Agentic Commerce Optimization (ACO)

Agentic Commerce Optimization (ACO) is to AI shopping agents what Search Engine Optimization (SEO) was to Google search in 2005. It names the discipline of optimizing a retailer's product data, feeds, and site infrastructure so AI agents (ChatGPT, Google AI Mode, Perplexity, Claude, Amazon Rufus) can reliably discover, understand, and recommend the retailer's products.

ACO is an evolution of SEO, not a replacement. The underlying goal (be findable) is shared. The specifics diverge sharply. SEO optimizes content for ranking algorithms that serve human readers. ACO optimizes structured product data for AI agents that read feeds and schemas, not rendered pages. The two disciplines share common infrastructure (clean URLs, canonical tags, fast page loads) but use different inputs and produce different outputs.

The term emerged in late 2025 as AI shopping traffic began growing fast enough to require dedicated vocabulary. By April 2026, Adobe Analytics reported AI referral traffic was up 393% year-over-year and converted 42% better than human-driven traffic (Adobe, April 2026). With growth that material, retailers needed a name for the work required to capture it.

ACO overlaps with Generative Engine Optimization (GEO) but is commerce-specific. GEO covers any content being recommended by generative engines; ACO is specifically about product data, transaction readiness, and commerce-protocol coverage.

The practical differences between ACO and SEO line up across four dimensions:

Input data. SEO optimizes HTML content a human will read: titles, headings, body copy, images. ACO optimizes structured data an AI will parse: product feeds, JSON-LD schemas, structured attributes, API endpoints. An SEO win is often a copy rewrite. An ACO win is often a feed enrichment.

Reader. SEO's reader is a human scanning a results page. ACO's reader is an AI agent that never opens the rendered page - it reads the feed, queries the API, or ingests the schema directly. The rendered page matters for human conversions but is secondary to the machine-readable layer.

Surface. SEO targets ranking in Google's 10 blue links. ACO targets being surfaced as a product card, product mention, or brand mention in conversational AI answers across ChatGPT, Google AI Mode, Perplexity, Copilot, and others. The "surface" is a conversational answer, not a ranked list.

Success metric. SEO measures keyword rankings, organic traffic, and CTR. ACO measures Found Rate (how often your product appears in queries), Visibility Score (how prominently), and Product Card Rate (how often you appear with image + price, not just a text mention). These are measured across multiple AI engines, not a single search results page.

A site can rank perfectly for "best running shoes" in Google and be completely invisible to ChatGPT's answer to "what running shoes work for flat feet?". That gap is what ACO exists to close.

A full ACO program works across four dimensions:

1. Product Data Depth. The heart of ACO. Every product needs 25-40 structured attributes (material, dimensions, fit, compatibility, certifications, care instructions, use-case tags). Most retailers ship 5-8. The attribute gap is what causes products to appear as text mentions rather than rich product cards.

2. Feed Coverage. ACO requires feeds to multiple AI surfaces. Google Merchant Center (feeds Google AI Mode and Gemini via UCP). OpenAI commerce feed (feeds ChatGPT Shopping via ACP). Category-specific feeds for Rufus, Perplexity, and other surfaces. Running one feed and ignoring the others cuts surface coverage in half.

3. Site Signals. The site itself still matters for confirmation and trust. AI crawlers need to be unblocked (robots.txt must allow GPTBot, Google-Extended, PerplexityBot, ClaudeBot). An llms.txt file helps. JSON-LD schema (Product, Offer, AggregateRating, MerchantReturnPolicy) gives agents a structured view of the rendered page that matches the feed.

4. Cross-Engine Monitoring. ACO is not fire-and-forget. Each AI engine weights signals differently and updates weekly. A disciplined ACO program runs the same queries across ChatGPT, Google AI Mode, and Perplexity weekly and tracks which products surface as product cards, text mentions, or disappear entirely. Changes in any of those three data points trigger remediation.

Organizationally, ACO sits awkwardly between three teams:

The SEO team often claims it because ACO feels like "the new SEO." But SEO's skillset (content, keywords, links) is mismatched against ACO's work (feed enrichment, protocol compliance, structured data at scale).

The paid search team often claims it because ACO surfaces are where traffic is moving. But paid search teams don't own the product catalog and can't drive the attribute-enrichment work that creates the readiness.

The PIM (Product Information Management) team often claims it because PIM is where product attributes live. But PIM teams typically don't think about AI visibility or multi-engine monitoring.

The successful model across early-adopter retailers is a dedicated ACO owner who coordinates with all three teams. The owner holds the readiness score, drives the attribute-enrichment roadmap, decides feed strategy, and owns the weekly monitoring cadence. SEO, paid search, and PIM all contribute inputs; the ACO owner converts those into prioritized work.

Emerging job titles in this space: Director of AI Commerce, Head of Agentic Visibility, Product Data Optimization Lead. Most are Q1-Q2 2026 hires at retailers ahead of the curve.