{
  "id": "C03",
  "url": "https://cartforagents.com/guides/ai-product-descriptions",
  "title": "AI product descriptions: open with type, use case, key numbers and limits",
  "description": "Write the first sentence of a product description for shoppers and AI agents: product type, use case, verified measurements and honest limits, with two worked rewrites.",
  "answer": "Open the description with the product type, the use it suits, two or three verified facts with units and conditions, and an honest limit, then put the detail underneath. Stated facts made a listing eligible in our supplied-snippet simulation, and they won where they answered a need the shopper stated that competing listings didn't. Put the deciding facts first, then check the record a catalog actually received.",
  "take": "The first sentence is the most valuable space on a product page. Write it for the shopper's hardest question (how loud, how warm, will it fit) and both people and agents get what they need without digging.",
  "summary": "The first sentence of a product description should say what the product is, who it suits, the numbers that decide the purchase and where it stops being the right choice. Here are two descriptions taken apart and rebuilt that way.",
  "claims": [
    {
      "text": "In a supplied-snippet simulation, an invented humidifier was placed among 7 real catalog listings and each model picked 2 of 8, so chance is 25%. With vague copy it was picked in 3 of 180 sessions across three Claude models, 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. With its noise, coverage, tank and runtime facts as a spec line it was picked in 134 of 180, 20 of 60, 19 of 59 and 15 of 55. An invented lip gloss went from 0 picks with vague copy on every model to 168 of 180, 60 of 60, 49 of 60 and 53 of 60 when its description stated its facts.",
      "source_title": "Cart for Agents: Vague vs specific product facts, simulated picks by Claude, GPT and Gemini models",
      "source_url": "https://cartforagents.com/research/vague-specific-product-facts-simulation",
      "source_date": "2026-10-27",
      "method": "controlled-simulation",
      "limitation": "Models chose from snippets we supplied, never opened pages and had no tools. The products and facts were invented, and every model had to pick two with no \"none fit\" option. The factual versions disclosed information the vague versions hid, so this does not show that rewording the same facts helps. Not live ChatGPT or Gemini visibility, rank or sales."
    },
    {
      "text": "On GPT-6.1-Sol, Gemini 3.8 Flash and Gemini 3.1 Pro, the humidifier spec line was picked in every session with the detailed shopper brief (20 of 20, 19 of 19, 15 of 15) and in none with a one-line ask or no shopper (0 of 40 on each model). Claude picked it in 59 of 60, 32 of 60 and 43 of 60 sessions for the same three briefs. In the snippets supplied, it was the only listing under USD 90 that stated a room size; several competitors stated runtimes, and competitors already stated 21, 26 or 28 dB, the 28 dB one at USD 39.99 with 310 ratings against the invented product's 18.",
      "source_title": "Cart for Agents: Vague vs specific product facts simulation (method)",
      "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 (a light sleeper with a 450 sq ft bedroom, under USD 90, who dislikes refilling daily) was written to fit the invented product. Brand familiarity, review counts and price differences may also explain the pattern."
    },
    {
      "text": "One real listing in that simulation, a USD 69.99 humidifier, was picked in 0 of 1,062 held-out humidifier sessions across all six models. Its full catalog description states a room rating of up to 400 square feet, but behind a long first sentence, and the 215-character cut we applied to every description removed it.",
      "source_title": "Cart for Agents: Vague vs specific product facts simulation",
      "source_url": "https://cartforagents.com/research/vague-specific-product-facts-simulation",
      "source_date": "2026-10-27",
      "method": "controlled-simulation",
      "limitation": "The cut was ours, not a platform's, so this is not evidence of how any assistant trims a listing. We did not test whether the listing would have been picked with the fact included."
    },
    {
      "text": "In the Shopify catalog snapshot for \"humidifier for dry winter air large room quiet top fill\", the description at position 4 opens with a 34-word sentence about the product name, cool mist and rotating nozzles; its room rating (\"up to 400 square feet\") starts in the second sentence, its runtime with a setting (\"up to 36 hours on low\") sits in a Key Specs list near the end, and it describes noise only as \"runs quietly enough to leave on overnight\". The record carried no generated metadata.",
      "source_title": "Cart for Agents: How we checked (this guide)",
      "source_url": "https://cartforagents.com/guides/ai-product-descriptions#how-we-checked",
      "source_date": "2026-10-26",
      "method": "observed",
      "limitation": "One record from one unpersonalized snapshot. Quoted from the catalog record, not measured by us; Shopify marks the description field as Inferred, so it may not be exactly the merchant's text."
    },
    {
      "text": "In that result, 4 of the 10 descriptions printed a decibel figure (positions 2, 3, 5 and 7), a room area appeared in 4 of 10 descriptions, and 8 of 10 descriptions gave a runtime, of which only 2 named the setting (positions 4 and 8).",
      "source_title": "Cart for Agents: Receipts cited by our guides (Shopify agent-catalog snapshot, Oct 26, 2026)",
      "source_url": "https://cartforagents.com/research/agent-shopping-research#receipts-cited-by-our-guides",
      "source_date": "2026-10-26",
      "method": "observed",
      "limitation": "Counts the description field only; adding Shopify-generated metadata, 6 of 10 records carried a decibel figure. One search; six of the ten listings came from one seller, so it is not a sample of the category."
    },
    {
      "text": "E-GEO, a controlled benchmark that rewrites product descriptions and re-ranks a fixed set of retrieved products, reports that hand-written rewriting heuristics did not reliably beat a one-sentence simple prompt: with GPT-4.1 as the rewriter, 11 of 15 heuristic prompts did worse, with the advertisement style among the worst. It also reports that the change in description length was essentially unrelated to rank improvement.",
      "source_title": "E-GEO: A Testbed for Generative Engine Optimization in E-Commerce (arXiv v2), sections 5.2 and 5.3",
      "source_url": "https://arxiv.org/html/2511.20867v2",
      "source_date": "2026-07-14",
      "method": "controlled-simulation",
      "limitation": "Authors report results from a post-retrieval benchmark with five model re-rankers; it does not measure deployed assistants, retrieval or sales. The rewriter does not see the shopper's query."
    },
    {
      "text": "Shopify's listing quality indicator in Sales channels > Agentic includes a \"Description completeness\" signal that it says measures the word count of the product description. Shopify also says agentic storefronts often re-rank results with their own logic, so its search preview is a directional signal rather than an exact prediction.",
      "source_title": "Shopify Help Center: Managing agentic storefronts",
      "source_url": "https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/agentic-home",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Living help page read on this date. It describes Shopify's admin tool, not how any assistant ranks, and does not say that word count alone improves ranking."
    },
    {
      "text": "Adding \"Not for rooms over 500 sq ft\" to the humidifier spec line lowered Claude's pooled picks from 134 of 180 to 118 of 180; on GPT-6.1-Sol, Gemini 3.8 Flash and Gemini 3.1 Pro the rates were 20 of 60, 17 of 57 and 17 of 57, close to the spec-line rates.",
      "source_title": "Cart for Agents: Vague vs specific product facts simulation",
      "source_url": "https://cartforagents.com/research/vague-specific-product-facts-simulation",
      "source_date": "2026-10-27",
      "method": "controlled-simulation",
      "limitation": "No brief required a room over 500 sq ft and there was no \"none fit\" option, so the test could not show whether a limit helps an agent rule a product out correctly."
    },
    {
      "text": "Shopify marks some Global Catalog response fields as Inferred, generated or enriched by Shopify, including the product description and the technical specifications, and says they might vary in accuracy and should be treated as discovery signals, not merchant-authored source text.",
      "source_title": "Shopify: Global Catalog MCP, Inferred response fields",
      "source_url": "https://shopify.dev/docs/agents/catalog/global-catalog",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Living developer documentation read on this date. It does not say how the fields are generated or how any assistant uses them."
    },
    {
      "text": "OpenAI's product feed specification asks for a factual product description in plain text and says to aim for at most 5,000 characters; it does not state a cutoff at which descriptions are read.",
      "source_title": "OpenAI Commerce: Products feed specification, Basic product data",
      "source_url": "https://developers.openai.com/commerce/specs/feed",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Living developer documentation read on this date. Absence of a stated cutoff in the feed spec does not prove that no later stage shortens descriptions."
    },
    {
      "text": "Peec AI reports from its own traces that in the US, ChatGPT shopping cuts every product description in a prompt block at 50 tokens, roughly 40 words or 250 characters, at a scoring stage.",
      "source_title": "Peec AI: ChatGPT Shopping optimization in 2026, US vs Europe",
      "source_url": "https://peec.ai/blog/chatgpt-shopping-optimization-in-2026-us-vs.-europe",
      "source_date": "2026-10-16",
      "method": "vendor-claim",
      "limitation": "A vendor's interpretation of observed traces, not an OpenAI specification or an independent replication. It may describe one stage, one market and one point in time."
    },
    {
      "text": "OpenAI says a product appears in ChatGPT's shopping carousel when ChatGPT perceives it to be relevant to the shopper's intent, considering the query and context such as Memory or Custom instructions.",
      "source_title": "OpenAI Help Center: Shopping with ChatGPT Search, How product results are selected",
      "source_url": "https://help.openai.com/en/articles/11128490-shopping-with-chatgpt-search",
      "source_date": "2026-10-27",
      "method": "documented",
      "limitation": "Living help article read on this date. It lists inputs, not weights."
    }
  ],
  "sources": [
    {
      "title": "Cart for Agents: 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"
    },
    {
      "title": "Cart for Agents: Receipts cited by our guides",
      "url": "https://cartforagents.com/research/agent-shopping-research#receipts-cited-by-our-guides",
      "accessed": "2026-10-27"
    },
    {
      "title": "E-GEO: A Testbed for Generative Engine Optimization in E-Commerce (arXiv v2, 2026-07-14)",
      "url": "https://arxiv.org/html/2511.20867v2",
      "accessed": "2026-10-27"
    },
    {
      "title": "Shopify Help Center: Managing agentic storefronts",
      "url": "https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/agentic-home",
      "accessed": "2026-10-27"
    },
    {
      "title": "Shopify: Global Catalog MCP",
      "url": "https://shopify.dev/docs/agents/catalog/global-catalog",
      "accessed": "2026-10-27"
    },
    {
      "title": "OpenAI Commerce: Products feed specification",
      "url": "https://developers.openai.com/commerce/specs/feed",
      "accessed": "2026-10-27"
    },
    {
      "title": "OpenAI Help Center: Shopping with ChatGPT Search",
      "url": "https://help.openai.com/en/articles/11128490-shopping-with-chatgpt-search",
      "accessed": "2026-10-27"
    },
    {
      "title": "Peec AI: ChatGPT Shopping optimization in 2026, US vs Europe",
      "url": "https://peec.ai/blog/chatgpt-shopping-optimization-in-2026-us-vs.-europe",
      "accessed": "2026-10-27"
    }
  ],
  "limitations": [
    "The opening pattern is our proposal. No platform requires it, and we have not tested it on live ChatGPT, Gemini or Copilot results.",
    "The simulation used invented products and snippets we supplied to six API models. It supports stating fit facts; it does not predict your rank, clicks or sales, and it did not test rewording facts a page already states.",
    "The real listing in the teardown comes from one unpersonalized Shopify catalog snapshot taken on October 26, 2026. We did not measure the product, and the rewrite uses only facts its own record stated, with gaps left in brackets.",
    "No source we read documents a universal cutoff for product descriptions. One vendor reports a 50-token cut at one stage; treat it as a reason to front-load, not as a rule.",
    "Store and brand names are withheld; listings are identified by their position in the result."
  ],
  "related": [
    "https://cartforagents.com/guides/product-specification-template",
    "https://cartforagents.com/research/vague-specific-product-facts-simulation",
    "https://cartforagents.com/guides/agent-ready-product-data",
    "https://cartforagents.com/guides/product-constraint-coverage",
    "https://cartforagents.com/research/quiet-humidifiers",
    "https://cartforagents.com/research/agent-shopping-research"
  ],
  "published": "2026-10-27",
  "updated": "2026-10-27",
  "checked": "2026-10-27",
  "kind": "guide"
}