Dark navy data visualization: a red −23% down arrow swinging up to a green +40% up arrow beneath the Amazon logo, illustrating the shift in AI-driven conversion between Prime Day 2025 and 2026
7 min read

From −23% to +40%: How AI Shopping Flipped Prime Day 2026

A year ago, sending shoppers to checkout via an AI assistant was a bad bet. On Prime Day 2025, traffic arriving from AI sources converted 23% worse than traffic from paid search, email and social. The robots browsed; they didn't buy.

On Prime Day 2026, that number didn't just recover — it inverted. AI-directed traffic converted 40% better than non-AI channels. In twelve months, the worst-converting source of traffic became the best. And it happened in the same window that Amazon quietly retired the Rufus name and rebuilt the experience as Alexa for Shopping.

−23%
how much worse AI-directed traffic converted vs. other channels on Prime Day 2025
+40%
how much better AI-directed traffic converted on Prime Day 2026
+89%
year-over-year growth in AI-referred retail traffic during the Prime Day 2026 window

Source: Adobe Analytics, reported by Forbes (Jun 27, 2026). See full sourcing below.

What the swing actually measures

The figures come from Adobe Analytics, which tracks visits and conversions across a large sample of U.S. retail sites and isolates traffic referred from generative-AI sources — ChatGPT, Gemini, AI-enabled browsers and on-site assistants like Amazon's. A year ago, those visitors behaved like researchers: high curiosity, low intent. They asked, compared, and left.

By mid-2026 the behaviour changed. The same class of traffic now converts 40% above the baseline of conventional paid channels, and there's simply far more of it: AI-referred visits to U.S. retail were up 235% year-over-year from January through May 2026, then jumped 89% again during the Prime Day window itself. Curiosity turned into carts.

In 2025, AI traffic was the worst-converting source on the site. In 2026, it's the best. The assistant stopped being a place people research and became a place people buy.

The year Rufus became Alexa for Shopping

The conversion swing didn't happen in a vacuum. Between the two Prime Days, Amazon overhauled the assistant that sits inside the buying flow.

DateWhat happened
Prime Day 2025Rufus is live for all U.S. shoppers. Adobe measures AI-directed traffic converting 23% worse than non-AI channels. Amazon later says Rufus helped 300M+ customers across 2025, with monthly active users up 115% YoY.
Feb 2026Amazon makes Alexa+ free for all U.S. Prime members, setting up the merge of conversational shopping and the broader assistant.
Jun 7, 2026Ahead of Prime Day, reports put Rufus past 250M users — and flag persistent readiness gaps in catalogs: empty Q&A, thin descriptions, unanswered reviews.
May 13, 2026Amazon discontinues the standalone Rufus chatbot and folds it into "Alexa for Shopping," embedded in the main search bar — no Prime membership or Echo device required.
Prime Day 2026 (Jun 23–26)$26.4B in U.S. online sales over four days (+9.3% YoY). AI-directed traffic now converts 40% better than other channels and is up 89% YoY.

Two things were happening at once. The technology got better at closing — more accurate answers, tighter product matching, fewer dead ends. And the distribution got wider — moving the assistant out of a separate chat panel and into the search bar every shopper already uses removed the friction that kept Rufus a research tool.

Shoppers were ready for it

Pre-event survey data lines up with the hard numbers. In a June 2026 survey of 1,500 U.S. shoppers, 58% said they planned to use — or were considering using — AI somewhere in their shopping journey, and half said they'd already tried Alexa for Shopping (the experience formerly known as Rufus).

What they use it for tells you why conversion improved. The top jobs are decision-support tasks that sit right next to the buy button:

65%
use AI shopping for price comparison
46%
use it to check price history
25%
use it to summarize reviews

These aren't idle queries. A shopper comparing prices and reading a review summary inside the assistant is already deep in a purchase decision. When the assistant answers well, the gap between "asking" and "buying" collapses — which is exactly what the +40% conversion figure captures.

The catch: half your shelf is invisible to the AI

Here's the number every brand should sit with. Adobe found that 46% of retail site content is not machine-readable — it can't be cleanly parsed by the AI agents now driving the best-converting traffic on the platform. Nearly half the digital shelf is illegible to the exact channel that's growing fastest and converting highest.

That's the whole game in one statistic. If AI-directed shoppers now convert 40% better, then any product the assistant can't confidently read or recommend is quietly losing the most valuable traffic on Amazon. The brands riding the swing aren't the ones with the biggest ad budgets — they're the ones whose listings, A+ content, reviews and Q&A give the assistant clear, structured evidence to act on.

What this means for brands

The bigger picture

For a decade, the Prime Day story was about price and position — the deepest discount, the best slot. The 2026 story is about readability. Amazon spent the year making its assistant trustworthy enough to buy from, and shoppers responded by routing their highest-intent moments through it.

The −23% to +40% swing isn't about AI getting hyped. It's about AI getting good enough to close — and rewarding the brands whose product story is clear enough for a machine to recommend out loud.

Fact-check & sourcing. The headline figures — AI-directed traffic converting 23% worse on Prime Day 2025 and 40% better on Prime Day 2026, plus the +89% Prime-Day and +235% Jan–May YoY AI-traffic growth, the 46% not-machine-readable figure, and $26.4B / +9.3% in sales — are from Adobe Analytics, reported by Forbes (Jun 27, 2026). The Rufus → Alexa for Shopping rebrand (May 13, 2026) is from CNBC and Amazon. The 300M customers / +115% MAU and Alexa+ free for Prime facts are from Stackline; the 58% / 50% / 65% / 46% / 25% shopper-behaviour figures are from a Stackline Consumer Pulse survey of 1,500 U.S. shoppers (Jun 2026); the 250M users milestone is from ChannelEngine reporting. Caveats: Adobe's +40%/+89% figures cover all generative-AI retail traffic in the U.S. (ChatGPT, Gemini, AI browsers and on-site assistants), not Alexa-for-Shopping alone — Amazon does not publish per-assistant transaction share. The 250M and 300M figures are different metrics (a pre-event snapshot vs. a cumulative-2025 total) and should not be summed. Stackline's survey figures are pre-event intent, not hard post-event conversion splits.

Sources

Share this article Share on LinkedIn Post on X

Is your shelf readable to the AI that's now buying?

WebQuest Digital helps brands stay visible, relevant and recommended as discovery shifts to AI. We turn what's changing into a clear consulting roadmap — so you're ready before your category is.

Talk to us about AI →