Agentic Commerce Visibility Research: 2026 Retailer Dominance

Across continuous platform research running on AI shopping channels, retailers own 85-92% of the #1 AI-recommended slot in the densest categories we measure. The pattern is consistent:

Source: Paz.ai continuous research, directional samples (Q1-Q2 2026). Sample size grows weekly as research agents expand category coverage.

This is the foundational gap in agentic commerce: AI agents trust retailer aggregations more than direct-from-brand catalogs by default. Brands that match retailer-grade catalog completeness, multi-protocol distribution, and review density start showing up. Brands that maintain marketing-rewritten product copy, nightly feeds, and incomplete attributes stay invisible.

The biggest single brand-level visibility gap we measure is between ChatGPT and Perplexity for Home and Furniture brands. Brand catalogs appear in:

• ~7% of ChatGPT answers for Home and Furniture queries
• ~30% of Perplexity answers for the same queries

That is a 4x visibility gap on the same brand catalog, same products, same week. The gap exists because:

1. Different signal weights. ChatGPT (running on ACP) prioritizes structured attribute completeness and feed freshness. Perplexity weights review volume and brand-level authority more heavily.
2. Different surface types. ChatGPT renders product cards with images, prices, ratings, and merchant links. Perplexity renders text recommendations with citations. Different content types win on each.
3. Catalog distribution mismatch. Many brands have ACP-compliant feeds but no equivalent Perplexity merchant program enrollment, or vice versa.

The gap is directly fixable. Closing the ChatGPT side of the gap to match Perplexity (~30%) typically takes one quarter of dedicated catalog optimization. The lift is multiplicative, not additive: better attributes lift both surfaces, but the structural fixes that close the ChatGPT gap (real-time inventory, GTIN coverage, structured variant data) compound on Perplexity too.

Three external data points shape where AI shopping is actually concentrating in 2026:

Adobe Q1 2026: home goods get 3x apparel AI traffic

Adobe Analytics, drawn from over 1 trillion U.S. retail visits in Q1 2026, found AI-driven traffic to home goods brands is 3x the level of apparel brands. AI agents prefer high-consideration categories where multi-attribute filtering (dimensions, finish, room fit) materially improves the consumer's decision quality. Apparel is more brand-driven and AI-resolvable through natural-language descriptions; home goods requires structured-attribute resolution that AI agents are uniquely suited for.

BrightEdge: best-X queries jumped 5% -> 83% in 12 months

Coverage of "best [X]" queries on Google AI jumped from 5% to 83% in twelve months (BrightEdge, 2026). This is the query format where AI agents replace traditional search results entirely - "best 4-season tent under $400," "best wireless earbuds for running," "best stand mixer for sourdough." Twelve months ago these queries returned ranked links. Today they return AI-synthesized recommendations.

OpenAI Shoptalk 2026: 50% of ChatGPT searches are discovery, 70% include constraints

OpenAI shared at Shoptalk 2026 that 50% of ChatGPT searches are product-discovery queries (vs informational), and 70% of shopping queries include explicit constraints (price ranges, use cases, attributes). The query patterns are different from Google: longer, more constrained, more conversational, and more outcome-focused.

The pattern across all three: AI shopping is not a "better search engine." It is a different intent surface where consumers state what they want by outcome and let the agent resolve it.

Catalog optimization for AI agent visibility is operationally distinct from SEO. The five highest-leverage moves we measure:

1. Complete structured attributes per category

For apparel: size (with body measurements), color (with hex), material composition (percentages), fit type, silhouette, care, country of origin, GTIN. For outdoor: activity-fit, weather rating, weight class, dimensions when packed and assembled. For beauty: skin type, undertone, finish, ingredient flags, formulation type. For home: dimensions including delivery clearances, lead time per zip, AR-ready 3D assets. Generic ecommerce attributes are necessary but not sufficient.

2. Real-time inventory and pricing feeds

Google's Shopping Graph refreshes 2 billion listings per hour. AI agents heavily weight freshness. A nightly feed leaves you wrong all day. Move to webhook-based feeds, Content API for Shopping push updates, or short-interval scheduled pulls.

3. Multi-protocol distribution

ChatGPT runs on ACP. Google AI Mode runs on UCP. Microsoft Copilot adopted UCP in April 2026. Optimizing only for one channel leaves half the AI shopping traffic on the table. ACP-only catalogs are visible on ChatGPT but invisible on Google AI Mode and vice versa.

4. Review syndication in structured form

AI agents weight review volume and sentiment heavily. Reviews must be in structured form (Schema.org AggregateRating + Review markup) and accessible to AI crawlers (GPTBot, Google-Extended, ClaudeBot, PerplexityBot must not be blocked in robots.txt).

5. Ongoing measurement and category authority content

The first four moves close the structural gap. Sustained visibility growth requires category-level content (review guides, comparisons, "best [X] for [Y]" inclusion) and continuous measurement of which queries surface your products vs which competitors take the share. AI visibility monitoring is the measurement layer; an AI Readiness check is the cheapest first audit.

Two notes on how to interpret these numbers:

Directional, not absolute

Per-category sample sizes are still small (Home and Furniture is ~15 brands as of Q2 2026). The patterns are consistent enough across categories and weeks that they are publishable as directional findings. We treat them as starting points for retailer conversations, not as final benchmarks.

Coverage growing

Continuous research agents monitor AI shopping behavior across categories: what queries are growing, what brands AI is recommending, where retailers dominate, where whitespace opens up. Coverage expands weekly. We refresh per-category data as the sample stabilizes.

Methodology: each tracked brand has its catalog ingested and cross-referenced against AI agent results across ChatGPT, Google AI Mode, and Perplexity for category-relevant queries. Visibility is measured as: did the brand appear in the agent's response, in what position, against which competitors. Aggregated across queries, weeks, and brands to surface the structural patterns above.