Some of the most influential moments in a buying journey have always happened well outside the prying eyes of your brand team. A friend mentions a product while wandering a store, someone drops a link into the group chat, or a family member makes the case over dinner.
It's a phenomenon that a panel of ecommerce experts dissected in "From Digital Shelf to Decision Shelf: How Brands Get Recommended by AI Agents,” a recent Salsify webinar: Commerce has always relied on hidden influences.
But now, AI and agentic commerce are bringing that same “unseen” dynamic front and center, and at unprecedented levels of scale and influence. According to Adobe Digital Insights, AI-driven visits to retail sites grew more than 4,700% from 2024 to 2025 and are showing no signs of slowing down.
AI tools can weigh options intelligently, interpret a shopper's needs, and effectively narrow the field, all before a customer even gets a fleeting glance at your product detail page (PDP).
The webinar’s expert panel shared how brands must go beyond merely surfacing products and should instead give these tools enough accurate, relevant context to make a recommendation — read on to learn more.
Speakers included:
Maria Morais — Head of Partnerships, Salsify (webinar moderator)
Swagat Choudhury — Global Digital Commerce Director, Perfect Digital Store, Mars
Rob Gonzalez — Co-founder, Chief Strategy and Innovation Officer, Salsify
Salsify’s own research, the “2026 Holiday Pulse Report,” indicates that more than half of shoppers (51%) report “feeling comfortable” with agentic commerce.
This comfort and the predominance of AI for shopping are quickly changing what a successful PDP looks like for online brands.
Getting found still matters, of course, but discovery is only the first step. AI agents aren’t merely recommending products; they’re ranking them.
This means AI visibility is now playing an increasingly outsized role in determining whether your product is worthy of a customer’s consideration at all.
But is AI optimization simply SEO by another name? Or is it an entirely new layer of ecommerce strategy? And, perhaps more importantly, what signals are AI agents actually looking for?
While AI visibility has plenty in common with search engine optimization (SEO) and answer engine optimization (AEO), the key difference lies in the kinds of things brands should be optimizing.
Traditional search helps a product rank and earn a click; AI agents can synthesize information from several sources, interpret what the shopper actually needs, and recommend a short list of options outright.
Both have their place in the shopper’s journey, and neither is going anywhere — at least, anytime soon. And that leaves brands now managing three overlapping environments at once.
“When ecommerce and the digital shelf started becoming a thing, it’s not like you could do that instead of physical retail,” Gonzalez says. “You had to do physical retail and [the] digital shelf. And now you’ve got to do the physical shelf, the digital shelf, and the agentic shelf.”
The agentic shelf adds something new to that mix: far richer shopper intent. A retailer search bar may receive a few keywords, while an AI agent can receive a detailed description of the shopper, the problem they need to solve, and the circumstances surrounding the purchase.
To remain relevant, product content must answer those more specific questions, not simply match a broad search term.
So: Is AI optimization the new SEO?
Sort of, but not exactly. It’s better understood as the next layer of ecommerce optimization, built on the same accurate product data, useful content, and authoritative sources that already support strong performance today.
AI agents don’t rely on one magic ranking factor. Instead, they assemble recommendations from a wide mix of product data, shopper context, third-party evidence, and commercial signals.
As a framework, Sinclair pointed to the “five C’s of agentic commerce optimization”: completeness, context, citations, correctness, and customer acquisition. Together, they offer a practical framework for understanding the various signals brands need to strengthen to improve their AI visibility.
Before an agent can recommend a product, it has to understand exactly what the product is and whether it’s actually available. That leaves accurate titles, pricing, images, video, availability, and backend attributes all solidly in the “must-have” category.
The standard also has to extend beyond a brand’s bestsellers. Agents can evaluate an entire catalog against highly specific requests, giving long-tail SKUs new opportunities to surface — provided their data is complete. As Gonzalez put it, brands should ensure “every single product [is] completely filled out, completely accurately, everywhere.”
Completeness tells an agent what a product is. Context helps it understand why that product fits a particular shopper.
Product content should clarify who an item is designed for, when it’s most useful, what problem it solves, and which needs or preferences make it a strong match. That becomes especially important as shoppers move beyond basic search terms and give AI agents detailed descriptions of what they want. The more specific the request, the less useful generic product copy becomes.
Even the strongest PDP is only one part of the evidence an agent may consider. Reviews, editorial coverage, PR, listicles, videos, and conversations across social and community platforms can all influence the final answer an AI produces for a customer.
“You need to be prepared for trust signals,” Choudhury says. “Machines love trust signals.”
Different platforms may favor different source ecosystems — from Reddit discussions to affiliate articles or professional reviews — so brands need credible, consistent signals wherever agents go looking.
Brands also need to monitor what AI agents are actually saying, not merely what they have published. When an answer contains an outdated claim, incorrect specification, or outright hallucination, teams must trace the misinformation back to its source and correct it there.
But factual accuracy is only part of the equation. Because agents can synthesize reviews, complaints, and other third-party evidence at scale, broader product quality and reputation become harder to obscure.
“With agentic commerce, the product has to be good,” Gonzalez says.
Ultimately, appearing in an AI answer is not the goal. Brands need to determine whether that visibility produces qualified traffic, conversions, new customers, and revenue.
Of course, attribution here might be messy: An agent may influence consideration long before the shopper purchases through a retailer or another channel. Still, brands should establish a baseline and track whether improvements to completeness, context, citations, and correctness produce meaningful commercial gains.
Winning on the decision shelf doesn’t require brands to entirely abandon their digital shelf strategy — or chase every shiny new AI optimization trick that appears. It starts with the same fundamentals outlined by the five Cs: complete and correct product data, useful context, credible citations, and measurement tied to real customer acquisition.
The questions shoppers ask, the sources agents trust, and the products gaining traction will not look identical everywhere. For global brands, that means these fundamentals should be centralized as shared standards, then adapted by category, channel, and market.
“Keep doing the basics brilliantly,” Gonzalez tells brands. “They’re going to help you in existing ecommerce no matter what — it’s a no-lose move.”