Google Merchant Center Feed Optimization

TL;DR — Your Merchant CenterGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. feed is the file of product data — titles, prices, images, identifiers — that Google reads to decide where to show your products. The two things that move the needle most: write product titles built for search (not just copies of your website title) and add GTINsProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. (the barcode numbers) to every branded product. Fix the required fields so nothing gets disapproved, then improve from there.

What a product feed is

When you sell products through Google — Shopping ads, the free product listingsFree product listings (originally launched as \"Surfaces across Google\" in 2020) are unpaid, organic product placements Google generates from your Merchant Center feed or on-page Product structured data. There's no bid and no CPC — Google matches your product data to a query and decides whether and where to show it — across the Shopping tab, Google Search (Popular Products grids), Images, Lens, Maps/Business Profile, YouTube, and Gemini; AI Mode and AI Overviews aren't on Google's official surfaces list, though practitioner reporting links them to the same eligibility pool. They're enabled by default in most cases for new Merchant Center accounts., or now AI shoppingAI shopping optimization is the practice of making a merchant's products discoverable, recommendable, and buyable across AI shopping surfaces — ChatGPT, Google AI Mode, Gemini, Copilot, and Perplexity — by treating the structured product feed as a first-class optimization surface alongside the human-facing webpage, not a replacement for it. answers — you don’t send Google your website. You send a feed: a structured file (a spreadsheet, XML, or CSV) with a row for each product and columns (“attributes”) describing it: title, description, price, availability, image, brand, and an ID number.

Google reads that feed and uses it to match your products to what people search for. Better feed data means your products show up more often, for more relevant searches, and look better when they do.

Two of the highest-priority fields

1. Product titles. The title is one of the biggest signals Google uses to match your product to a search. You don’t have to copy your website’s product title word-for-word — Google’s own guidance says feed titles “don’t always need to be identical to the content on your landing pages,” they just need to refer to the same product. A strong pattern is Brand + Product Type + Key Details (color, size, material, model). Third-party research puts the share of top-performing advertisers using a feed-specific title (rather than copying the website title) at 81% — treat that as a reason to test a feed-specific title, not a rule every catalog must follow.

2. GTINs. A GTIN is the barcode number on a branded product (UPC in North America, EAN elsewhere). Google reports that retailers who add correct GTINs see an average 20% increase in clicks, and branded products without GTINs get limited visibility. If a product genuinely has no GTIN (handmade, vintage, custom), tell Google with identifier_exists = no — but never guess or borrow a GTIN from a similar product. Google checks them and wrong ones get your products disapproved. Evidence for this claim Merchants should submit correct product identifiers and must not invent or borrow identifier values. Scope: Identifier requirements vary based on whether the manufacturer assigned identifiers. Confidence: high · Verified: 2026-07-18 Google: Unique product identifiers

Keep it accurate so nothing gets disapproved

Google’s own rule: “Incorrect, inaccurate, or missing product information can cause disapprovals, limited eligibility, incorrect displays for your products.” The most common cause of trouble is your feed price not matching your website price. Google crawls your product page to check. Update your feed often (daily at least) so prices and stock stay in sync.

A few quick wins:

• Images: use a clear, high-resolution photo (Google recommends large, e.g. 1500×1500). No watermarks, promo text, or borders.
• Titles: no ALL CAPS, no “Free Shipping” or “Sale,” no gimmicky symbols.
• Brand and description: fill them in and make the description match your landing page.

Why this isn’t just for ads

The same feed powers Google’s free product listings (an organic, unpaid channel) and increasingly shapes AI shopping answers in Google’s own tools like AI OverviewsAI Overviews are the AI-generated summary box Google shows above or within its regular search results, written by Gemini models from pages retrieved out of Google's normal Search index. It's a Search feature, not a separate platform or index. and AI Mode. Research also finds heavy overlap between what shows in Google Shopping and what shows in third-party AI shopping tools like ChatGPT — Google doesn’t publish a direct feed relationship with those third parties, but a well-optimized Google feed correlates with showing up there too. So your feed is now part of your SEO and your AI visibilityLLM visibility (or AI visibility) is the aggregate measure of how often and how prominently a brand or page shows up in AI-generated answers — across AI Overviews, ChatGPT, Perplexity, Copilot, and Gemini. It's the AI-search analog of organic visibility, but it's driven by different signals. — not only your paid campaigns.

Want the full attribute-by-attribute treatment, title formulas by category, supplemental data sources, and Attribute Rules? Switch to the Advanced tab.

TL;DR — Feed optimization is structuring and enriching the product data in Google Merchant CenterGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. so Google can match your products across Shopping ads, free listingsFree product listings (originally launched as \"Surfaces across Google\" in 2020) are unpaid, organic product placements Google generates from your Merchant Center feed or on-page Product structured data. There's no bid and no CPC — Google matches your product data to a query and decides whether and where to show it — across the Shopping tab, Google Search (Popular Products grids), Images, Lens, Maps/Business Profile, YouTube, and Gemini; AI Mode and AI Overviews aren't on Google's official surfaces list, though practitioner reporting links them to the same eligibility pool. They're enabled by default in most cases for new Merchant Center accounts., and AI surfaces. Audit first to set a benchmark, then optimize as controlled tests — priority order depends on your catalog and goals, but feed-specific product titles (built for query matching, not a copy of your page H1An H1 tag is the HTML `<h1>` element that marks a page's primary heading — the big visible headline at the top of the content. It helps users, search engines, and screen readers understand what the page is about.) and GTINProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. coverage (Google reports an average 20% click increase from adding correct GTINs; missing GTINs limit visibility for branded products) are consistently high-priority. Get the required attributes clean to stay eligible, meet image standards, then enrich with product_highlight/ product_detail, custom labels, supplemental data sources (targeted, bulk overrides), and Attribute Rules (cascading pattern transforms — test stacked rules, since layering can produce unexpected values). Free listings make the feed an organic SEO asset, and the Shopping GraphThe Shopping Graph is Google's machine-learning-powered, real-time database of the world's products and sellers — the commerce equivalent of the Knowledge Graph. Built from Merchant Center feeds, crawled Product structured data, StoreBot verification, and broad web signals, it powers Shopping results, AI Overviews, AI Mode, and Gemini shopping answers. now feeds AI Overviews and AI Mode; research shows heavy overlap with third-party AI shopping surfaces like ChatGPT, though Google doesn’t publish a direct feed relationship to those third parties. Never invent GTINs or product facts.

What the feed actually is — and why it’s now an SEO surface

A product feed is a structured file (XML, TXT/CSV, or Google Sheets, or data pushed via the Merchant API — the successor to the Content API for Shopping, which is being sunset August 18, 2026) with one row per product and a set of attributes describing each. Evidence for this claim Merchant Center accepts several product-data source methods and formats. Scope: Supported methods and APIs evolve; use current Merchant Center documentation. Confidence: high · Verified: 2026-07-18 Google: Create a product data source Google ingests it into the Shopping Graph — 50B+ product listings — and uses that data to decide which products match a query, to compare prices, to populate rich resultsRich results (formerly 'rich snippets') are enhanced search listings — stars, images, prices, breadcrumbs, video thumbnails, and more — that Google and Bing build from structured data. They're a display feature, not a ranking factor, and eligibility never guarantees they'll show., and to power AI-driven shopping. Merchant API v1beta is already discontinued; current calls must use v1 or v1alpha. That is an integration-version rule, not an HTML or Search structured- data issue. Evidence for this claim Google says Merchant API v1beta is discontinued and Content API for Shopping is scheduled to sunset August 18, 2026; Merchant API calls must use v1 or v1alpha. Scope: Google Merchant API integration versions and Content API migration; not an HTML, structured-data, or Search ranking issue. Confidence: high · Verified: 2026-07-19 Google Merchant API: Latest updates

That last part is the shift. Feed quality used to be a paid-search concern. It isn’t anymore:

• Free listings are an organic, unpaid channel across Search, Images, the Shopping tab, and YouTube. They use the same feed.
• Google’s own AI shoppingAI shopping optimization is the practice of making a merchant's products discoverable, recommendable, and buyable across AI shopping surfaces — ChatGPT, Google AI Mode, Gemini, Copilot, and Perplexity — by treating the structured product feed as a first-class optimization surface alongside the human-facing webpage, not a replacement for it. surfaces — AI OverviewsAI Overviews are the AI-generated summary box Google shows above or within its regular search results, written by Gemini models from pages retrieved out of Google's normal Search index. It's a Search feature, not a separate platform or index. and AI Mode — draw on the Shopping Graph directly. Third-party AI shopping tools like ChatGPT are a different story: Google doesn’t publish a direct feed relationship there, but research cited by Search Engine Land found up to 83% of ChatGPT shopping carousel products match Google Shopping’s organic results, with 60% of those matches from Shopping positions 1–10 — a correlation, not a confirmed ingestion pipeline. That same research called feed titles a “highest-impact lever” for matching products to queries.

Net: if you do SEO for an ecommerce site, your feed is part of your job now, whether or not you run ads.

Audit before you optimize

The most common mistake is changing data without a benchmark. As FeedOps frames it, audit first, then optimize the feed “as a sequence of controlled tests.” A useful three-tier order:

1. Eligibility — are products approved and serving? Clear disapprovals and missing required attributes first (Diagnostics → Issue Details).
2. Visibility — are approved products actually showing, and for the right queries? This is where titles, GTINs, and categorization do the work.
3. Click-through — of the products that show, which earn clicks? Images, price competitiveness, and title clarity drive this.

Don’t reformat everything at once; you won’t know what worked. Which specific fields deserve attention first within each tier depends on your catalog and goals — a branded electronics catalog usually gets more lift from GTIN and identifier cleanup, while a private-label apparel catalog often gets more from title and category work. Titles and GTINs come up often because they’re frequently the weakest fields, not because every catalog ranks them the same.

Required attributes: compliance and optimization

Google: “Providing accurate and correctly formatted product data is essential for creating successful ads,” and “Incorrect, inaccurate, or missing product information can cause disapprovals, limited eligibility, incorrect displays for your products.” Walk each required attribute with an optimization angle, not just a checkbox:

• id — unique, max 50 chars; use your SKU. Keep IDs stable; changing them on a reformat loses performance history and breaks supplemental-feed links.
• title — max 150 chars. Usually one of the highest-priority fields to get right (deep dive below).
• description — max 5,000 chars; should “accurately describe your product and match the description from your landing page.” Front-load the first ~500 characters with the details that matter.
• link — the landing page; mind tracking parameters and canonical handling.
• image_link — the main image (standards below).
• price — ISO 4217 format. In the US/Canada, “Don’t include any taxes”; elsewhere include VAT/GST. Make sure “any customer can buy the product for the submitted price, without having to sign up for a membership program.”
• availability — submit one of in_stock, out_of_stock, preorder, or backorder. Keep it real-time; availability_date is required when availability is preorder (Google also recommends it for backorder).
• brand — required for all new products except movies, books, and musical recording brands.
• gtin — strongly recommended for branded products with an assigned identifier; usually a high-priority field for those products (below).
• mpn — required only if a product lacks a manufacturer-assigned GTIN.

Conditional ones worth flagging: item_group_id ties variants together and is required for free listings for all product variants; apparel in many countries needs color, size, gender, age_group. Evidence for this claim Variant and apparel attributes have conditional requirements in Google's product data specification. Scope: Requirements depend on country, product category, and whether items are variants. Confidence: high · Verified: 2026-07-18 Google: Product data specification

Title optimization — a high-priority lever

TIP Validate the catalog structure before rewriting titles

Title improvements cannot rescue duplicated product identities or image URLs that fail basic fetchability.

Run a representative XML or TSV sample through my free AI Commerce Validator Free

1. Test a representative slice before changing the full feed.
2. Fix duplicate stable IDs without resetting valid historical IDs.
3. Replace missing or invalid image values with absolute HTTP(S) URLs, then rerun before title experiments.

Tap to enlarge

The AI Commerce Validator passes the required merchant-feed fields, then warns that product ID sku-1 is duplicated in rows one and two. It also warns that both rows have missing or invalid absolute HTTP or HTTPS image URLs.

The product title is one of the primary signals Google uses to match your product to a query, so it’s usually worth dedicated effort — how much relative to other fields depends on your catalog.

• Length: max 150 characters, but only roughly the first 70 show in ads — front-load the critical terms.
• Scope: 150 characters is the Merchant title/structured_title attribute cap. The feed description/structured_description cap is 5,000 characters. Neither number is a rule for the HTML <title> or meta descriptionThe meta description is an HTML head tag — `<meta name=\"description\" content=\"…\">` — that suggests a short summary of the page for the search snippet. It's not a Google ranking factor, and Google rewrites it the majority of the time, but a good one can still lift click-through. on the landing page. Evidence for this claim Google Merchant Center caps feed title/structured_title at 150 characters and feed description/structured_description at 5,000 characters; these are product-data attributes, not HTML title or meta-description rules. Scope: Google Merchant Center product data specification; these limits do not apply to landing-page HTML metadata. Confidence: high · Verified: 2026-07-19 Google Merchant Center: Product data specification
• You don’t have to mirror your page H1 word-for-word. Google’s own title guidance says feed titles “don’t always need to be identical to the content on your landing pages,” but they must “refer to the same product.” Marpipe’s research found 81% of high-performing advertisers use different, more keyword-optimized titles in their feeds than on their product pages — a useful signal that differentiation is common among strong performers, but treat it as a hypothesis to test with a supplemental data source, not a rule to apply blind. Titles must still describe the same product as the landing page; don’t drift into a different product identity for query-matching’s sake.
• Use a structured formula by category:
  • Apparel: Brand + Gender + Product Type + Color + Size + Material
  • Electronics: Brand + Model + Product Type + Key Feature + Spec
  • Home/Garden: Brand + Product Type + Material + Color + Size
  • General: [Brand] + [Product Type] + [Key Attributes] + [Descriptive Details]
• Avoid what Google prohibits: titles must not include “promotional text like ‘free shipping’, all capital letters, or gimmicky foreign characters,” and the title must “clearly identify the product you are selling.” Don’t keyword-stuff — the goal is structured, natural, descriptive language.
• Test variations with a supplemental data source rather than editing the primary feed.

GTIN coverage — a high-priority identifier

For branded products with an assigned GTIN, it’s one of the most important optional attributes you can submit. Google’s own guidance reports that retailers who add correct GTINs see an average 20% increase in clicks (figures like “up to 40%” show up in third-party marketing content, but Google’s published number is 20% on average — use that as the defensible figure).

• Validity: GTINs are 8, 10, 12 (UPC), 13 (EAN/JAN/ISBN-13), or 14 (ITF-14) digits, validated against the GS1 database. Wrong values trigger disapprovals or even suspension.
• Never guess or borrow. Google: “Provide correct product identifiersProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. (GTIN, MPN, and brand) whenever possible. Don’t make up, guess, or include values from similar products.” Leave it blank rather than wrong.
• Custom / private-label products without a GTIN: “use your store name as the brand along with an MPN with a unique identifier number of your choice.”
• Genuinely identifier-less (handmade, vintage, collectibles): set identifier_exists = no.

Images — the most overlooked quick win

Google: images must “accurately display the product,” don’t “scale up an image or submit a thumbnail,” and don’t “include promotional text, watermarks, or borders.”

• Resolution: Google recommends large images (1500×1500 for best quality); a new minimum of 500×500 pixels for all product images has enforcement beginning January 31, 2027. Until then, report this as a dated readiness notice; recheck the live specification before turning it into an enforced issue. Evidence for this claim Google Merchant Center announced a minimum 500×500-pixel product-image requirement beginning January 31, 2027 and lists JPEG, WebP, PNG, non-animated GIF, BMP, and TIFF as accepted formats. Scope: Google Merchant Center product-image eligibility and dated future enforcement; not a general Google Images file-format rule. Confidence: high · Verified: 2026-07-19 Google Merchant Center: Product data specification — Image link
• Accepted formats: JPEG, WebP, PNG, non-animated GIF, BMP, and TIFF. This is Merchant product-image eligibility, not a general Google Images format rule. Evidence for this claim Google Merchant Center announced a minimum 500×500-pixel product-image requirement beginning January 31, 2027 and lists JPEG, WebP, PNG, non-animated GIF, BMP, and TIFF as accepted formats. Scope: Google Merchant Center product-image eligibility and dated future enforcement; not a general Google Images file-format rule. Confidence: high · Verified: 2026-07-19 Google Merchant Center: Product data specification — Image link
• Background: white/neutral for apparel and accessories; put lifestyle shots in additional_image_link (up to 10), not as the main image.
• Multiple angles (front, back, detail) can unlock richer displays.
• AI-generated images must carry IPTC metadata — TrainedAlgorithmicMedia, CompositeSynthetic, or AlgorithmicMedia.

Categorization: google_product_category and product_type

• google_product_category — pick the most specific leaf node from Google’s taxonomy. It’s an underused but meaningful contextual signal.
• product_type — your taxonomy, hierarchical with > separators (e.g. Home > Kitchen > Appliances > Coffee Makers > Drip). Doesn’t have to match Google’s; great for bid segmentation.

Optional attributes that pull weight (especially for AI)

• product_highlight (2–100 features, max 150 chars each) and product_detail (structured section/attribute/value specs) — these support semantic matching in conversational AI queries.
• sale_price / sale_price_effective_date for promotions.
• custom_label_0–custom_label_4 — your own free-text segmentation (margin tier, season, clearance, best-sellers). Invisible to shoppers; purely for your bidding/reporting.
• additional_image_link, video_link (6–240s, 720p+), loyalty_program.

Supplemental data sources vs. Attribute Rules

Two bulk-editing tools that are easy to confuse. Google’s current terminology calls the older “supplemental feed” a supplemental data source, and “feed rules” are now Attribute Rules — same underlying concepts, updated names.

Supplemental data sources enhance or override information in your primary data source, matched by product id. They cannot add new products — only modify existing ones — and only the id is required; everything else overrides the primary source when present. Store Growers’ note on the core benefit (still accurate under the new name): “Making individual changes to your primary feed can be very time-consuming. With supplemental feeds, you’re able to make bulk changes.” Use them for title rewrites, bulk GTIN/MPN fixes, custom labels, promotion IDs, and fixing disapprovals fast. Google Sheets is the easiest format to iterate on. Some custom-matched supplemental sources are limited to certain merchants and carry a 30,000-row limit — check availability in your account before relying on one at scale. Confirm the current deletion workflow in your own Merchant Center account before assuming a specific number of steps; the UI has changed and older two-step-deletion guidance may be stale.

Attribute Rules apply automatic transformations to your product data. Unlike the older one-rule-per-attribute limit, you can now stack multiple rules on the same attribute — Google runs them in cascading order (first rule, then the second, and so on), so test your rule stack before applying it; layered rules can produce an unexpected final value. Verified operations include Set to, Set to multiple, Extract, Take latest, Prepend, Append, Add repeated field, Find & Replace, Calculate, Split & Choose, Clear, and Optimize URL. Use them for bulk pattern work (strip “Sale” from titles, standardize color names, append brand, populate missing values from existing data). For targeted, product-level changes, a supplemental data source is more flexible. Advanced options here can be scoped to eligible accounts — confirm what your account has access to before building a workflow around a specific operation.

Free listings as an organic surface

Free listings use the same feed but are a separate, unpaid surface across Search, Images, the Shopping tab, and YouTube. Eligibility leans on accurate price, availability, shipping, and return policy, and item_group_id is required for all free-listing variants. Most teams treat free listings as a byproduct of paid setup; treating them as their own SEO surface is the opportunity. One agency test reported by Search Engine Land saw a 92% revenue increase for free listings in product-level testing and a CTR 55% higher than paid.

AI search and the new feed imperative

The Shopping Graph that your feed populates now feeds Google’s own AI shopping surfaces — AI OverviewsAI Overviews are the AI-generated summary box Google shows above or within its regular search results, written by Gemini models from pages retrieved out of Google's normal Search index. It's a Search feature, not a separate platform or index. and AI Mode. Google doesn’t publish a direct feed relationship to third-party AI shopping tools like ChatGPT, but given the overlap research above, a well-optimized Google feed correlates with performing well there too. Practical implications:

• Optimize titles and descriptions for conversational, natural-language queries, not just bidding keywords.
• product_highlight and product_detail carry the structured specs that conversational AI matches against.
• Google’s Conversational Attributes feature lets merchants add AI-optimized descriptions directly in Merchant Center.
• Whatever an AI writes for you, it should “not invent facts, claims, dimensions, materials, or compatibility details” (FeedOps) — feed data still has to match your landing pages.

Feed health and maintenance

Check Diagnostics regularly across three buckets: item issues (disapprovals), account issues, and feed issues. The most common disapprovals are price mismatch, landing-page unavailability, missing required attributes, invalid GTINs, and policy violations in titles. Resolve via the Issue Details Page. Update daily at minimum; use scheduled hourly fetches or the Merchant API for fast-changing price and inventory.

International notes

Submit separate feeds per country/language using feedLabel and contentLanguage. Include local currency and VAT/GST where applicable. As of July 1, 2025, merchants no longer need to submit the US sales-tax attribute.

Related reading on this site

Feed titles and descriptions overlap heavily with product page SEOProduct page SEO is the practice of optimizing an individual product detail page (PDP) so it ranks in organic search and earns rich results. It blends unique product copy, structured data, variant canonicalization, image SEO, and customer reviews — but the structured data earns rich results and eligibility for free product listings, it doesn't make the page rank.; availability management connects to out-of-stock productsAn out-of-stock product page is a product URL whose item can't currently be bought. The right SEO treatment depends on whether the stockout is temporary, indefinite, or permanent — keep temporary stockouts live at 200 with OutOfStock schema; 301 permanently discontinued products to a relevant replacement.; and the broader context lives on the Ecommerce SEOEcommerce SEO is the practice of optimizing an online store so its product and category pages rank in organic search and attract purchase-intent visitors. It uses the same Google algorithm as any other site, but compounds the usual SEO work with commerce-specific challenges like faceted navigation, product variants, and platform-imposed URLs. hub and in ecommerce site architectureEcommerce site architecture is how an online store's pages — categories, subcategories, and products — are organized and linked. Google reads the link structure (not the URL path) to work out hierarchy and relative importance, so a logical pyramid plus good internal links matters more than how deep the URLs look..

AI summary

A condensed take on the Advanced version:

• The feed is structured product data (XML / TXT-CSV / Sheets / Merchant API — the successor to the sunsetting Content API for Shopping) Google ingests into the Shopping GraphThe Shopping Graph is Google's machine-learning-powered, real-time database of the world's products and sellers — the commerce equivalent of the Knowledge Graph. Built from Merchant Center feeds, crawled Product structured data, StoreBot verification, and broad web signals, it powers Shopping results, AI Overviews, AI Mode, and Gemini shopping answers. (50B+ listings) to match products to queries. It now powers Shopping ads, free listingsFree product listings (originally launched as \"Surfaces across Google\" in 2020) are unpaid, organic product placements Google generates from your Merchant Center feed or on-page Product structured data. There's no bid and no CPC — Google matches your product data to a query and decides whether and where to show it — across the Shopping tab, Google Search (Popular Products grids), Images, Lens, Maps/Business Profile, YouTube, and Gemini; AI Mode and AI Overviews aren't on Google's official surfaces list, though practitioner reporting links them to the same eligibility pool. They're enabled by default in most cases for new Merchant Center accounts., AND Google’s own AI shoppingAI shopping optimization is the practice of making a merchant's products discoverable, recommendable, and buyable across AI shopping surfaces — ChatGPT, Google AI Mode, Gemini, Copilot, and Perplexity — by treating the structured product feed as a first-class optimization surface alongside the human-facing webpage, not a replacement for it. (AI OverviewsAI Overviews are the AI-generated summary box Google shows above or within its regular search results, written by Gemini models from pages retrieved out of Google's normal Search index. It's a Search feature, not a separate platform or index., AI Mode). Third-party AI shopping tools show a strong correlation, not a confirmed feed relationship: up to 83% of ChatGPT shopping carousel products match Google ShoppingGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. organic results.
• Audit before optimizing. Run it as controlled tests in order: eligibility → visibility → click-through. Changing data without a benchmark is the classic mistake — which fields matter most within that order depends on your catalog.
• High priority — feed-specific titles. Build them for query matching (Brand + Product Type + Key Details); Google says they don’t need to be identical to the landing page, just refer to the same product. Third-party research (Marpipe) foundA 302 (\"Found\") is a temporary redirect: it forwards users to a new URL while telling search engines the original URL should stay in the index. It's a weak canonicalization signal, not the zero-equity dead end of SEO folklore. 81% of top performers use different feed titles — treat that as a testable pattern, not a universal rule. Max 150 chars, ~70 show; front-load. No ALL CAPS, no promo text, no gimmicky characters.
• High priority — GTINProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. coverage. Google reports an average 20% click increase from correct GTINs; missing GTINs limit branded-product visibility. Never guess/borrow them (GS1-validated; wrong = disapproval). Custom products: store brand + your MPN. None at all: identifier_exists = no.
• Required attributes must be clean and accurate to stay eligible. Keep id stable; match price/availability to the landing page (price mismatch is a leading disapproval cause).
• Images: 1500×1500 recommended; 500×500 minimum enforced Jan 31, 2027; no watermarks/borders; AI images need IPTC tags.
• Supplemental data sources = targeted bulk overrides by id (can’t add new products). Attribute Rules = pattern transforms that can now stack multiple rules per attribute, run in cascading order — test the stack.
• Free listings = organic surface (same feed; item_group_id required for variants). Optimize for conversational queries for AI; never let AI invent product facts. Update daily minimum; check Diagnostics routinely.

Official documentation

Primary-source documentation from Google Merchant CenterGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. and the Merchant API.

• Product data specification — every attribute, format rules, and requirements; the master reference.
• Unique product identifiers (GTIN, MPN, brand) — when each is required and how to handle products without GTINsProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute..
• Title and structured title — Google’s current guidance on structuring titles.
• Provide high-quality data — submission, freshness, and verification recommendations.
• Supplemental data source — adding/overriding data on existing products (the current name for “supplemental feeds”).
• Free listings for products — eligibility and requirements for the unpaid surface.
• Merchant API — programmatic feed and product management for developers and feed pipelines; the current successor to the Content API for Shopping, which is being sunset August 18, 2026.

Quotes from the source

On-the-record guidance from Google’s documentation and practitioner sources. Each Google link is a deep link that jumps to the quoted passage.

Google — product data specification

• “Providing accurate and correctly formatted product data is essential for creating successful ads.” Jump to quote
• “Incorrect, inaccurate, or missing product information can cause disapprovals, limited eligibility, incorrect displays for your products.” Jump to quote
• Titles must not include “promotional text like ‘free shipping’, all capital letters, or gimmicky foreign characters.” Jump to quote

Google — unique product identifiersProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute.

• “Provide correct product identifiersProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. (GTIN, MPN, and brand) whenever possible. Don’t make up, guess, or include values from similar products.” Jump to quote

Google — title and structured title

• “While the titles you provide for products in your product data don’t always need to be identical to the content on your landing pages, they should refer to the same product.” Source
• Retailers who’ve added correct GTINs “have seen a 20% increase in clicks on average.” Source

Practitioner sources (secondary — confirm against the live articles)

• Kirk Williams, ZATO Marketing (via Search Engine Land): supplemental feeds are a “really handy way to get in promotion IDs.” Coverage
• Store Growers: a supplemental feed is “a file that supplements or adds data to your primary product feed,” and the benefit is that “with supplemental feeds, you’re able to make bulk changes.” Source
• FeedOps: audit first, then optimize “as a sequence of controlled tests”; LLMsA large language model (LLM) is a deep-learning model trained on massive text corpora to predict the next token and generate human-like text. LLMs use the transformer architecture and power AI search features like Google's AI Overviews (Gemini) and Bing Copilot (GPT-4). “should not invent facts, claims, dimensions, materials, or compatibility details.” Source
• Search Engine Journal: “The quality, accuracy, and completeness of your product data determine how often and where your ads appear.” Source
• Marpipe: “81% of high-performing advertisers use different, more keyword-optimized titles in their feeds compared to their product pages.” (Marpipe’s primary methodology and sample aren’t published, so treat this as a directional finding worth testing rather than a verified universal ratio.) Source

Feed optimization checklist

Work top to bottom: eligibility first, then the high-priority fields for your catalog, then enrichment.

Eligibility (don’t get disapproved)

• All required attributes present: id, title, description, link, image_link, price, availability, brand.
• Feed price matches landing-page price (a leading disapproval cause).
• availability reflects real stock (in_stock/out_of_stock/preorder/ backorder); availability_date set for preorders.
• Price excludes tax (US/Canada); includes VAT/GST elsewhere; buyable without membership.
• Merchant CenterGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. Diagnostics reviewed — item, account, and feed issues.

Titles (usually a high-priority field)

• Feed titles are purpose-written for query matching, not copied from the page H1An H1 tag is the HTML `<h1>` element that marks a page's primary heading — the big visible headline at the top of the content. It helps users, search engines, and screen readers understand what the page is about..
• Structured by category (e.g. Brand + Product Type + Key Attributes); critical terms in the first ~70 characters.
• No ALL CAPS, no promo text (“Free Shipping,” “Sale”), no gimmicky characters, no keyword stuffing.

Identifiers (high priority for branded products)

• GTINsProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. present and valid (8/10/12/13/14 digits) for all branded products.
• No guessed or borrowed GTINs; blank beats wrong.
• Custom/private-label: store name as brand + your mpn.
• Identifier-less products: identifier_exists = no.

Images

• High resolution (1500×1500 recommended; ≥500×500 before the Jan 31, 2027 enforcement).
• No watermarks, promotional text, or borders; not a scaled-up thumbnail.
• Main image on white/neutral background; lifestyle shots in additional_image_link.
• AI-generated images carry the required IPTC metadata.

Attributes & enrichment

• google_product_category set to the most specific leaf node.
• product_type hierarchy defined for segmentation.
• product_highlight / product_detail populated (helps AI matching).
• item_group_id set for all variants (required for free listingsFree product listings (originally launched as \"Surfaces across Google\" in 2020) are unpaid, organic product placements Google generates from your Merchant Center feed or on-page Product structured data. There's no bid and no CPC — Google matches your product data to a query and decides whether and where to show it — across the Shopping tab, Google Search (Popular Products grids), Images, Lens, Maps/Business Profile, YouTube, and Gemini; AI Mode and AI Overviews aren't on Google's official surfaces list, though practitioner reporting links them to the same eligibility pool. They're enabled by default in most cases for new Merchant Center accounts.).
• Custom labels applied for campaign/reporting segmentation.

Process & monitoring

• Audit/benchmark captured before making bulk changes.
• Feed refreshes daily minimum (hourly/Merchant API for fast-moving stock).
• Supplemental data sources used for targeted bulk overrides (title tests, GTIN fixes).
• Attribute Rules used for catalog-wide pattern transforms (test the rule stack before applying).

Attribute quick reference

Required (most products)

Strongly recommended / conditional

Optional, high-value

GTIN digit lengths: UPC-A = 12 (North America), EAN-13/JAN = 13 (JAN can also be 8), ISBN = 10 or 13 (books), ITF-14 = 14 (multipacks). Valid lengths: 8, 10, 12, 13, or 14.

Title formulas

• Apparel: Brand + Gender + Product Type + Color + Size + Material
• Electronics: Brand + Model + Product Type + Key Feature + Spec
• Home/Garden: Brand + Product Type + Material + Color + Size
• General: [Brand] + [Product Type] + [Key Attributes] + [Descriptive Details]

Supplemental data sources vs. Attribute Rules

Validate your feed before you upload

A little local validation catches disapprovals before Google does. These work on a tab-delimited TXT feed exported from a spreadsheet (header row first). Replace feed.txt with your file.

Check that every row has the required attributes filled

Required columns checked here: id, title, description, link, image_link, price, availability, brand.

macOS / Linux

# Print any data row where a required column is empty
awk -F'\t' '
NR==1 {
  for (i=1;i<=NF;i++) col[$i]=i
  split("id title description link image_link price availability brand", req, " ")
  next
}
{
  for (r in req) {
    c = col[req[r]]
    if (c=="" || $c=="") { print "Row "NR": missing "req[r]; break }
  }
}' feed.txt

Windows (PowerShell)

$req = 'id','title','description','link','image_link','price','availability','brand'
Import-Csv .\feed.txt -Delimiter "`t" | ForEach-Object {
  $row = $_
  $missing = $req | Where-Object { [string]::IsNullOrWhiteSpace($row.$_) }
  if ($missing) { "Missing $($missing -join ', ') for id=$($row.id)" }
}

Flag titles that violate Google’s rules (ALL CAPS / promo text)

macOS / Linux

# Column-agnostic: scan the title field for banned patterns
awk -F'\t' '
NR==1 { for (i=1;i<=NF;i++) if ($i=="title") t=i; next }
{
  bad=""
  if ($t ~ /[Ff]ree [Ss]hipping|[Ss]ale|% off|[Bb]est [Pp]rice/) bad="promo text"
  # crude ALL CAPS check: a long run of uppercase words
  if ($t ~ /[A-Z]{5,}( [A-Z]{2,}){2,}/) bad=bad (bad?"; ":"") "ALL CAPS"
  if (bad!="") print "id "$1": "bad" -> "$t
}' feed.txt

Windows (PowerShell)

Import-Csv .\feed.txt -Delimiter "`t" | ForEach-Object {
  $t = $_.title
  $flags = @()
  if ($t -match '(?i)free shipping|sale|% off|best price') { $flags += 'promo text' }
  if ($t -cmatch '([A-Z]{5,})(\s[A-Z]{2,}){2,}') { $flags += 'ALL CAPS' }
  if ($flags) { "id $($_.id): $($flags -join '; ') -> $t" }
}

Sanity-check GTIN lengths (8, 10, 12, 13, or 14 digits)

This only checks length/digits — it does not validate against GS1, and you should never invent a GTINProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. to pass it.

macOS / Linux

awk -F'\t' '
NR==1 { for (i=1;i<=NF;i++) if ($i=="gtin") g=i; next }
$g!="" {
  n=$g; gsub(/[^0-9]/,"",n)
  L=length(n)
  if (L!=8 && L!=10 && L!=12 && L!=13 && L!=14) print "id "$1": suspicious GTIN ("L" digits) -> "$g
}' feed.txt

Windows (PowerShell)

Import-Csv .\feed.txt -Delimiter "`t" | Where-Object { $_.gtin } | ForEach-Object {
  $n = ($_.gtin -replace '\D','')
  if ($n.Length -notin 8,10,12,13,14) { "id $($_.id): suspicious GTIN ($($n.Length) digits) -> $($_.gtin)" }
}

For programmatic feed management at scale, use the Merchant API rather than file uploads — it’s the current successor to the Content API for Shopping, which is being sunset August 18, 2026.

SOP: recurring Merchant Center feed health review

Run this on a fixed schedule and after any catalog, template, or market change. Daily refreshes keep price and availability current; the review catches the systemic issues a successful fetch can still contain.

1. Confirm the source refreshed. Check the data source’s last processing time and item count. Done means the expected source processed on schedule and an unexplained catalog-count change is investigated.
2. Clear eligibility problems first. Open Products → Needs attention (formerly Diagnostics) and review account, data-source, and item issues. Done means new disapprovals have an owner and the highest-impact required-field, price, availability, and policy failures are queued before optimization work.
3. Audit identifier coverage. Export products missing or failing GTINProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute., MPN, brand, or identifier_exists checks. Done means no identifier was guessed and every correction traces to packaging, the manufacturer, or an approved catalog source.
4. Review visibility inputs. Sample approved products by category for title structure, categorization, product details, and variant grouping. Done means the feed-specific title and attributes match the actual landing page and the category’s approved formula.
5. Review click-through inputs. Check the main image, price clarity, and title readability for products receiving visibility but weak engagement. Done means proposed changes are isolated as controlled tests rather than bundled into a catalog-wide rewrite.
6. Publish and document the next refresh. Apply targeted overrides through the approved primary source, supplemental source, or Attribute Rule. Done means the change, affected IDs, baseline, owner, and expected processing time are recorded for validation.

The mental models

Eligibility → visibility → click-through

Work in that order. A disapproved item cannot earn visibility, and an invisible item cannot earn a click. Required attributes and policy come first; query matching through titles, identifiers, and categorization comes second; imagery, price presentation, and title clarity come third.

Catalog truth before optimization

The feed is a structured claim about the product. Every title, identifier, price, availability value, image, and specification has to agree with a real source and the landing page. A more persuasive value that is not true is not optimization—it is a feed defect.

Controlled overrides, not mystery transformations

Use a supplemental source for targeted product-ID overrides and Attribute Rules for repeatable catalog-wide transformations. Keep IDs stable, change one meaningful variable at a time, and preserve a record of what generated the final value.

Feed quality is surface-independent

The same product data supports paid Shopping, free listingsFree product listings (originally launched as \"Surfaces across Google\" in 2020) are unpaid, organic product placements Google generates from your Merchant Center feed or on-page Product structured data. There's no bid and no CPC — Google matches your product data to a query and decides whether and where to show it — across the Shopping tab, Google Search (Popular Products grids), Images, Lens, Maps/Business Profile, YouTube, and Gemini; AI Mode and AI Overviews aren't on Google's official surfaces list, though practitioner reporting links them to the same eligibility pool. They're enabled by default in most cases for new Merchant Center accounts., and AI shoppingAI shopping optimization is the practice of making a merchant's products discoverable, recommendable, and buyable across AI shopping surfaces — ChatGPT, Google AI Mode, Gemini, Copilot, and Perplexity — by treating the structured product feed as a first-class optimization surface alongside the human-facing webpage, not a replacement for it. discovery. Treat the feed as product-discovery infrastructure rather than a file owned only by the ads team.

Audit feed titles for query matching

Audit these Merchant Center feed rows for title quality. For each row:
1. Identify missing high-value attributes that are present in the supplied product data
2. Flag ALL CAPS, promotional language, keyword repetition, and unsupported claims
3. Recommend a category formula
4. Draft one feed title, keeping the most important product-identifying terms first
5. List every source field used in the draft

Use only the product facts in my input. Do not invent a brand, model, material, size,
compatibility claim, offer, or specification. Keep feed titles distinct from page H1s
when query matching benefits, but do not contradict the landing page.

Category rules and feed rows:
[PASTE CATEGORY + ID + CURRENT TITLE + VERIFIED PRODUCT ATTRIBUTES]

Plan a safe title experiment

Create a controlled Merchant Center title test from this product cohort. Return:
- The single title variable to test
- Control and treatment formulas
- IDs suitable for each cohort and any rows to exclude
- The supplemental-source columns required
- Eligibility and policy checks before upload
- The visibility and click-through fields to compare after processing

Do not change product IDs, prices, availability, identifiers, or unsupported facts.
Do not claim statistical significance from the input unless I provide a valid analysis.

Baseline export:
[PASTE DATA]

Required-attribute preflight

Test to run: Run the article’s required-field validator against the exact export scheduled for upload. Expected result: Every row contains the required id, title, description, link, image_link, price, availability, and brand values for its product type. Failure interpretation: The source mapping or transformation produced incomplete rows. Monitoring window: Immediately before every upload. Rollback trigger: Abort the upload when any required value is blank or a catalog-wide column count changes unexpectedly.

Identifier integrity preflight

Test to run: Check GTINProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. format, then verify corrected GTIN ownership against packaging, the manufacturer, or GS1 rather than relying on length alone. Expected result: Submitted identifiers belong to the exact products and genuinely identifier-less items use the approved fallback. Failure interpretation: A guessed, borrowed, malformed, or mis-mapped identifier may cause warnings or disapproval. Monitoring window: Before upload and whenever identifier source data changes. Rollback trigger: Do not ship a GTIN override that cannot be tied to the exact product.

Processed-feed confirmation

Test to run: After the refresh, inspect the data source status, processed item count, and Products → Needs attention for the affected ID cohort. Expected result: The source processes successfully, expected IDs remain present, and no new feed, account, or item issue appears. Failure interpretation: The uploaded file, rule, or supplemental join changed eligibility or dropped products. Monitoring window: As soon as Merchant CenterGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. completes processing. Rollback trigger: Revert the change if affected products disappear, become disapproved, or receive a new systemic issue.

Landing-page parity check

Test to run: Sample changed products and compare final feed title, price, availability, image, and product facts with the live landing page. Expected result: All commercial and descriptive facts agree, while the feed title may use a query-oriented order. Failure interpretation: A stale source or transformation created contradictory product data. Monitoring window: Immediately after processing and after the next inventory/price refresh. Rollback trigger: Remove the override when it publishes an inaccurate price, availability state, identifier, or product claim.

Resources worth your time

My related writing (on this site)

• Ecommerce SEOEcommerce SEO is the practice of optimizing an online store so its product and category pages rank in organic search and attract purchase-intent visitors. It uses the same Google algorithm as any other site, but compounds the usual SEO work with commerce-specific challenges like faceted navigation, product variants, and platform-imposed URLs. — the pillar this feed work sits inside.
• Product page SEOProduct page SEO is the practice of optimizing an individual product detail page (PDP) so it ranks in organic search and earns rich results. It blends unique product copy, structured data, variant canonicalization, image SEO, and customer reviews — but the structured data earns rich results and eligibility for free product listings, it doesn't make the page rank. — feed titles and descriptions overlap heavily with on-page product content.
• Out-of-stock productsAn out-of-stock product page is a product URL whose item can't currently be bought. The right SEO treatment depends on whether the stockout is temporary, indefinite, or permanent — keep temporary stockouts live at 200 with OutOfStock schema; 301 permanently discontinued products to a relevant replacement. — the availability-attribute side of feed health.
• Ecommerce site architectureEcommerce site architecture is how an online store's pages — categories, subcategories, and products — are organized and linked. Google reads the link structure (not the URL path) to work out hierarchy and relative importance, so a logical pyramid plus good internal links matters more than how deep the URLs look. — product taxonomy and structured dataStructured data is a standardized way of labeling page content (using the schema.org vocabulary in JSON-LD, Microdata, or RDFa) so search engines can understand its meaning. It's not a direct ranking factor — its value is rich results and entity understanding. context.

My speaking

• Talks and decks on ecommerce and technical SEOTechnical SEO is the practice of making a site easy for search engines to crawl, render, index, and (now) be eligible for AI answers. It's the foundation that lets your content and links rank — not a ranking trick of its own. — see the Ecommerce SEOEcommerce SEO is the practice of optimizing an online store so its product and category pages rank in organic search and attract purchase-intent visitors. It uses the same Google algorithm as any other site, but compounds the usual SEO work with commerce-specific challenges like faceted navigation, product variants, and platform-imposed URLs. hub for the current set.

From around the industry

• Product feeds as an organic / AI-search strategy (Search Engine Land) — the ChatGPT/Shopping overlap, GTINProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. click lift, and free-listing case-study numbers.
• Supplemental feeds and feed rules (Search Engine Land) — Kirk Williams on practical supplemental-feed use.
• Set up feed rules in Google Merchant Center (Search Engine Land) — a walkthrough of feed-rule operations.
• Google Shopping feed optimization guide (FeedOps) — the audit-first, controlled-tests framework.
• Google Shopping optimization tips (Search Engine Journal) — quality/accuracy/completeness framing.
• The complete guide to product feed optimization (Marpipe) — comprehensive attribute coverage and the feed-specific-title stat.
• All Merchant Center feed attributes explained (Store Growers) — a thorough attribute reference.
• Product feed optimization guide (Optmyzr) — the audit → fix → optimize flow.

Test yourself: Merchant Center feed optimization

Five quick questions on optimizing your Google Merchant CenterGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. product feed. Pick an answer for each, then check.

Google Merchant Center Feed Optimization

Google Merchant Center feed optimization is the process of improving the quality, accuracy, and completeness of the product data you submit to Google Merchant Center — titles, identifiers, images, and attributes — so your products show up more often and in more relevant places across Shopping ads, free listings, and AI shopping surfaces.

Related: Google Merchant Center, Ecommerce SEO, Product Page SEO

Google Merchant Center Feed Optimization

Google Merchant CenterGoogle Merchant Center (GMC) is a free platform where retailers upload and manage product data so their products can appear across Google — Shopping, organic Search product grids, Images, Lens, and AI surfaces. Since 2020 it powers free (organic) product listings, not just paid Shopping ads. (GMC) feed optimization is the work of structuring and enriching your product feed — the file (XML, TXT/CSV, or Google Sheets) of attributes describing every product you sell — so Google can match your products to the right queries across Shopping ads, free listingsFree product listings (originally launched as \"Surfaces across Google\" in 2020) are unpaid, organic product placements Google generates from your Merchant Center feed or on-page Product structured data. There's no bid and no CPC — Google matches your product data to a query and decides whether and where to show it — across the Shopping tab, Google Search (Popular Products grids), Images, Lens, Maps/Business Profile, YouTube, and Gemini; AI Mode and AI Overviews aren't on Google's official surfaces list, though practitioner reporting links them to the same eligibility pool. They're enabled by default in most cases for new Merchant Center accounts., Google Search, Images, YouTube, and increasingly Google’s own AI surfaces like AI OverviewsAI Overviews are the AI-generated summary box Google shows above or within its regular search results, written by Gemini models from pages retrieved out of Google's normal Search index. It's a Search feature, not a separate platform or index. and AI Mode.

A feed is more than a compliance artifact. Google uses its 50B+ listing Shopping GraphThe Shopping Graph is Google's machine-learning-powered, real-time database of the world's products and sellers — the commerce equivalent of the Knowledge Graph. Built from Merchant Center feeds, crawled Product structured data, StoreBot verification, and broad web signals, it powers Shopping results, AI Overviews, AI Mode, and Gemini shopping answers. to decide where and how often each product appears, and feed data quality is effectively a ranking factor within that graph. Optimization runs from fixing required-attribute errors and disapprovals up to strategic work: writing feed-specific product titles (usually a high-priority field), maximizing GTINProduct identifiers are the standardized values — GTIN (Global Trade Item Number), MPN (Manufacturer Part Number), and brand — that shopping feeds like Google Merchant Center and Microsoft Merchant Center use to match a product listing to the correct item in their catalog. A GTIN alone is usually enough; without one, brand + MPN is the fallback; products with none declare that with the identifier_exists attribute. coverage where you have assigned identifiers (Google reports an average 20% click increase from adding correct GTINs; branded products without them see limited visibility), meeting image standards, enriching optional attributes, and using supplemental data sources and Attribute Rules to make bulk changes without rebuilding the primary feed.

Two distinctions matter. Feed optimization is not just a paid-search task — free listings make the feed an organic SEO asset, and Google’s own AI shoppingAI shopping optimization is the practice of making a merchant's products discoverable, recommendable, and buyable across AI shopping surfaces — ChatGPT, Google AI Mode, Gemini, Copilot, and Perplexity — by treating the structured product feed as a first-class optimization surface alongside the human-facing webpage, not a replacement for it. surfaces pull from the same Shopping Graph (third-party AI shopping tools show strong correlation with Google Shopping results, though Google doesn’t publish a direct feed relationship to them). And the feed title doesn’t need to be identical to your on-page titleThe title tag is the HTML title element in a page's head that specifies the document's title. It's the primary source for the SERP title link and a confirmed light ranking factor — but since August 2021 Google doesn't always show it verbatim. — Google’s guidance only requires that it refer to the same product, and third-party research finds many high performers write differentiated, query-matched titles rather than copying the page <h1>.

Related: Google Merchant Center, Ecommerce SEO, Product Page SEO

Changelog