For a decade, the formula for Amazon visibility was predictable enough to build a business around: rank organically, supplement with pay-per-click, keep the listing clean, and manage your bids. The channel was stable. The only real variables were budget and execution, and a competent operator could learn the rules and win by working them harder than the next seller.
That stability is gone. The discovery layer of ecommerce — the part that decides which products a shopper ever sees — is being rebuilt underneath the brands that depend on it, and the pace of the rebuild is no longer subtle.
The traffic signal
AI-referred traffic to US retailers grew 393% year-over-year in the first quarter of 2026, with March alone up 269%, according to Adobe Analytics. A growth rate like that does not describe a fringe referral stream. It describes a channel moving from the margins to the mainstream inside a single year — the kind of curve that, in retrospect, marks the moment a behavior became normal.
And the traffic performs. By March 2026, AI-referred visits were converting 42% better than non-AI traffic — a record in Adobe’s dataset — with those shoppers spending 48% more time on the page and browsing 13% more pages per visit. The shopper arriving via an AI agent is not a lower-intent browser killing time. They are further down the funnel before they ever land, because the agent has already done the filtering that a human would otherwise do across a dozen open tabs.
The behavior underneath the numbers
Consumer adoption explains the surge. Adobe found that 39% of US consumers have used AI for online shopping, and 85% of that group said it improved their experience. That satisfaction figure is the one that should worry any brand relying on the old playbook, because satisfaction is what turns a one-time experiment into a default habit. During Cyber Week 2025, Salesforce reported that AI and agents drove $67 billion in sales and influenced 20% of all global orders — and that retailers using its Agentforce agents grew sales 32% faster than those without.
The shopper’s behavior is changing in a way that quietly breaks the old playbook. They no longer type “wireless earbuds under $80” into a search bar and scroll a results page where a seller can buy placement at the top. They ask an agent to find the best option for their stated need. The agent assembles a shortlist — often three or four products, not a page of twenty. The shortlist determines the sale, and there is no sponsored slot at the top of it that a brand can simply purchase its way onto.
Amazon is building the agent, not just watching it
Amazon is not a bystander here. On its Q1 2026 earnings call, the company said monthly active users for its Rufus shopping assistant were up more than 115% year-over-year and engagement was up nearly 400%, with nearly 20% of shoppers who interact with a brand prompt continuing the conversation about that brand. That last figure matters: it means the assistant is not just answering questions, it is steering brand consideration inside the app.
When the platform’s own discovery agent engages shoppers that deeply, the incentive to route more of them through it — and fewer through the traditional search results where sellers bid — only grows. Every query resolved by Rufus is a query that did not scroll a sponsored results page. The paid-search real estate that sellers have optimized for a decade is being quietly reallocated to a surface they cannot buy their way onto, and the reallocation is happening on Amazon’s timetable, not the seller’s.
What actually optimizes for an agent
Optimizing for AI-mediated discovery is not the same discipline as SEO or PPC, and treating it as a new keyword game is the mistake that will cost brands the most. Agents pull structured product data, review signals, pricing consistency, and listing completeness — and weight them differently than a keyword-driven results page. They reward specificity: accurate attributes, honest dimensions, clear compatibility information, and specifications complete enough that the agent can confidently match the product to a nuanced request.
Consider what that means in practice. When a shopper asks an agent for “a durable travel backpack that fits under an airline seat and has a laptop sleeve,” the agent is not matching keywords — it is checking attributes. A listing that specifies dimensions, carry-on compatibility, laptop-sleeve size, and materials will be surfaced; an otherwise identical product whose listing says only “spacious premium backpack” will be invisible to that query, no matter how good the product actually is. The brand that wrote the boring, complete spec sheet wins the sale from the brand that wrote the vague marketing copy. That is a reversal of the incentives that governed a decade of listing optimization.
A brand cannot bid its way to the top of an agent’s shortlist. It earns the position by being the clearest, most accurate, best-structured answer to the shopper’s actual question. That reality has produced an accidental advantage for brands that invested in rich, precise product content and consistent pricing — and an accidental penalty for those that cut corners on listing quality or gamed reviews to inflate ratings. The transition is not abrupt; it is running in parallel with traditional search for now, and paid placement still works today. But the direction is not in doubt, and the long-range stakes are large: McKinsey projects agentic commerce could reach $3–5 trillion globally by 2030.
Why the shift is hard to reverse
What makes this transition durable, rather than another channel fad, is that it compounds on habit. Once a shopper delegates the work of comparing options to an agent and gets a good result, the friction of going back to scrolling a results page feels like a downgrade. Each successful query trains the user to trust the agent with the next one, and that trust is far stickier than a bookmarked search or a remembered brand name. The 85% satisfaction figure is not a snapshot; it is the fuel for a habit loop.
For sellers, that means the window to adapt is not open-ended. The brands that make their catalog machine-legible now — complete attributes, consistent pricing across channels, structured specifications, and reviews that reflect genuine use — are teaching the agents to prefer them while the field is still forming and the rankings are still fluid. The ones who wait until AI traffic is the dominant channel will be optimizing into a shortlist whose habits are already set and whose favored brands are already learned. In an agent-mediated market, being early to structured accuracy is itself a moat, and it is one of the few moats a smaller brand can still build without an enormous budget.
Final Thought
The channel has always shaped the content that wins on it. Radio rewarded the voice. Television rewarded the image. Search rewarded the keyword. AI agents reward something harder to fake: structured truth — products that are exactly what they claim to be, described with precision, priced consistently, and backed by genuine reviews. The brands that built their visibility on manipulation of the old signals are not losing to a smarter competitor. They are losing to the architecture of the new channel, which no longer has a slot for the shortcut they relied on.
Sources
- Adobe Analytics (via Yahoo Finance) — AI traffic +393%, +42% conversion
- Salesforce (official) — Cyber Week $67B / 20% / +32%
- McKinsey — The agentic commerce opportunity
- PYMNTS — Rufus MAU +115%, engagement +400%
- Digital Commerce 360 — Prime Day 2026 / AI traffic
Erene Araojo