What just happened: product cards arrive
Claude has reportedly begun surfacing product cards directly within its answers, with early sightings across categories from electronics to pet to fashion and beauty. The implementation looks early, and Anthropic has not formally announced it, but it lands exactly where the company has been pointing for months.
In a public post on its plans, the company wrote that it is "particularly interested in the potential of agentic commerce, where Claude acts on a user's behalf to handle a purchase or booking end to end", adding that it will "continue to build features that enable our users to find, compare, or buy products, connect with businesses, and more".
Why Claude is moving into commerce now
Claude has rapidly grown to become one of the largest AI platforms in the world, and the trajectory behind that shift is steep. Anthropic's run-rate revenue crossed $47 billion in May 2026, up from $9 billion at the end of 2025, driven largely by enterprise adoption and Claude Code. The same month it raised $65 billion in Series H funding at a $965 billion post-money valuation, surpassing OpenAI, and days later filed confidentially for an IPO targeting a listing as early as October 2026, on track to be one of the largest offerings in market history. Around 70% of the Fortune 100 now use Claude, including 8 of the Fortune 10.
That scale matters for commerce because of how Claude is used. Sessions are longer and engagement heavier than on consumer-first assistants, with a large and growing share of activity on mobile. That is the profile of a tool people use for considered decisions, including what to buy. With a high-intent audience already researching products inside Claude, and a company under fresh public-market pressure to convert that attention into revenue, commerce surfaces are the logical next step: they keep the shopper inside the assistant from question to purchase rather than handing them to a retailer. Product cards are the first visible move.
The Agentic Commerce Optimisation playbook for Claude
Mapping our 5 Cs framework to Claude, here is what we recommend to get chosen by the shopping agent.
First, a point that shapes everything else: Claude does not crown a single winner the way other assistants do. It surfaces three or four credible options with hedged language like "options worth considering include", and will withhold a recommendation entirely when it lacks confidence. So the goal is not to win the single recommendation. It is to be consistently one of the named options.
C1: Completeness
When web-enabled, Claude pulls from indexed pages and treats well-structured product data as a high-confidence input. That includes your retailer product pages: Amazon and Walmart listings are among the most heavily indexed and frequently synthesised sources in any shopping query, so the structured fields on those PDPs feed Claude directly. Schema markup on your own site tells Claude explicitly "this is a price", "this is a rating", rather than leaving it to infer. Structured data is cited more often than unstructured content, and Product, Review and FAQ schema all help Claude parse a listing cleanly.
Core action: audit structured product data everywhere it lives, your own site and your Amazon and Walmart PDPs alike, and fill the backend attributes and detail-page fields these surfaces expose. Complete, consistent data across retailers is what makes you eligible to appear. Without it, the other four Cs have nothing to work with.
C2: Context
Claude reads buyer intent granularly. "Best trainers for flat feet and marathon training" is read as separate constraints, each weighed individually. The brands that win answer the real questions shoppers ask, mapped to a clear buyer profile.
Two things help with Claude specifically. It pulls comparison tables almost verbatim and reads structured comparisons as evidence of expertise, so honest "better for X, less suited to Y" framing gets lifted more often than "why we're the best" copy. And specificity beats polish: vague marketing language reduces its confidence, concrete detail raises it.
Core action: identify contextual gaps and generate brand-compliant content that answers shopper questions, framed around who each product is for.
C3: Citations
This is where Claude diverges most. It actively cross-verifies, and multi-source corroboration is the single biggest weight it applies: the same claim across your listing, third-party articles, review platforms and community threads counts for far more than a claim that only lives on your own site. The review depth on your Amazon and Walmart PDPs feeds into this too, adding corroborating signal Claude can weigh. One practitioner dataset puts roughly 68% of AI citations as third-party versus 32% brand-owned, with the skew more pronounced for Claude because it cross-references before citing, reportedly checking claims against review platforms like G2, Capterra and Trustpilot first. Its training also biases against promotional language: content that reads like analysis earns more citations than content that reads like a landing page.
Core action: build and amplify off-platform presence across every channel Claude synthesises from, and get your product claims echoed in independent, credible places. This pulls in PR and corporate comms, not just ecommerce.
C4: Correctness
Claude makes this easier and harder. Easier, because it provides direct citations, so you can see which source a claim came from. Harder, because cross-verification means a wrong claim on a third-party site can override your accurate listing.
Run your target shopper prompts regularly and watch whether the spec, price and positioning Claude repeats are right, and which source it pulled them from. Because assistants rotate sources between runs, patterns over weeks tell the real story, not single readings.
Core action: monitor Claude outputs systematically to detect and correct hallucinations, and trace inaccuracies back to the citation Claude trusted.
C5: Customer Acquisition
For Claude this is directional today. Web-enabled answers drive referral traffic that shows up in GA4 as AI referrals, and share of voice can be tracked by logging target prompts and measuring how often your brand and SKUs appear. Attribution will sharpen as the surface matures, especially if product cards develop click-out or purchase.
Core action: track directional metrics now, and position for fuller attribution as the surface matures.
Where this is heading
If product cards develop, the open question is whether Claude keeps sourcing from corroborated web content or starts pulling from structured product feeds. Either way, the 5 Cs hold. Complete data is what a card renders. Context gets it matched to a query. Citations earn it a place. Correctness keeps it accurate. Customer Acquisition is what it drives.
Claude rewards exactly this discipline, with one twist: it leans harder on third-party corroboration than any other assistant, which makes Citations disproportionately important.
Azoma's agentic commerce optimisation platform helps you understand how you show up in Claude and optimise your product data and content so you win agentic shopping as it emerges. ➡️ Get in touch now for a full demo.
Article Author: Max Sinclair