{
  "id": "G01",
  "url": "https://cartforagents.com/guides/agent-commerce-optimization",
  "title": "Agentic commerce optimization (ACO): test product evidence before chasing rank",
  "description": "A five-stage ACO test plan for one product: confirm eligibility, read the record agents receive, check each shopper requirement against a sourced fact, test one truthful change, then verify the offer. With stop rules, outcome measures and prior art.",
  "answer": "Agentic commerce optimization (ACO) is a working name, used by others before us, for making a product's facts and buying path usable by AI shopping agents. Run it on one product in five stages: confirm eligibility, read the record agents actually receive, check each shopper requirement against a sourced fact, test one truthful change against a control, and verify the offer. Count \"this product doesn't fit,\" \"the test was inconclusive\" and \"fix eligibility first\" as results too.",
  "take": "Start with the facts, not the ranking. In our simulations, vague listings were almost never picked, and a listing's stated facts got it picked when they answered something the shopper asked and its neighbours didn't, so the first job is finding the missing proof.",
  "summary": "Find the missing proof. Fix what is true. Test the right outcome. We didn't coin the term. This is how we would actually run it on one product, including the results that say \"stop.\"",
  "claims": [
    {
      "text": "ReFiBuy publishes a guide that defines Agentic Commerce Optimization and states that ACO is a discipline it named and defined.",
      "source_title": "What is Agentic Commerce Optimization? (ReFiBuy)",
      "source_url": "https://refibuy.ai/guides/what-is-agentic-commerce-optimization",
      "source_date": "2026-10-27",
      "method": "vendor-claim",
      "limitation": "Shows prior public use of the term. We have not verified who used it first, or the guide's market-size and engine claims. Undated page; the date is our check date."
    },
    {
      "text": "Search Engine Journal published \"Agentic Commerce Optimization: A Technical Guide To Prepare For Google's UCP\" on February 23, 2026, in which the author floats ACO as a possible new acronym.",
      "source_title": "Agentic Commerce Optimization: A Technical Guide To Prepare For Google's UCP (Search Engine Journal)",
      "source_url": "https://www.searchenginejournal.com/agentic-commerce-optimization-a-technical-guide-to-prepare-for-googles-ucp/566969/",
      "source_date": "2026-02-23",
      "method": "documented",
      "limitation": "Evidence of public use of the term by that date, not of its origin or of any method's effect."
    },
    {
      "text": "eCommerce Insights published a guide on May 18, 2026 that credits ReFiBuy with the term and describes ACO's unit of work as the product record rather than the marketing page.",
      "source_title": "What is Agentic Commerce Optimization (ACO)? (eCommerce Insights)",
      "source_url": "https://ecommerceinsights.ai/resources/guides/what-is-agentic-commerce-optimization/",
      "source_date": "2026-05-18",
      "method": "vendor-claim",
      "limitation": "A third-party vendor's definition. It agrees with ReFiBuy's attribution but is not independent proof of origin."
    },
    {
      "text": "To be included in Shopify Catalog, products need a title, at least one image, a price above 0, publication to the online store, Hydrogen or Headless channels, and must not be unlisted or hidden from search engines. The store must be on the Starter plan or higher and not in private mode, except that an Agentic-plan store that has not yet published its online store to a paid plan can still be included through agentic storefronts. Account standing is reviewed over time.",
      "source_title": "Requirements for being included in Shopify Catalog",
      "source_url": "https://help.shopify.com/en/manual/shopify-catalog/requirements",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Living help page; date is our check date. Meeting the requirements makes a product eligible, not displayed or ranked."
    },
    {
      "text": "Shopify says ChatGPT and Microsoft Copilot find products through Shopify Catalog, while Google AI Mode and Gemini get them through the Google & YouTube sales channel and Meta surfaces through the Facebook and Instagram channel.",
      "source_title": "Shopify Catalog and product discovery for agentic storefronts",
      "source_url": "https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/products",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Living help page. Products can also be found by web crawling and merchant feeds, so one route is not the whole picture."
    },
    {
      "text": "Shopify marks the catalog description and the spec, feature and selling-point lists as Inferred: generated or enriched by Shopify and possibly absent or inaccurate.",
      "source_title": "Global Catalog MCP (Shopify developer documentation)",
      "source_url": "https://shopify.dev/docs/agents/catalog/global-catalog",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Describes the interface, not how any assistant weighs these fields."
    },
    {
      "text": "OpenAI says merchant offers are ranked on factors such as availability, price, quality and whether the seller is the maker or primary seller, that ChatGPT can simplify titles and generate labels, and that price updates may lag.",
      "source_title": "Shopping with ChatGPT Search (OpenAI Help Center)",
      "source_url": "https://help.openai.com/en/articles/11128490-shopping-with-chatgpt-search",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Living help page with no published weights. Re-read October 27, 2026, including \"How merchants are selected.\""
    },
    {
      "text": "In a supplied-snippet simulation where each model picked 2 of 8 listings (an invented product among 7 real catalog listings, so chance is 25%), vague copy was shut out by every model: the invented humidifier was picked in 3 of 180 Claude sessions, 0 of 60 on GPT-6.1-Sol, 6 of 60 on Gemini 3.8 Flash and 3 of 54 on Gemini 3.1 Pro, and the vague lip gloss in none. With facts in its description, the gloss was picked in 168 of 180 Claude sessions, 60 of 60 on GPT-6.1-Sol, 49 of 60 on Gemini 3.8 Flash and 53 of 60 on Gemini 3.1 Pro. The humidifier with facts as a spec line was picked in 134 of 180 Claude sessions and in 20 of 60, 19 of 59 and 15 of 55 on GPT-6.1-Sol and the two Gemini models. Those GPT and Gemini totals split cleanly by shopper context: every session with a detailed brief asking for coverage and runtime (20 of 20, 19 of 19, 15 of 15; only the invented product under USD 90 stated a room size), and none with a one-line ask or no shopper (0 of 40 each), where the models chose real competitors that already stated quieter or cheaper decibel figures.",
      "source_title": "Vague vs specific product facts: simulated picks by Claude, GPT and Gemini models",
      "source_url": "https://cartforagents.com/research/vague-specific-product-facts-simulation#method",
      "source_date": "2026-10-27",
      "method": "controlled-simulation",
      "limitation": "Claude counts pool three models (Haiku, Sonnet and Opus 5.5). Invented target products, snippets we supplied, no page reading or retrieval, and a forced pick of two with no \"none of these\" option. Failed calls are excluded, so some cells are below 60. Not live ChatGPT or Gemini shopping, visibility or sales."
    },
    {
      "text": "On GPT-6.1-Sol and both Gemini models, the humidifier spec line was picked in every session with a detailed shopper brief and in none with a one-line ask or no shopper at all.",
      "source_title": "Vague vs specific product facts: simulated picks by Claude, GPT and Gemini models",
      "source_url": "https://cartforagents.com/research/vague-specific-product-facts-simulation#method",
      "source_date": "2026-10-27",
      "method": "controlled-simulation",
      "limitation": "About 20 sessions per model and brief. The detailed brief was written to match the invented product."
    },
    {
      "text": "In an earlier Claude-only screening round, adding \"New 2026 / Bestseller\" to vague copy scored 0 of 108 for each of the two products.",
      "source_title": "Vague vs specific product facts: simulated picks by Claude, GPT and Gemini models",
      "source_url": "https://cartforagents.com/research/vague-specific-product-facts-simulation#method",
      "source_date": "2026-10-27",
      "method": "controlled-simulation",
      "limitation": "Screening round used to choose which versions to confirm, not the held-out confirmation, and not repeated on GPT or Gemini."
    },
    {
      "text": "In a Shopify Catalog search for \"humidifier for dry winter air large room quiet top fill\" on October 26, 2026, the listing at position 4 said in its description that it is rated for \"single rooms up to 400 square feet and runs quietly enough to leave on overnight\", with no decibel figure; four other descriptions on that result gave a decibel figure, and two more listings gave one only in Shopify-generated metadata.",
      "source_title": "Cart for Agents: receipts cited by our guides",
      "source_url": "https://cartforagents.com/research/agent-shopping-research#receipts-cited-by-our-guides",
      "source_date": "2026-10-26",
      "method": "observed",
      "limitation": "One unpersonalized catalog search. It shows what the record said, not what any assistant recommended, and the figures are quoted, not measured."
    },
    {
      "text": "In the E-GEO benchmark, 11 of 15 rewriting heuristics performed worse than a simple rewriting prompt.",
      "source_title": "E-GEO: A Testbed for Generative Engine Optimization in E-Commerce (v2)",
      "source_url": "https://arxiv.org/html/2511.20867v2",
      "source_date": "2026-07-14",
      "method": "controlled-simulation",
      "limitation": "Authors report a controlled benchmark with ten retrieved products fixed per query; it does not measure a deployed assistant's ranking or retrieval."
    },
    {
      "text": "In E-GEO's red-team tests, one hidden-instruction attack improved mean rank on DeepSeek and Llama and was penalized on GPT-5, Claude and Gemini.",
      "source_title": "E-GEO: A Testbed for Generative Engine Optimization in E-Commerce (v2)",
      "source_url": "https://arxiv.org/html/2511.20867v2",
      "source_date": "2026-07-14",
      "method": "controlled-simulation",
      "limitation": "A 200-query benchmark slice. It shows defenses differ by model; it is not a measure of attacks on live shopping assistants, and we do not reproduce any attack here."
    },
    {
      "text": "In the ACES sandbox, the authors report that randomized listing position causes large, model-specific changes in selection, and that one seller's description edit produced significant gains in 33% of category and model experiments.",
      "source_title": "What Is Your AI Agent Buying? (ACES, v3)",
      "source_url": "https://arxiv.org/html/2508.02630v3",
      "source_date": "2025-12-15",
      "method": "controlled-simulation",
      "limitation": "Simulated selection in small assortments; not real market shares, and the edited seller received competitor sales feedback a merchant will not normally have."
    },
    {
      "text": "Google says sites do not need new machine-readable files, AI text files or markup to appear in its AI features.",
      "source_title": "AI features and your website (Google Search Central)",
      "source_url": "https://developers.google.com/search/docs/appearance/ai-features",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Covers Google Search AI Overviews and AI Mode supporting links, not every shopping surface."
    }
  ],
  "sources": [
    {
      "title": "What is Agentic Commerce Optimization? (ReFiBuy)",
      "url": "https://refibuy.ai/guides/what-is-agentic-commerce-optimization",
      "accessed": "2026-10-27"
    },
    {
      "title": "Agentic Commerce Optimization: A Technical Guide To Prepare For Google's UCP (Search Engine Journal)",
      "url": "https://www.searchenginejournal.com/agentic-commerce-optimization-a-technical-guide-to-prepare-for-googles-ucp/566969/",
      "accessed": "2026-10-27"
    },
    {
      "title": "What is Agentic Commerce Optimization (ACO)? (eCommerce Insights)",
      "url": "https://ecommerceinsights.ai/resources/guides/what-is-agentic-commerce-optimization/",
      "accessed": "2026-10-27"
    },
    {
      "title": "Requirements for being included in Shopify Catalog",
      "url": "https://help.shopify.com/en/manual/shopify-catalog/requirements",
      "accessed": "2026-10-27"
    },
    {
      "title": "Shopify Catalog and product discovery for agentic storefronts",
      "url": "https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/products",
      "accessed": "2026-10-27"
    },
    {
      "title": "Global Catalog MCP (Shopify developer documentation)",
      "url": "https://shopify.dev/docs/agents/catalog/global-catalog",
      "accessed": "2026-10-27"
    },
    {
      "title": "Shopping with ChatGPT Search (OpenAI Help Center)",
      "url": "https://help.openai.com/en/articles/11128490-shopping-with-chatgpt-search",
      "accessed": "2026-10-27"
    },
    {
      "title": "AI features and your website (Google Search Central)",
      "url": "https://developers.google.com/search/docs/appearance/ai-features",
      "accessed": "2026-10-27"
    },
    {
      "title": "E-GEO: A Testbed for Generative Engine Optimization in E-Commerce (v2)",
      "url": "https://arxiv.org/html/2511.20867v2",
      "accessed": "2026-10-27"
    },
    {
      "title": "What Is Your AI Agent Buying? (ACES, v3)",
      "url": "https://arxiv.org/html/2508.02630v3",
      "accessed": "2026-10-27"
    },
    {
      "title": "Cart for Agents: Shopify Catalog snapshot of 162 shopper searches",
      "url": "https://cartforagents.com/research/agent-shopping-research",
      "accessed": "2026-10-27"
    },
    {
      "title": "Vague vs specific product facts: simulated picks by Claude, GPT and Gemini models",
      "url": "https://cartforagents.com/research/vague-specific-product-facts-simulation",
      "accessed": "2026-10-27"
    }
  ],
  "limitations": [
    "The five-stage plan is a proposal. It is not an adopted standard and has not been measured as a method on live stores.",
    "We did not coin ACO and do not settle who did. We cite public uses we read; others may predate them.",
    "The worked example maps one real catalog record from a single October 26, 2026 snapshot. We made no change to that listing and measured no effect.",
    "The selection evidence cited comes from controlled simulations and benchmarks, not from deployed assistants. It cannot predict your rank, traffic or sales.",
    "Official documentation describes stated inputs and requirements, not ranking weights. Platforms can change them without notice.",
    "Good evidence does not guarantee a pick. Models respond to position, ratings, price and wording in ways that vary by model and version."
  ],
  "related": [
    "https://cartforagents.com/research/agent-shopping-research",
    "https://cartforagents.com/research/vague-specific-product-facts-simulation",
    "https://cartforagents.com/guides/ai-product-visibility",
    "https://cartforagents.com/guides/agentic-commerce",
    "https://cartforagents.com/guides/agent-ready-product-data",
    "https://cartforagents.com/research/quiet-humidifiers"
  ],
  "published": "2026-10-27",
  "updated": "2026-10-27",
  "checked": "2026-10-27",
  "kind": "guide"
}