Product Review Feeds for Google & Microsoft Shopping

TL;DR — A product review feedA product review feed is a structured XML file (following a Google-defined schema) that merchants or approved aggregators submit to Google Merchant Center — and separately to Microsoft Merchant Center — to populate the 1–5 star ratings and review counts shown on Shopping ads and free product listings. It is a different file from the main product feed, and different from Review/AggregateRating schema markup. is a separate file you send to Google (and separately to Microsoft/Bing) so that little star ratings show up next to your products in Shopping. It’s not the same file as your regular product feed, and it’s not the review-star code you might have on your website. To qualify on Google you generally need at least 50 reviews and to sign up first.

What a product review feed is

When you shop on Google and see a product with a row of gold stars and a number like “(214)” underneath it, that data usually traces back to a product review feed — a file the merchant (or a reviews company acting on their behalf) sent to Google. Google says it compiles these Product Ratings “from merchants, review aggregators, review sites, and Google users,” so the feed you submit is one input into a broader program, not the only conceivable source — and submitting one doesn’t guarantee your reviews will display; Google is explicit that providing a feed “does not guarantee that your content will be included.”

Here’s the part that trips almost everyone up: this is a different file from the main product feed that carries your prices, titles, and availability. Your store actually maintains two separate things: Evidence for this claim Product Ratings review feeds are separate from the primary product data source. Scope: This concerns Google Product Ratings, not on-page review markup or store ratings. Confidence: high · Verified: 2026-07-14 Google: Product Ratings basics

1. The product feed — price, title, image, availability, 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.. (That’s covered in the 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 optimization piece in this same cluster.)
2. The product review feed — the actual customer reviews and their star ratings.

Google and Bing then match reviews to products using a shared ID — usually the GTIN (the barcode number), or the Brand + MPN, or the product’s URL. If those IDs don’t line up between your two feeds, Google may not match the review to the intended product. Evidence for this claim Google uses product identifiers and product URLs in review feeds to match reviews to products. Scope: A review can fail to match when identifying data is missing, invalid, or inconsistent. Confidence: high · Verified: 2026-07-14 Google Product Review Feeds: Schema

The other thing people confuse it with

If you’ve added a reviews widget to your product pages — the kind that shows stars right on the page — that is schema markupSchema markup is code that uses the schema.org vocabulary to label what your content means so search engines can understand it and show rich results. It's most often written in JSON-LD, and it's not a direct ranking factor. (Review/AggregateRating), and it’s a completely separate thing. Google’s own docs describe that markup as producing organic review 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 Knowledge PanelThe Knowledge Graph is Google's database of entities — people, places, organizations, and things — and the factual relationships between them. It's separate from any single website's structured data: your schema markup is one of many possible inputs to the graph, not the graph itself. display — they don’t document any path from on-page schema into Google Shopping’s Product Ratings program, which draws from submitted review feeds, approved aggregators, review sites, and Google users instead. I’ll say that again because it’s the single biggest myth here: adding review schemaReview schema (schema.org/Review) is structured data for a single critic's or user's evaluation of one specific thing — one author, one itemReviewed, one reviewRating — distinct from AggregateRating, which summarizes many reviews into an average. to your pages is not how you get Shopping stars — submitting a review feed is.

How you actually get the stars

The short version:

• On Google: either build and submit your own review feed (after signing up through Google’s Product Ratings Interest Form), or use an approved reviews company (Yotpo, Judge.me, Trustpilot, Bazaarvoice, Okendo, and many more) that sends the feed for you. Google needs a minimum of 50 reviews across your products before stars show at all.
• On Microsoft/Bing: submit a Google-formatted review file over FTP/SFTP — or, if your store already runs reviews through Bazaarvoice, you skip the feed entirely and Bing surfaces the ratings automatically once your product IDs are accurate.

The honesty rules

You can’t just send your good reviews. Both Google and Bing want the full, honest set. Google explicitly says you can’t submit only your 5-star reviews. You also can’t quietly delete old bad reviews from the feed. Incentivized reviews (where you gave someone something to leave a review) are allowed, but only if you disclose them and the reward wasn’t tied to leaving a positive review.

Want the full XML schema, the “full feed replaces everything” landmine, and the Bing-specific file-naming rules? Switch to the Advanced tab.

TL;DR — A product review feedA product review feed is a structured XML file (following a Google-defined schema) that merchants or approved aggregators submit to Google Merchant Center — and separately to Microsoft Merchant Center — to populate the 1–5 star ratings and review counts shown on Shopping ads and free product listings. It is a different file from the main product feed, and different from Review/AggregateRating schema markup. is a standalone XML file (Google schema, v2.4 at time of writing) that populates star ratings in Shopping ads and 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. — a different file from your product feed, and different from Review/AggregateRating schema (organic-only) and from Google’s algorithmic “reviews system.” Google needs ~50 reviews + Interest Form signup, always expects a full feed (omitted reviews get deleted), and matches reviews to products by 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., Brand+MPN, or product URL — the #1 failure point. Microsoft reuses Google’s XML but adds its own ProductReviews file-naming and FTP/SFTP transport, and auto-surfaces BazaarVoice ratings. No cherry-picking, incentivized reviews must be disclosed and sentiment-independent, and AI-generated reviews must be flagged as spam.

First: three different things all called “reviews”

Most of the confusion in this topic is definitional. Keep these four systems apart and almost everything else falls into place:

1. Product review feed → Product Ratings — the XML feed you submit to Merchant Center that drives per-product star ratings in Shopping ads and free listings. Google compiles Product Ratings from merchants, review aggregators, review sites, and Google users (Product Ratings basics). Evidence for this claim Google compiles Product Ratings from multiple review sources for eligible Shopping placements. Scope: Ratings and review counts are not guaranteed to display. Confidence: high · Verified: 2026-07-14 Google: Product Ratings basics This article is about this.
2. Merchant review feed → Store/Seller Ratings — a separate, seller-level feed (Merchant Review Feeds) that drives the business-level rating. Store Ratings reflect the whole business’s post-fulfillment experience, not individual products, and need a 3.5+ average to display (Store ratings overview).
3. Review/AggregateRating schema markupSchema markup is code that uses the schema.org vocabulary to label what your content means so search engines can understand it and show rich results. It's most often written in JSON-LD, and it's not a direct ranking factor. — on-page 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. that only affects organic review rich results and has no effect on Shopping. (Note the self-serving restriction: pages where the entity being reviewed controls the reviews about itself are ineligible for the star rich result — so marking up your own on-site reviews won’t reliably earn organic stars, and definitely won’t earn Shopping stars.)
4. Google’s “reviews system” (the Product Reviews Update) — an algorithmic ranking system for first-party editorial review content (blog posts and articles that review products). Per Google, it does not evaluate third-party reviews such as those posted in the reviews section of a product page. It has nothing to do with 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. feeds — despite sharing the word “review.”

I’ll cover the review-feed vs. schema distinction, and the Product Reviews Update mix-up, again in the Anti-Patterns tab because they’re the two mistakes I see most.

Why the feed matters

Star ratings and review counts are the most visible trust signal you can add to a Shopping ad or free product listing, and they sit right in the grid where the click decision happens. Getting them is a feed-and-eligibility exercise, not a ranking one — which is exactly why it gets neglected. It’s plumbing that pays off in CTR.

Google product review feeds

Eligibility

• Minimum 50 reviews across all products before ratings display (Product Ratings eligibility).
• You must sign up through the Product Ratings Interest Form first.
• Your review data source has to be uploaded at least once a month and contain updated reviews — it’s an ongoing requirement, not a one-and-done upload.
• The domains in the product_url of your review data source must match the registered domain on your Merchant Center account.
• Onboarding takes a few weeks; status shows “inactive” until approved, then “active,” and subsequent updates take roughly 7–10 days to reflect. Plan client expectations around that lag.

These are prerequisites to participate, not a promise of display. Google’s own docs state that providing a feed “does not guarantee that your content will be included” — meeting the 50-review threshold and getting your Interest Form approved make you eligible; validation, matching, and moderation still have to succeed before any star shows up.

Direct feed vs. approved aggregator

Two routes (How to submit product reviews data):

1. Submit your own feed — build a compliant XML file and submit it directly (needs Interest Form approval). Full control, but you own the schema literacy and the monthly full-feed discipline.
2. Go through an approved review aggregator/partner — Bazaarvoice, Trustpilot, Yotpo, Feefo, PowerReviews, Judge.me, Loox, Okendo, Stamped, and many others feed Google on your behalf. Less work, but it costs money and coverage varies by country, so confirm regional availability with the aggregator.

There’s a decision tree for choosing between them in the Decision Trees tab.

XML schema essentials

The feed is a validated XML file — schema version 2.4 at time of writing (XML Schema Reference). The elements worth knowing:

• Feed-level: <feed version="2.4">, <publisher> (required), <reviews>, and an optional <aggregator> block.
• Per review: <review_id> (a permanent unique ID), <review_timestamp>, <reviewer><name> (supports is_anonymous), <content> (the review text), plus optional <title>, <review_language> (BCP 47), and <review_country> (ISO 3166-1).
• Trust/quality flags: <is_verified_purchase>, <is_incentivized_review>, <is_spam>, and <collection_method> (“unsolicited” or “post_fulfillment”), plus an optional <transaction_id>.
• Ratings: <ratings><overall> with min/max attributes.
• Qualitative: <pros>/<cons> — Google asks you to omit boilerplate like “pro:”.
• Review URL: <review_url> with a type of “singleton” (a page showing just this review) or “group” (a page listing several).
• Product association: <products><product> (one or more), each with a <product_url>, optional <product_name>, and a <product_ids> container holding <gtins>, <mpns>, <skus>, and <brands> — each required whenever that identifier is known for the product (<mpns> must be paired with <brands>), plus a genuinely optional <asins>.

Matching itself is hierarchical, not a flat list of equal options. Google’s schema reference says explicitly: ratings and reviews match “based on a number of factors, the strongest being globally 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. such as GTINs,” and without a GTIN, Google will attempt SKU, Brand+MPN, or product URL — but “these identifiers generally do not provide as good a match as GTINs.” Provide the GTIN whenever you have one; treat the alternatives as fallbacks, not equivalents. Evidence for this claim Google's product review feed schema supports GTIN, MPN plus brand, SKU, ASIN, and product URL identifiers. Scope: Submit identifiers only when valid and applicable to the reviewed product. Confidence: high · Verified: 2026-07-14 Google Product Review Feeds: Schema

Submission methods and file limits

• Upload by scheduled fetch (http/httpsHTTPS is the encrypted version of HTTP — it uses TLS to authenticate the server and protect data in transit between a browser and a website. Google announced it as a lightweight ranking signal in 2014 and today conditionally prefers HTTPS pages as canonical; Chrome marks plain HTTP pages 'Not Secure.'/sftp), SFTP, Google Cloud Storage, or a direct upload from your computer (Upload a product reviews data source).
• Under 4 GB per data source; multiple XML files can be zipped.

The “full feed replaces everything” gotcha

This is the landmine. Google’s system always expects a full feed, and any review previously submitted but not included in your next full feed is deleted from Google’s submitted-review dataset (not from your own reviews platform — your source data is untouched, but Google’s copy is gone). Teams that build an incremental-only pipeline — “just send the new reviews” — silently wipe out their entire review history on Google’s side with the next upload. Send the full, current set every time.

Content policy

The Product Ratings policies are stricter than most people expect:

• Authenticity: Google removes reviews it believes have a conflict of interest or were written by employees or people with a vested interest in the product.
• No cherry-picking, no deletion: submit the full honest set — you can’t include only 5-star reviews, and you may not delete old reviews from the feed.
• Incentivized reviews are allowed if disclosed via <is_incentivized_review>, and the incentive can’t be dependent on the sentiment of the review — pay-for-a-review is fine, pay-only-if-positive is not.
• AI-generated reviews must be marked as spam (<is_spam>). Google calls out automated content explicitly — worth flagging in 2026 as AI review generators proliferate.
• Prohibited content includes PII, obscene/hateful material, regulated/dangerous products, counterfeit goods, malware/phishing links, plagiarized content, and impersonation. Reviews should be owned by the retailer, not syndicated across unrelated feeds.

Microsoft (Bing) product review feeds

Microsoft Merchant CenterMicrosoft's free platform (inside Microsoft Advertising, formerly 'Bing Merchant Center') where retailers create a store and submit a product feed. One feed powers both free Product Listings on the Bing Shopping tab and paid Microsoft Shopping Campaigns — the Bing/Copilot-side twin of Google Merchant Center. As of 2026 it's also the product-data foundation Copilot draws on for AI-assisted shopping and checkout. runs a parallel but separately-administered system. A note on sourcing: Microsoft’s own help pages are client-rendered and return no usable content to automated fetchers (confirmed again as of this update pass — the pages resolve to empty search-result shells rather than the documented content), so the specifics below are corroborated through official Microsoft Advertising support responses in Microsoft Q&A threads rather than first-party doc permalinks. Treat the figures below (review thresholds, file-size limits, transport details) as current-as-of-research rather than independently confirmed against Microsoft’s primary documentation, and verify against the live UI before you implement.

Merchant Ratings vs. Product Ratings

Microsoft splits ratings the same way Google does:

• Merchant Ratings — business/seller-level, shown automatically as an automated extension on Search campaigns once you’re eligible. No manual submission. Sourced from third-party review providers, and eligibility runs on a minimum of 30 unique reviews within the past 12 months per country at a composite 3.5+ stars, with your business name and website consistent across sources. Stars typically take 3–6 weeks to appear after you meet the requirements (Microsoft Q&A support thread, REVIEWS.io — Bing seller ratings requirements).
• Product Ratings — product-level, an automated extension on Shopping campaigns, with two paths below.

The BazaarVoice auto-integration path

If your site’s reviews already run through BazaarVoice, you don’t submit a separate product review feed at all — you just provide accurate product identifiers (GTIN, Brand, MPN) in your shopping feed, and your product ads display the BazaarVoice ratings automatically. That’s a genuinely different, lighter path than Google offers.

Direct feed submission

For everyone else, submit your own XML feed (Microsoft Q&A — feed XML schema and SFTP guide):

• The file must be an existing Google-formatted .xml file — Microsoft reuses Google’s product-review-feed schema rather than defining its own. A well-built Google feed is most of the way to Bing-ready.
• File naming matters and is a common failure point: the file must start with ProductReviews (full feed, all inventory, submitted daily) or ProductReviewsupdate (update-only, just new reviews). The file name you configure in Merchant Center settings must match the uploaded file exactly and without the extension (configure “merchant”, not “merchant.xml”). For compressed uploads, the inner file and the archive must share the same name, one file per archive.
• Transport: SFTP to ftps://feeds.adcenter.microsoft.com in passive mode, using your store’s FTP credentials.
• File size: max 1 GB via FTP/SFTP (recommended for files over 4 MB), up to 3 GB after decompression if zipped.
• Requirement to display: a product needs at least 1 review.

The #1 Bing failure mode

A separate Microsoft Q&A thread documents a merchant who uploaded a valid XML over FTP and still saw nothing weeks later — and the support troubleshooting centered on the exact file-name match between the uploaded file and the configured feed name, plus the size/compressionCompression (HTTP content encoding) shrinks text-based responses — HTML, CSS, JS, JSON, SVG, XML sitemaps — before they're sent over the network, using an algorithm like Gzip, Brotli, or Zstd, so the browser or crawler downloads fewer bytes. It's not a ranking factor, but it speeds up page loads and helps pages stay under crawler fetch limits. rules. If your Bing feed isn’t appearing, check the file name first. It’s the Bing equivalent of Google’s GTIN-matching problem.

Keeping product and review feeds in sync

This is the whole game once you’re live. Reviews attach to products primarily through GTIN — the strongest match. When a GTIN is missing or inaccurate, Google will attempt SKU, Brand+MPN, or product URL, but by Google’s own description those alternatives “generally do not provide as good a match as GTINs” — don’t treat them as equally reliable fallbacks. Identifier mismatches are the top cause of “reviews submitted but stars not showing.” If the GTIN in the review feed doesn’t match the GTIN in the product feed (reused, wrong, or missing GTINs are the usual culprits), the reviews simply don’t attach. That’s the same root cause behind the Merchant Center identifier errors covered in the feed optimization piece in this cluster — get your main feed’s GTINs right first, then mirror them exactly in the review feed.

It’s also worth separating the stages a submission passes through, since success at one doesn’t prove success at the next: your XML has to be well-formed and schema-valid, then the feed has to be accepted by Merchant Center, then individual reviews have to clear policy moderation, then reviews have to match a product by identifier, then matched reviews get aggregated into a rating, and only then — assuming your account meets program eligibility — does a rating actually display. A clean upload confirms the first stage or two; it says nothing about the rest.

Timing to communicate to stakeholders: 2–4 weeks for Google’s initial onboarding, 7–10 days for subsequent Google syncs, and 3–6 weeks for Bing merchant ratings to appear once you’re eligible.

Where this sits in the cluster

Everything above assumes you’ve already got a working Merchant Center account and a clean main product feed — the 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. hub and the feed optimization deep dive in this cluster cover those. This piece owns the one thing they don’t: the separate review feed that turns your customer reviews into Shopping stars.

AI summary

A condensed take on the Advanced version:

• A product review feedA product review feed is a structured XML file (following a Google-defined schema) that merchants or approved aggregators submit to Google Merchant Center — and separately to Microsoft Merchant Center — to populate the 1–5 star ratings and review counts shown on Shopping ads and free product listings. It is a different file from the main product feed, and different from Review/AggregateRating schema markup. is a standalone XML file (Google schema, v2.4 at time of writing) that drives per-product star ratings in Shopping ads and 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..
• Four things share the word “review” — keep them apart: the product review feed (Product Ratings, per-product), the merchant review feed (Store/Seller Ratings, business-level), Review/AggregateRating schema (organic 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. only), and Google’s algorithmic “reviews system” / Product Reviews Update (ranks editorial review content, unrelated to feeds).
• Schema markupSchema markup is code that uses the schema.org vocabulary to label what your content means so search engines can understand it and show rich results. It's most often written in JSON-LD, and it's not a direct ranking factor. ≠ Shopping stars. On-page review stars come from 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. and only affect organic results; Shopping stars require the feed.
• Google eligibility: ~50 reviews minimum, Interest Form signup, at least monthly uploads, product_url domain must match the 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. domain — these are prerequisites to participate, not a guarantee of display.
• Full feed always replaces: any review not in your latest full upload is deleted from Google’s submitted dataset (your own review platform is unaffected) — incremental-only pipelines silently wipe review history on Google’s side.
• Matching is by 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., or Brand+MPN, or product URL — mismatches are the #1 reason stars don’t show. GTIN is the strongest match.
• Policy: no cherry-picking (can’t submit only 5-star reviews), no deleting old reviews, incentivized reviews must be disclosed and sentiment-independent, AI-generated reviews must be flagged as spam.
• Microsoft/Bing reuses Google’s XML but adds its own ProductReviews file-naming and FTP/SFTP transport (ftps://feeds.adcenter.microsoft.com, 1 GB max), needs only 1 review per product to display, and auto-surfaces ratings for BazaarVoice-integrated stores. Exact file-name matching is the #1 Bing failure point.
• Direct feed vs. aggregator is a real strategic choice: control vs. convenience/cost.
• Timing: Google onboarding 2–4 weeks, updates 7–10 days; Bing merchant ratings 3–6 weeks once eligible.

Official documentation

Primary-source documentation from Google and Microsoft.

Google

• Product Review Feeds — Introduction — what the feed is and who can submit it.
• Product Review Feeds — XML Schema Reference — the full v2.4 element reference and the “full feed replaces everything” rule.
• Merchant Review Feeds — Introduction — the separate, seller-level feed (Store/Seller Ratings).
• Upload a product reviews data source to Merchant Center — submission methods.
• How to submit product reviews data — direct feed vs. approved aggregator.
• Product Ratings basics — what Product Ratings are and where they come from.
• Product Ratings eligibility — the 50-review minimum and onboarding.
• Product Ratings policies — authenticity, no cherry-picking, incentivized-review disclosure, AI/spam flagging.
• Store ratings overview — the business-level counterpart.
• Review Snippet (Review, AggregateRating) structured data — the organic-only markup this feed is not.
• Google Search’s Reviews System — the algorithmic ranking system, distinct from 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. feeds.

Microsoft / Bing

• Microsoft Q&A — Product Reviews Feed XML schema and the SFTP submission guide — official support response covering file naming, size, and SFTP transport.
• Microsoft Q&A — Product reviews feed uploaded via FTP not appearing — the exact file-name-match failure mode.
• Microsoft Q&A — How to get review stars in Bing merchant for product ads — merchant vs. product ratings paths.

Quotes from the source

On-the-record statements from Google’s documentation. Each deep link jumps to the quoted passage on the source page.

Google — Product Ratings policy

• “Reviews are most valuable when they are honest and unbiased. We remove reviews that we believe have conflicts of interest and/or have been written by employees or people with a vested interest in the product.” — 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. Help, Product Ratings policies. Jump to quote

Which path should you take?

Two decisions do most of the work: how you submit on Google, and which system you even need. Walk the tree.

Mistakes I see over and over

1. Expecting on-page review schema to produce Shopping stars

The mistake: Adding Review/AggregateRating markup to product pages and assuming the stars will show up in 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..

Why it’s wrong: Review SnippetReview schema (schema.org/Review) is structured data for a single critic's or user's evaluation of one specific thing — one author, one itemReviewed, one reviewRating — distinct from AggregateRating, which summarizes many reviews into an average. 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. only affects organic review rich results. It has no impact on Google Shopping surfaces. Shopping stars come exclusively from the separate Merchant Center product review feedA product review feed is a structured XML file (following a Google-defined schema) that merchants or approved aggregators submit to Google Merchant Center — and separately to Microsoft Merchant Center — to populate the 1–5 star ratings and review counts shown on Shopping ads and free product listings. It is a different file from the main product feed, and different from Review/AggregateRating schema markup. / Product Ratings program.

Do instead: Keep the schema for your organic listings if you like, but build and submit an actual review feed (or use an approved aggregator) to earn Shopping stars. They’re two independent pipelines.

2. Blaming the “Product Reviews Update” for missing Shopping stars

The mistake: Thinking Google’s 2021+ Product Reviews Update (the “reviews system”) governs whether your Merchant Center review feed shows stars.

Why it’s wrong: The reviews system is an algorithmic ranking system for first-party editorial review content — articles and blog posts that review products. Google states it does not evaluate third-party reviews such as those in the reviews section of a product page. It’s unrelated to Merchant Center feeds.

Do instead: Debug feed problems as feed problems (eligibility, identifiers, full-feed discipline), not as an algorithmic ranking issue.

3. Building an incremental-only feed pipeline

The mistake: Sending “just the new reviews” each time to save bandwidth or dev effort.

Why it’s wrong: Google’s system always expects a full feed. Any review previously submitted but not in your next full upload is deleted from Google’s submitted dataset (your own review platform is unaffected). An incremental-only pipeline silently wipes your review history on Google’s side.

Do instead: Regenerate and send the complete, current review set every upload, at least monthly.

4. Cherry-picking only positive reviews

The mistake: Submitting only the 5-star reviews to inflate the average.

Why it’s wrong: Google policy requires the full, honest set — you explicitly cannot include only 5-star reviews, and you may not delete old reviews from the feed. Selective submission is a policy violation.

Do instead: Submit everything you honestly collected. Fix the product, not the feed.

5. Paying only for positive reviews (or hiding that you incentivized at all)

The mistake: Offering a discount only if the customer leaves a positive review, or including incentivized reviews without flagging them.

Why it’s wrong: Incentivized reviews are allowed, but only when disclosed via <is_incentivized_review> and only when the incentive isn’t contingent on sentiment.

Do instead: Offer the incentive for any honest review, and mark those reviews with the incentivized flag.

6. Letting review-feed and product-feed identifiers drift apart

The mistake: Assuming reviews will attach as long as both feeds “have the product.”

Why it’s wrong: Reviews match to products by 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. (or Brand+MPN, or product URL). A mismatched, reused, or missing GTIN means the reviews never attach and no stars appear — the single most common “I submitted reviews but nothing shows” cause.

Do instead: Get your main product feed’s GTINs correct first (see the feed optimization piece in this cluster), then mirror them exactly in the review feed.

7. Getting the Bing file name wrong

The mistake: Uploading a valid XML to Microsoft over FTP but naming it something that doesn’t match the configured feed name (or leaving the extension in the settings field).

Why it’s wrong: Microsoft requires the file to start with ProductReviews and to match the configured feed name exactly, without the extension. A name mismatch means the feed never processes — and support threads show merchants losing weeks to exactly this.

Do instead: Name the file ProductReviews.<yourname>.xml, configure the feed name without the extension, and verify the two match before you wait.

A minimal Google product review feed skeleton

Enough valid v2.4 XML to see how the pieces fit. Fill in your own real values — never invent 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., review text, or IDs. This is a scaffold, not a copy-paste feed.

<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns:vc="http://www.w3.org/2007/XMLSchema-versioning"
      xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
      version="2.4">
  <publisher>
    <name>Your Store Name</name>
    <favicon>https://www.example.com/favicon.ico</favicon>
  </publisher>
  <reviews>
    <review>
      <review_id>REVIEW-UNIQUE-PERMANENT-ID</review_id>
      <reviewer>
        <name is_anonymous="false">Real Reviewer Name</name>
      </reviewer>
      <review_timestamp>2026-06-01T12:00:00Z</review_timestamp>
      <content>The actual review text goes here.</content>
      <pros>
        <pro>Concise pro, no "pro:" boilerplate</pro>
      </pros>
      <cons>
        <con>Concise con</con>
      </cons>
      <review_url type="singleton">https://www.example.com/reviews/REVIEW-ID</review_url>
      <ratings>
        <overall min="1" max="5">5</overall>
      </ratings>
      <is_verified_purchase>true</is_verified_purchase>
      <is_incentivized_review>false</is_incentivized_review>
      <collection_method>post_fulfillment</collection_method>
      <products>
        <product>
          <product_ids>
            <gtins>
              <gtin>REAL-GTIN-HERE</gtin>
            </gtins>
            <brands>
              <brand>Your Brand</brand>
            </brands>
          </product_ids>
          <product_name>Product Title</product_name>
          <product_url>https://www.example.com/products/the-product</product_url>
        </product>
      </products>
    </review>
  </reviews>
</feed>

The exact allowed elements, cardinality, and attributes are in Google’s XML Schema Reference — treat that page, not this skeleton, as canonical.

Validate your XML is well-formed before you upload

Schema validity is Google’s job, but catching a broken XML file locally saves an upload cycle.

macOS / Linux

# Checks the file is well-formed XML (not schema-valid, just parseable)
xmllint --noout ProductReviews.example.xml && echo "well-formed"

Windows (PowerShell)

# Throws if the XML is malformed; prints "well-formed" otherwise
try { [xml](Get-Content -Raw .\ProductReviews.example.xml) | Out-Null; "well-formed" }
catch { "MALFORMED: $($_.Exception.Message)" }

Confirm review and product feeds share GTINs

The #1 sync failure is a GTIN in the review feed that isn’t in the product feed. Pull the GTINs from each and diff them.

macOS / Linux

# Extract <gtin> values from the review feed
grep -oE '<gtin>[^<]+</gtin>' ProductReviews.example.xml \
  | sed -E 's|</?gtin>||g' | sort -u > review-gtins.txt

# Compare against the GTIN column of your product feed (adjust column as needed)
# Lines showing only in review-gtins.txt are reviews that will fail to attach.
comm -23 review-gtins.txt product-gtins.txt

Any GTIN that appears only in the review feed is a review that won’t find its product — fix it in the source before the next full upload.

SOP: maintain a full product review feed

Run this at least monthly for Google, and on the submission schedule configured for each destination. The pipeline must rebuild the complete current review set rather than append only the latest reviews.

1. Freeze the source window. Export the complete, current review corpus with permanent review IDs. Done means old valid reviews and new reviews are both present, with no sentiment-based filtering.
2. Apply policy flags. Preserve verified-purchase, incentive, spam, collection-method, and other source flags. Done means incentives are disclosed and AI-generated review content is marked as spam rather than presented as customer testimony.
3. Join products using stable identifiers. Match 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. first when known, otherwise the approved brand-plus-MPN or product URL. Done means review identifiers agree with the live product feed and destination domain.
4. Generate the full XML. Include publisher, reviews, ratings, review URLs, and product associations under the approved schema. Done means every review intended to remain active is in the new file.
5. Validate before transport. Parse the XML and run the article’s review-feed/product-feed identifier comparison. Done means the file is well-formed and unmatched identifiers are corrected or intentionally held.
6. Name and deliver per destination. Submit Google through the configured data source. For Microsoft direct submission, use the required ProductReviews or ProductReviewsupdate prefix and match the configured name exactly. Done means the destination accepts the file and the uploaded filename is recorded.
7. Reconcile processing. Review destination status, counts, errors, and product attachment after processing. Done means an unexpected review-count drop or unmatched-product cohort has an owner before the next run.

Reviews are submitted but Shopping stars do not appear

Symptom: The feed processes, but product listings show no ratings. Likely cause: Review-feed 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., brand-plus-MPN, or product URL values do not match the main product feed; Google eligibility or onboarding may also be incomplete. Fix: Diff the identifiers, correct the source join, confirm the registered domain and program status, then allow for the documented processing/onboarding lag. Validate attachment by product instead of assuming a successful file upload is enough.

Microsoft accepts the transfer but never processes the feed

Symptom: A valid XML file is present on the server, yet Microsoft shows no product ratings data. Likely cause: The file prefix or configured feed name does not match the uploaded file exactly, or archive/inner filenames differ. Fix: Use the documented ProductReviews or ProductReviewsupdate prefix, configure the name without the extension, and make compressed and inner file names agree. Confirm processing status after the corrected upload.

Historical reviews disappear after an upload

Symptom: The review count drops to only the newest batch. Likely cause: An incremental file was sent to Google’s full-feed pipeline, where omitted reviews are treated as deleted from Google’s submitted dataset (your own review platform still has them). Fix: Rebuild and submit the complete current review set using the permanent review IDs. Confirm the next processed feed restores the expected corpus; do not keep an incremental-only Google pipeline.

Reviews attach to the wrong product or variant

Symptom: Ratings appear on an unintended item, or sibling variants have inconsistent counts. Likely cause: A reused GTIN, shared MPN, or stale product URL maps reviews to the wrong catalog entity. Fix: Reconcile identifiers against packaging/manufacturer data and the current product feed, then correct the review source mapping. Confirm the affected review IDs attach only to the intended product.

Google Product Ratings readiness

• At least 50 reviews across all products are available before expecting ratings to display.
• Product Ratings Interest Form submitted and program status tracked through onboarding.
• Review source scheduled at least monthly with updated data.
• Each upload is the full current feed, not only new reviews.
• product_url domains match the registered 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. website.
• 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., or brand plus MPN, or product URL matches the main product feed.
• Review IDs are permanent and the XML is well-formed under the current supported schema.
• Full, honest review set included; no positive-only selection or silent deletion.
• Incentivized reviews disclosed and independent of sentiment; AI-generated reviews flagged as spam.

Microsoft direct-submission checks

• Determine whether BazaarVoice integration already removes the need for a separate feed.
• Reuse a valid Google-format XML file when submitting directly.
• Filename starts with ProductReviews for a full feed or ProductReviewsupdate for an update feed.
• Configured feed name matches the upload exactly and omits the extension.
• Compressed archive and inner XML share the required name and contain one file.
• SFTP credentials, passive transfer, and current destination settings confirmed in the live account.
• 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. in the review feed match the Microsoft shoppingMicrosoft's free platform (inside Microsoft Advertising, formerly 'Bing Merchant Center') where retailers create a store and submit a product feed. One feed powers both free Product Listings on the Bing Shopping tab and paid Microsoft Shopping Campaigns — the Bing/Copilot-side twin of Google Merchant Center. As of 2026 it's also the product-data foundation Copilot draws on for AI-assisted shopping and checkout. feed.
• Processing status checked before waiting for stars to appear.

The four review systems

Product review feed → Product Ratings

This is product-level XML sent to a commerce platform. It powers per-product stars in Shopping surfaces and matches reviews to products through identifiers or product URLs.

Merchant review feed → Store/Seller Ratings

This measures the business and post-fulfillment experience, not an individual product. It is a separate data pipeline with separate eligibility.

Review structured data → organic rich results

Review and AggregateRating markup live on web pages and relate to organic review snippetsReview schema (schema.org/Review) is structured data for a single critic's or user's evaluation of one specific thing — one author, one itemReviewed, one reviewRating — distinct from AggregateRating, which summarizes many reviews into an average.. Adding them does not submit product ratings to 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..

Google’s reviews system → editorial review rankings

The search ranking system evaluates first-party editorial product-review content such as articles. It does not decide whether third-party customer reviews in a Merchant Center feed earn Shopping stars.

When stars are missing, identify the system first. Debug eligibility, transport, identifiers, and full-feed behavior for Product Ratings; do not reach for schema or algorithm-update explanations that belong to another pipeline.

Product review feed quick reference

Google XML essentials

Submission behavior

• Google: full current feed every time; omitted reviews are deleted.
• Google cadence: upload updated data at least monthly.
• Google transport: scheduled fetch, SFTP, Cloud Storage, or direct upload.
• Microsoft direct: Google-format XML over the configured transport.
• Microsoft filename: ProductReviews… for full inventory or ProductReviewsupdate… for updates; configured name omits the extension.

Google vs. Microsoft

First diagnostic: successful upload ≠ successful product attachment. Check identifiers and destination processing before waiting.

Test yourself: Product Review Feeds

Five quick questions on how product review feedsA product review feed is a structured XML file (following a Google-defined schema) that merchants or approved aggregators submit to Google Merchant Center — and separately to Microsoft Merchant Center — to populate the 1–5 star ratings and review counts shown on Shopping ads and free product listings. It is a different file from the main product feed, and different from Review/AggregateRating schema markup. earn star ratings in Google and Microsoft Shopping. Pick an answer for each, then check.

Product Review Feed

A product review feed is a structured XML file (following a Google-defined schema) that merchants or approved aggregators submit to Google Merchant Center — and separately to Microsoft Merchant Center — to populate the 1–5 star ratings and review counts shown on Shopping ads and free product listings. It is a different file from the main product feed, and different from Review/AggregateRating schema markup.

Related: Google Merchant Center, Google Merchant Center Feed Optimization

Product Review Feed

A product review feed is a 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. file — XML, following a Google-defined schema — that merchants or approved review aggregators submit to 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 separately to Microsoft Merchant CenterMicrosoft's free platform (inside Microsoft Advertising, formerly 'Bing Merchant Center') where retailers create a store and submit a product feed. One feed powers both free Product Listings on the Bing Shopping tab and paid Microsoft Shopping Campaigns — the Bing/Copilot-side twin of Google Merchant Center. As of 2026 it's also the product-data foundation Copilot draws on for AI-assisted shopping and checkout., to populate the 1–5 star ratings and review counts shown on Shopping ads and 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.. Reviews in the feed are matched to catalog products through shared identifiers (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., or Brand + MPN, or product URL).

It is not the same thing as three commonly confused neighbors. It is not the main product feed, which carries price, availability, GTIN, and title. It is not Review/AggregateRating schema.orgSchema markup is code that uses the schema.org vocabulary to label what your content means so search engines can understand it and show rich results. It's most often written in JSON-LD, and it's not a direct ranking factor. markup on a product page, which only affects organic review 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 has no effect on Shopping surfaces. And it is not the same as Google’s algorithmic “reviews system” (the Product Reviews Update), which ranks first-party editorial review content, not customer reviews in a feed. Google’s own framing is that these feeds let review publishers provide product reviews to Google, but providing a feed does not guarantee inclusion.

Related: Google Merchant Center, Google Merchant Center Feed Optimization

Changelog