Subscribe
Share
Share
Embed
10 minutes of expert insights every weekday. Your morning ritual for staying ahead in retail media.
Amazon is playing agentic commerce chicken. Other retailers should not.
===
[00:00:00] Kiri Masters: Long time ad industry analyst Eric Sert threw down the gauntlet against AG Agentic shopping earlier this year. In his essay, ag Agentic Commerce is a Mirage. He laid out multiple reasons why AI [00:00:15] shopping agents won't reshape retail the way the hype suggests. Recently he revisited part of his argument on LinkedIn, talking about Amazon as a gatekeeper.
[00:00:27] The logic is that Amazon controls much [00:00:30] of us. E-commerce already has its own agent with Rufuss and has zero incentive to open up its walls to third party agents Thereby hampering its ad business. If Amazon won't play ball ag, agentic commerce [00:00:45] can't reach critical mass. Now I've learned a tremendous amount from Eric over the years, and I think his broader skepticism about agentic commerce hype is healthy.
[00:00:56] But this specific argument, which is Amazon as a [00:01:00] bellwether, is one that I wanna push back on,
[00:01:03] not because it's wrong in ~its own terms. Amazon probably won't open up to independent agents anytime soon, but Amazon won't cooperate. Isn't the same as agent commerce fails and the assumption that Amazon's no, just remove, sorry.~
[00:01:04] ~Not because it's wrong on ~its own terms. Amazon probably won't open up to independent agents anytime soon. But the assumption that Amazon's participation is necessary is where I think [00:01:15] the logic breaks down. Let's jump in.
[00:01:17]
[00:01:18] Kiri Masters: let's start with Eric's argument. I'm gonna lay it out for you. So the first is around Amazon's structural power. Amazon controls around 40% of US [00:01:30] E-commerce, GMV rufuss shows that Amazon is already an agentic winner. Rufuss is used by about 250 million customers. And apparently makes users 60% more likely to buy, [00:01:45] which a lot of people are skeptical of how that is calculated.
[00:01:49] But we'll leave that aside. Um, , but it is certainly widely regarded as one of the best, if not the best. Onsite [00:02:00] retailer agents. So they've done very well there. And another part of this structural power argument is that Amazon's advertising and its first party data business gives it a huge [00:02:15] reason to block independent agents.
[00:02:17] This is a very strong argument, which is Amazon derives huge value from its ad business. So seeding discovery and transaction to third party agents [00:02:30] weakens that moat. So this argument is cemented around a lack of incentive for Amazon to fling open its doors, which makes.
[00:02:41] Logical sense. Eric goes on to argue that [00:02:45] independent agents, such as a chat, GBT, instant checkout can't scale without Amazon. ~paraphrasing from his post, he says that independent agents.~
[00:02:51] ~He says, ~paraphrasing from his post, he argues that in order for ag agentic commerce to really take off, independent agents must get meaningful [00:03:00] traction without Amazon. Then These actions must divert enough transactions away from Amazon to really matter Then.
[00:03:09] some other retailer that does integrate with agents must grow operations enough to threaten [00:03:15] Amazon. And finally, that threat must be so acute that Amazon is forced to subordinate its first party customer relationship and $60 billion ad business. So this is a strong logic chain, but. It also assumes an [00:03:30] either or outcome that Amazon either cooperates or that agents are irrelevant.
[00:03:36] And it assumes that all retailers share Amazon's same incentive structure. So here is my counterpoint, and it is that retailers [00:03:45] shouldn't be tempted into a game of chicken with Amazon. ~that Amazon is the only car on the road, and if Amazon won't swerve towards agent interoperability, the argument goes, then the agents will crash and burn.~
[00:03:50] ~But.~
[00:03:50] ~Eric's argument. Eric's logic assumes that Amazon is the only car on Eric's. Eric's logic assumes that Amazon is the only car on the road. Eric's argument is ~Eric's argument is that Amazon is the only car on the road, and if Amazon won't swerve, then agents crash and burn.
[00:03:58] But I don't think that's how this [00:04:00] game works. So part one, agents as the new demand generators, not just channels. Today, AI isn't massively stealing traffic from Google search. As a lot of research has shown, it's more [00:04:15] like an extra layer.
[00:04:17] So first I ask AI for help with things like meal planning or gift ideas, but I still go to retailer dot com's and apps to make the purchase once I've [00:04:30] decided what I wanted. But as agents get better at going from, help me to think to, okay, now place the order for me. More and more actual purchase decisions will move into that [00:04:45] agent layer.
[00:04:46] So think of that as old demand versus new demand. The old demand is that most shopping journeys have for a long time started in Amazon search bar. That's the world [00:05:00] where Amazon owns 40%. Of us e-commerce, that is 40% of legacy e-commerce. But also we need to think about the new demand.
[00:05:11] When journeys start in an agent instead, and [00:05:15] the agent decides where to send the order. As more retailers, payment providers, and commerce platforms plugged directly into agents, a growing share of future demand will be created and routed [00:05:30] outside. Amazon's walls, even if Google search and Amazon search volumes stay high. Miracle Ads is the Ad Tech [00:05:45] solution trusted by Rakuten and over 50 global enterprise retailers. That's because Miracle Ads was built with both three P Marketplace sellers and one P suppliers in mind. Both advertiser audiences [00:06:00] demand a seamless advertising journey from onboarding to reporting.
[00:06:04] You can offer everything from sponsored products to video ads all in one solution. Learn more@miracle.com. That's M-I-R-A-K [00:06:15] l.com.
[00:06:15]
[00:06:17] Kiri Masters: Part two, Amazon lags on grocery. And here's where Amazon's 40% share of e-commerce becomes less relevant than it looks, and it is that the company continues to falter [00:06:30] in a key category, which is grocery so who does hold the keys to the kingdom in online grocery? It is in fact Walmart e-marketer projects that Walmart will hold about 31.6% of US [00:06:45] grocery e-commerce sales this year ahead of Amazon at 22.6%, and Kroger at 8.6%.
[00:06:54] This matters because grocery is a prime candidate for ag agentic shopping. [00:07:00] It's not one of these like fun shopping experiences like other categories like beauty or fashion, where shoppers enjoy the the thrill of the chase, the popularity of Instacart buy online, pickup [00:07:15] in store, and other home delivery grocery methodologies.
[00:07:20] The popularity of those suggest that many consumers are okay with allowing someone else to pick out their avocados. And finally, a lot of purchases [00:07:30] in this category are either replenishment or could be integrated into meal planning systems. It is this category of shopping that most consumers when polled say they would like to [00:07:45] theoretically offload to an agent.
[00:07:47] So if Walmart and other grocers are agent forward and Amazon is lagging behind, then Amazon's dominant share of total e-commerce doesn't translate into veto [00:08:00] power over the most agentic prone category.
[00:08:04] Part three, maybe Amazon sees the writing on the wall. Amazon has an incredible ads business and they stand to lose that golden goose if they [00:08:15] concede their users' shopping journeys.
[00:08:18] But Amazon is recently behaving as if onsite lower funnel sponsored product ad revenue is under threat with or without agents, and it has been rapidly diversifying into [00:08:30] offsite CTV and non-endemic ad units. I analyzed the headline announcements that Amazon made at their unboxed event this year, and I found that a significant percentage of new features and [00:08:45] upgrades have nothing to do with sponsored product ads or onsite bottom of funnel ad units. Let's also consider that Rufuss itself is an platform agent that by design reduces the [00:09:00] number of old school sponsored product ad tiles. That are needed per transaction.
[00:09:04] Even if there are some ad units inside of a Rufus conversation, it's gonna be far, far fewer than a traditional search result page. From these [00:09:15] actions, it looks like Amazon already sees the writing on the wall, that lower funnel sponsor product ads are already under threat. ~So the idea that Amazon will permanently refuse a gen, they are already diversifying.~
[00:09:24] ~Into connected TV and trying to tap into non-endemic brand budgets and national advertising budgets. ~So the idea that Amazon will permanently refuse agen access just to protect sponsored [00:09:30] product ad revenue is a short horizon assumption. Part four, interoperability economics and why many retailers should want agents to work.
[00:09:42] Travis Klinger, who is the Chief [00:09:45] Connectivity and Ecosystem Officer at LiveRamp gave me an interesting perspective. In the long run, interoperability tends to win and history backs him up.
[00:09:56] In the early internet companies like America [00:10:00] Online and CompuServe tried to keep users inside closed portals with curated content and shopping. Their whole pitch was the internet is scary and confusing. Stay here where it's safe and curated, but the open web built [00:10:15] on shared standards, one.
[00:10:17] Users preferred a world where any site could link to any other site, and no single portal controlled everything. Amazon's Rufuss is a portal model. Rufuss says, [00:10:30] we'll be your internet. Just stay here today. LLMs and open protocols like a CP are the open web model. They're saying we'll connect you to many merchants and publishers through [00:10:45] one interface.
[00:10:46] ~The A CP reality is already,~
[00:10:46] ~already LLMs are wiring directly into a.~
[00:10:46] LLMs are already wiring directly into a long tail of merchants and platforms, not just one or two mega retailers. And this is why retailers should want Agentic Commerce to [00:11:00] work amazon is choosing for now closed rufuss, whereas other retailers who are really in need of incremental budgets and differentiation versus Amazon.
[00:11:11] Now they have the opposite incentive, which is to be [00:11:15] extremely agent friendly.
[00:11:16] And part five, partial adoption is enough. Agentic commerce doesn't need to absorb all or even most online spending to matter if 10 or 20% of spend in a few key categories [00:11:30] go agentic. It's already a massive shift in media planning. We don't need to wait on a sci-fi future to argue that retailers should be thinking about this.
[00:11:39] Now, I like to say that brands are fair [00:11:45] weather friends, their ad budgets drift to wherever the consumer is hanging out and making purchase decision. For the past few years, that place has been retail media, but a shift in just some shopping missions [00:12:00] for some groups of customers and a brand's wandering eye is awakened.
[00:12:06] So who swerves first? Eric's Amazon argument isn't wrong on its own terms. Amazon probably won't open its gates [00:12:15] to independent agents. They have every reason to keep Rufuss as a walled garden and and they have the market power to make that bet. But here's where the chicken game gets interesting.
[00:12:26] Amazon isn't the only car on the road. [00:12:30] Walmart, Kroger.
[00:12:31] Target, Instacart, they're all approaching the intersection from the opposite direction, and for them the calculus is inverted. They don't have Amazon's ads cushion. Their crash isn't a head on collision [00:12:45] with agents. It's irrelevance. It's watching a new demand channel form and realizing that they weren't invited.
[00:12:52] So while everyone waits to see if Amazon blinks, the ecosystem is wiring itself together around the [00:13:00] holdouts.
[00:13:00] Eric is right that Agen Commerce can't succeed by going through Amazon, but it doesn't have to. It just has to go around them and capture enough of the new demand to make brands pay attention.
[00:13:14]