversation is the conversion path, transaction completion replaces click-through as the metric that matters, and the brand with the strongest organic presence inside Alexa’s recommendations may not need to pay at all. That is a different game than the one most media plans are built for.

This analysis covers what Amazon actually launched versus what is still only announced, how the voice-commerce funnel collapses, why Amazon’s roughly $68–70 billion ad business and transaction data give it a structural edge, the privacy contradiction sitting under the launch, how it maps against the other agentic ad formats unveiled at Cannes, and what a marketer should actually do about it. Every figure is sourced and dated; where a claim is vendor-stated or a market estimate spans a wide range, we say so plainly.

At Cannes Lions 2026, Amazon unveiled Alexa+ Agentic Ads and put it into beta on Echo Show devices. Amazon describes it as the first ad format that takes a customer from seeing an ad to completing a purchase entirely within an organic conversation without ever having to leave the ad — a vendor framing worth quoting precisely, because the “entirely within” part is the whole novelty. The shopper never lands on a website, never opens an app, never scans a code. The ad and the store are the same surface.

Two beta use cases anchor the launch. Papa Johns brings food ordering by voice. The Orchard — Sony Music Entertainment’s independent distribution subsidiary — brings concert-ticket purchasing. Worth keeping straight: The Orchard is the label and distribution layer, while individual artists Beck, Jill Scott, and Omar Courtz are separate artist partners whose fans can buy concert tickets via voice through Ticketmaster, with tickets delivered to the customer’s Ticketmaster account. The label layer and the artist layer are distinct partnerships, even where they overlap.

The mechanics are simple to describe and consequential to plan around. A customer sees an ad on the Echo Show screen, starts a voice conversation with Alexa, refines the options out loud — preferred seating, number of tickets, the usual toppings — and completes the purchase without leaving the Alexa interface. No QR code. No phone. No handoff to a browser. Amazon says the system does not require “rigid phrases” to know when to complete an action; it reads subtler conversational cues through the underlying large language model.

What that does to the funnel is the part marketers should sit with. Every stage that traditional digital advertising instruments — the click, the landing page, the form, the cart — either collapses into the conversation or disappears. The table below maps the voice-commerce funnel against its traditional digital equivalent and names the friction each stage removes.

Read down the middle column and the pattern is unmistakable: the entire apparatus of click-and-landing-page advertising is what gets removed. That is genuinely useful for the shopper, and genuinely disorienting for the marketer, because the measurement model that sits on top of clicks goes with it. The conversation is more convenient and less legible at the same time — which is exactly why the question of what replaces the click is not a footnote but the central planning problem.

Amazon is not the only company chasing agentic ads, but it arrives with assets the others do not have. Its advertising division generated approximately $68–70 billion in 2025, making it the third-largest digital ad platform behind Google and Meta — a reported figure, and one Amazon’s own executive referenced loosely as a “$70 billion engine” at Cannes, so we cite the range rather than a single precise number. Layered on top is years of closed-loop transaction data and a demand-side platform most rivals cannot match.

The shopping intelligence is already in place. The Rufus AI shopping assistant — which logged 300 million-plus customer interactions in 2025 (interactions, not unique users) — was folded into Alexa+ to create “Alexa for Shopping,” spanning the Amazon Shopping app, the website, and Echo Show devices. Amazon’s own programmatic signal points the same direction: it says 20% of shoppers who interact with a Sponsored Brands conversational prompt continue the conversation, and that adding those prompts produces a 6% increase in conversions. Both are vendor-stated, and we treat them as directional rather than audited. For the wider landscape of AI shopping assistants like Rufus, our comparison guide maps the field.

Most coverage of this launch repeats Amazon’s privacy framing uncritically, and it is worth slowing down on. Amazon’s VP of devices content and advertising, Charlotte Maines, described a “firewall” between conversational data used for personalization and the ad-tech stack, leaning on Amazon’s decades of building products that are, in her telling, both personalized and safe for customer privacy. That is a vendor claim, not an independently verified fact, and it should be read as one.

It also runs into a verifiable policy change. With the Alexa+ rollout, Amazon eliminated the “Do Not Send Voice Recordings” option that had previously allowed local-only processing on some devices; reporting indicates all queries now route to Amazon’s cloud. That is a tightening of data collection, and it cuts against the consumer-friendly “firewall” framing. To be precise: it is a policy change Amazon disclosed, not a proven violation — but it is the kind of fact a marketer thinking about consumer trust should weigh alongside the reassurance. The contrast at Cannes is instructive: OpenAI, entering advertising the same week, pledged not to share conversational information with advertisers at all. Our read on OpenAI’s Cannes advertising debut walks through that opposite bet.

Agentic advertising flooded Cannes Lions 2026, and Amazon’s launch is best read against the cohort it arrived with. Fox Advertising launched an end-to-end agentic platform. TikTok unveiled Symphony Agent with Dentsu. Yahoo’s DSP introduced its Agent Network. Publicis acquired LiveRamp. OpenAI declared itself an advertising business. The table below maps five of these side by side on the dimensions that distinguish them — no single trade outlet has put them in one frame with their privacy stances attached.

Set side by side, the strategic fork is clear. Amazon’s differentiator is the transaction layer — it owns the purchase, the measurement, and the demand-side platform that ties them together, which is precisely the advantage its executives lean on. OpenAI is betting the opposite way on data, making a no-sharing pledge the point of difference. Fox, TikTok, and Yahoo are largely at the announcement stage, building toward audience, creative, and programmatic agents respectively. For how this agent layer is reshaping the buying side of ad tech, see our piece on the emerging agentic ad-buying layer.

Voice commerce is the headline market for this format, and its size is genuinely uncertain — not because the data is bad, but because the definitions diverge. For 2026, estimates run from roughly $22.4 billion (US only, narrow definition) to about $77 billion (global, broad definition). Anyone citing a single figure is quietly picking a methodology; the honest move is to cite the range and note that voice shopping is projected to drive around 30% of ecommerce revenue by 2030. The opportunity is real; the precision is not there yet.

The programmatic backdrop is firmer. US programmatic ad spending is expected to exceed $220 billion in 2026, with walled gardens — Amazon, Google, and Meta — projected to capture 80.4% of total programmatic spend. And the direction of travel toward autonomy is steep: PubMatic estimates 25% of ad buys will be executed autonomously by AI agents by 2028, rising above 50% by 2030. Against a backdrop where roughly 34% of consumers already use AI platforms daily, an agentic ad format from the third-largest walled garden is not a side experiment — it is a positioning move for where spend is heading.

The sharpest unanswered question is one almost no coverage names cleanly: the organic tax. If Alexa recommends products on organic relevance first and paid placement second, brands without a strong organic presence in Alexa’s ecosystem will pay to be inserted into conversations where they have not earned a slot. It mirrors the “pay to play” dynamic that took over Amazon search after 2018 — except now it is happening inside a conversation that remembers the customer. The advertisers who benefit most may be the ones who least need to buy; the ones who most need the boost may find the paid placement does the least.

So what should a marketer do while the format is in beta and the economics are undefined? Not nothing, and not a budget commitment. The defensible posture depends on where you sit — your organic strength, your category’s fit for voice, how performance-led your budget is, and how sensitive your brand is on data. The matrix below sorts the realistic moves.

Standing up that kind of readiness — earning organic presence, sizing a contained test, and keeping the measurement gap a known limitation rather than a post-mortem surprise — is exactly the work our paid media management and ecommerce growth engagements are built around. The goal is not to chase the loudest Cannes headline; it is to be ready for a genuinely new purchase surface without mistaking a beta for a media plan. For the wider context, our comparison of agentic commerce platforms maps where voice fits among the rest.