Agentic Commerce Optimization (ACO)

Agentic Commerce Optimization (ACO) is to AI shopping assistants what SEO was to Google search in 2005. It names the work of preparing a retailer's product data, feeds and site for agentic commerce, so that ChatGPT, Google AI Mode, Perplexity, Claude and Amazon Rufus can find, understand and recommend the retailer's products.

ACO grows out of SEO rather than replacing it. The goal, being findable, is shared. The inputs differ. SEO works on page content for ranking algorithms that serve human readers. ACO works on structured product data for AI systems that read page content, feeds, schema and APIs. The two share plumbing (clean URLs, canonical tags, fast pages) and diverge from there.

The term appeared in late 2025 as AI shopping traffic grew fast enough to need its own vocabulary. By April 2026 Adobe Analytics reported AI referral traffic to US retailers up 393 percent year over year, converting 42 percent better than other traffic (Adobe, April 2026).

ACO overlaps with GEO but is specific to commerce. GEO covers any content a generative engine might use. ACO is about product data, transaction readiness and the commerce protocols.

The differences line up across four dimensions:

Input. SEO works on HTML a person reads: titles, headings, body copy, images. ACO works on data an AI parses: product feeds, JSON-LD, structured attributes, APIs. An SEO win is often a copy rewrite. An ACO win is often a completed attribute set.

Reader. SEO's reader is a person scanning a results page. ACO's reader is a system that may pull from the page, the feed, an API or structured data. Clear product copy helps both.

Surface. SEO targets a rank among Google's links. ACO targets appearing as a product card, product mention or brand mention inside a conversational answer in ChatGPT, Google AI Mode, Perplexity or Copilot.

Metric. SEO measures rankings, organic traffic and click-through. ACO measures how often your products appear for the questions that matter, how prominently, and whether they appear as a full product card with image and price rather than a passing mention, across several engines rather than one results page.

A site can rank first for "best running shoes" in Google and be absent from ChatGPT's answer to "what running shoes work for flat feet". That gap is what ACO closes.

A full ACO program works across four dimensions:

1. Product data depth. Complete, category-relevant attributes: material, dimensions, fit, compatibility, certifications, care, use case. This is the biggest lever and the most work.

2. Feed coverage. The feeds and data sources each target surface accepts. Google Merchant Center and OpenAI product feeds have different specifications. Confirm each destination's path.

3. Site signals. The site still confirms what the feed says. Googlebot access for Google Search, OAI-SearchBot access for ChatGPT search (GPTBot training permission is separate), and JSON-LD schema (Product, Offer, AggregateRating, MerchantReturnPolicy) that matches the feed. Google requires no special AI file for AI Mode or AI Overviews.

4. Cross-engine monitoring. Platforms change how they retrieve and select. Run the same shopper questions across ChatGPT, Google AI Mode and Perplexity on a schedule and track which products appear as cards, as mentions, or not at all. A change is a prompt to investigate the question, the source data and the business priority before deciding what to do.

For a long-form companion, see Paz.ai's complete guide to Agentic Commerce Optimization and the underlying case for why structured product data wins in agentic commerce.

ACO sits awkwardly between three teams.

The SEO team often claims it because it feels like the new SEO. But SEO's skills (content, keywords, links) are a poor match for feed enrichment and structured data at scale.

The paid search team often claims it because that is where traffic is moving. But paid search does not own the catalog and cannot drive the attribute work.

The catalog or PIM team often claims it because that is where attributes live. But catalog teams are rarely set up to track AI visibility across engines or to decide what to prioritize commercially.

What works is one owner who holds the readiness picture, runs the enrichment roadmap, decides feed strategy and keeps the monitoring cadence, with input from all three teams. Some retailers hire for it. Many lean teams are short of the headcount for it, and that is the gap Paz fills: a Paz expert owns this work for you, on a platform built for it, inside your priorities and within agreed guidelines, with exceptions brought to you. See how the AI visibility service works.