Latest / Retail Media Breakfast Club / Inside Amazon’s AI Shopping Funnel: Why Rufus Converts 3.5x Better
Transcript
- Rufuss: AI Shopping's Revenue Shift === [00:00:00] Kiri Masters: New research from Sensor Tower. Analyzing over a hundred thousand real Amazon shopping sessions during the 2025 holiday season shows that AI assisted shopping isn't a novelty anymore. [00:00:15] It's becoming a default consumer behavior. On Black Friday. Rufuss assisted sessions captured approximately 40% of total sessions, but then drove 66% of [00:00:30] purchases [00:00:30] sessions involving the assistant converted at three and a half times the rate of non rufuss sessions, a gap that held steady across October, November, and December. [00:00:40] Sensor towers. Data even shows that nearly all incremental [00:00:45] holiday growth on Amazon came from AI assisted shopping. This data matters because Amazon has been making very bold claims about rufuss that the industry has so far struggled to verify in 2025 [00:01:00] on the Q3 earnings call, [00:01:02] cEO. Andy Jassy reported that Rufuss had reached over 250 million active customers with monthly users up 140% year over year, and interactions up 210%. [00:01:15] And the headline number, rufuss was on track to deliver over a hundred billion dollars in incremental annualized sales. Those claims got people talking mostly because no one could figure out what those numbers [00:01:30] actually meant. [00:01:31] What defines a Rufuss user? Does sales mean Amazon's net revenue or gross merchandise value? What baseline are we comparing against? Without answers to these questions. Many industry observers [00:01:45] filed the numbers away as corporate PR and moved along. So sensor tower's independent research picks up where those questions left off and suggests that Jass E'S claims while, in typical Amazon style [00:01:59] where [00:02:00] Perhaps deliberately fuzzy may have directionally undersold what's actually happening. This podcast episode is based on an article that I wrote for my column at the drum called While Everyone Debates, a Agentic Shopping, Amazon's [00:02:15] Rufuss is Racking Up Sales, and that piece was published on January 29th, 2026. [00:02:20] [00:02:22] Kiri Masters: In a recent interview with the publication, the information, Andy Jassy doubled down on Rufuss as Amazon's answer to third party [00:02:30] AI shopping agents. Physical retail still has advantages in discovery. He acknowledged the ability to walk in, ask questions, refine those questions, and have somebody point you to different things. [00:02:42] Jesse argued that Rufuss addresses [00:02:45] this gap within the digital experience. With regards to competitors like open AI building shopping capabilities, he was diplomatic but pointed. He said, many third party agents don't have your buying history, your preferences [00:03:00] or accurate information about pricing and products. [00:03:03] He's not wrong as it pertains to onsite shopping assistance. But notice what Amazon did not do with rufuss. It didn't add recipe planning, lifestyle content, or [00:03:15] other features of questionable relevance to the core shopping mission. Amazon recognizes that people come to their platform to shop, and Rufus stays focused on that job. [00:03:26] This restraint sets rufuss apart compared to [00:03:30] some other retailer AI experiments I've observed. Instead of making assistance, genuinely useful alternatives to search unnecessary features, get bolted on. The result is bloated. Assisting [00:03:45] tools that don't actually even help us accomplish our primary goal. [00:03:49] Amazon avoided that trap by keeping Rufus tightly scoped to what customers actually need, which is finding products, comparing options, answering questions, [00:04:00] and building the confidence to purchase. So that is something that Amazon got completely right with Rufuss. Now, beyond the headline conversion numbers that I mentioned before, sensor Tower also identified [00:04:15] 10 distinct shopper behavior patterns that reveal when and why shoppers invoke the Rufuss assistant. [00:04:23] Now listening to a podcast, all of these numbers. Are gonna be hard to follow and are not gonna make a lot of sense. [00:04:30] So if you wanna dig into this, certainly check out my post, and the original research from Sensor Tower, [00:04:36] but I'll just tell you about one, which they call the cart reconsider a funnel journey. And [00:04:45] this is a small number of sessions that they observed only 7%, but it had a 50% conversion rate. In this shopper journey, the shopper has some items in the cart. But they hesitate, [00:05:00] they invoke rufuss to overcome objections and confirm the fit. [00:05:05] So this is a late funnel rescue sale, and how about that 50% conversion of that rescue. [00:05:14] But like I said, [00:05:15] there's nine other different funnel patterns that Sensor Tower observed. Things like early decider research, conversationalist the cart, reconsider that I mentioned the search assistant and the product [00:05:30] validator. The insight here is that AI enabled shopping is not one funnel. The early decider asks roofers a quick question at the start. [00:05:38] They get the confidence and then they convert at 85%. [00:05:43] What retailers can learn [00:05:45] from this is that just adding a chat bot and hoping for the best isn't the answer. Search assistant type behavior needs different support than the cart reconsiders. [00:06:00] Miracle Ads is the only retail media solution designed for both one P and three P Marketplace brands. Why does that matter? Marketplace sellers demand a seamless [00:06:15] advertiser experience that still offers full funnel ad formats, and retailers need a flexible solution that allows you to scale your media business. [00:06:25] Learn more@miracle.com. That's [00:06:30] M-I-R-A-K l.com. [00:06:32] [00:06:34] Kiri Masters: Now, here's a necessary caveat, and this came up in the comments when, um, sensor towers Ian Simpson first shared this data on [00:06:45] LinkedIn [00:06:45] and so here's where some methodological caution is warranted. Rufus usage could correlate with higher intent rather than cause higher conversion. Heavy Amazon shoppers, those [00:07:00] most likely to convert regardless, would also be most likely to use rufuss. [00:07:05] And so this research defines Rufus Sessions as any session with even brief interaction, which could inflate gaps, and the [00:07:15] methodology relies on modeled panel data that may be revised. [00:07:19] This data shows AI assisted journeys are tightly linked to high conversion and likely to improve outcomes for at least some shopper segments. Consider [00:07:30] a strong directional signal, not a causal Proofpoint. So the next KPI to watch. If Rufuss is genuinely improving product understanding and purchase confidence, not just accelerating decisions, shoppers would've made anyway, [00:07:45] we should see evidence in post-purchase outcomes, lower return rates, fewer cancellations [00:07:52] Especially in high research categories where the assistant guides complex decisions. [00:07:58] This is the measurement that will prove [00:08:00] whether AI actually makes shopping better. Or faster retailers and branches start demanding this data split AI assisted versus non assisted post-purchase outcomes. Until we see that analysis conversion [00:08:15] lift remains an incomplete picture. Now, what does all this mean for retail media? [00:08:21] If AI assistance become primary decision aids, the monetizable unit shifts from search result placement to [00:08:30] inclusion in the assistance answers. Amazon, of course, has already introduced sponsored ad units within Rufuss and I'll be reporting soon on how those early placements are performing for brand advertisers, but that's a topic for a [00:08:45] future column. [00:08:46] The more immediate point is this. We don't need fully autonomous shopping agents to see meaningful disruption. Rufuss isn't making purchases on anyone's behalf. It's not a true agent in that sense, [00:09:00] but it is already reshaping how a significant share of Amazon shopping sessions unfold and capturing a disproportionate share of actual purchases. [00:09:09] Amazon can afford to play the long game here. Other retailers watching these [00:09:15] numbers should recognize they don't have that luxury. Everyone's waiting for the agent. Future. Don't miss what's already changed. [00:09:23]