Who gets to shop for you?

On Friday, the Ninth Circuit hears oral arguments in Amazon v. Perplexity , the first real legal test of agentic commerce. The outcome will set the early rules for a market McKinsey & Company thinks will hit $900B–$1T in U.S. retail by 2030. Whatever the judges decide, the shift is already underway: 35% of U.S. consumers now use AI at the product discovery stage versus 13.6% who use traditional search.

If you market a brand, this is the issue where the abstract "AI is changing search" headlines start producing concrete winners and losers. Here's what's actually worth your attention.

1. The agentic commerce case everyone is watching

Quick recap: In March, Judge Maxine Chesney granted Amazon a preliminary injunction blocking Perplexity's Comet browser agent from making purchases inside Amazon accounts. Her reasoning was narrow but consequential — Comet had the user's permission but not Amazon's authorization, and Amazon argued Perplexity disguised the agent as a regular Chrome session rather than identifying itself.

The Ninth Circuit stayed the injunction a week later. Perplexity filed a 96-page appeal. Amazon responded on April 23. The hearing is Friday.

Why it matters for brand visibility: Two futures hang on this. In one, retailers can wall off their catalogs from third-party agents and force every AI shopper through their own properties. In the other, agents roam freely and your discoverability depends on whether your product data is structured for machines to interpret — schema, feeds, APIs, the unglamorous plumbing. Either way, the brands that win will be the ones whose data is agent-readable before this gets decided for them.

2. Google AI Mode hit critical mass — and the citation map redrew itself

AI Overviews now trigger on roughly half of tracked Google queries, up 58% year over year. The bigger story is AI Mode, the full conversational experience Google rolled out broadly in early 2026: 93% of AI Mode sessions end without a single click to an external site.**

The sector damage is uneven and brutal. Education queries went from 18% to 83% AI Overview saturation. B2B Tech climbed from 36% to 82%. Lifestyle and travel publishers have seen 40–70% organic traffic declines year over year. About a third of publishers have started blocking AI Overviews entirely — a defensive posture that doesn't actually restore traffic, but does push the conversation toward licensing and content provenance.

And here's the data point that should reframe your SEO meetings: the overlap between pages ranking in Google's top 10 and pages cited in AI answers has collapsed from roughly 76% in mid-2025 to about 38% by Ahrefs' count this year — and as low as ~17% in BrightEdge's data. Ranking #1 no longer means being seen.

3. OpenAI 's Agentic Commerce Protocol keeps gaining merchants

While Amazon is in court, ChatGPT's Instant Checkout is quietly compounding. Through Shopify's auto-enrollment, more than a million merchants are live on the Agentic Commerce Protocol (ACP), alongside Etsy, Glossier, Vuori, Spanx, and SKIMS. OpenAI takes 4% per Instant Checkout transaction on top of standard payment processing fees.

OpenAI open-sourced the ACP spec, published eight extension RFCs, and confirmed plans to expand beyond the U.S. this year. Google countered at NRF in January with its own Universal Commerce Protocol.

Translation: there are now competing standards for how AI agents read your catalog, your pricing, and your trust signals — and both are being shipped to merchants now, not eventually. If you sell anything, this is a "show up structured or don't show up" moment.

4. The new citation playbook

The hardest mindset shift for marketing teams in 2026 is letting go of position as the unit of success and embracing mention. Research from Brandlight pegs the overlap between top Google links and AI-cited sources at under 20%, down from 70%. Different game, different scoreboard.

What's actually getting cited:

• Comparison content drives 32.5% of AI citations — the single largest format. Detailed, even-handed "X vs. Y" pages outperform almost everything else.
• Original data and proprietary research punch dramatically above their weight; LLMs hunt for novel numbers because they distinguish themselves from the generic web.
• Earned media drives 80–90% of LLM responses, not your owned content. If you're not being talked about on Reddit, G2, YouTube, and category roundups, the model doesn't see you as a real player — no matter how good your blog is.
• Canonical definitions win. Brands that own the definition of a category concept get cited disproportionately whenever that concept comes up.
• The implication is uncomfortable: most content programs are still optimized for a click economy that's quietly evaporating. The brands pulling ahead are running a distribution strategy, not a publishing one.

What it means for the week ahead

Three things worth doing before next issue:

• Audit your AI visibility. Pick ten high-intent queries in your category and ask ChatGPT, Perplexity, and Google AI Mode. Are you mentioned? Where? With what framing? That's your baseline.
• Check whether your product data is agent-ready. Schema, structured feeds, public API surface. Agentic commerce isn't waiting for the Ninth Circuit.
• Find your comparison gap. If competitors own "X vs. you" pages and you don't, fix that this month. It's the highest-leverage GEO move available.

The discovery layer is being rebuilt in real time. Brand visibility is no longer something you earn from a search engine — it's something you engineer across the entire surface area an AI sees.

We'll be back next issue with what the Ninth Circuit decides and how merchants are responding.

— The Discovery Shift, powered by Quadrant