Merchant Feed Checker and Fixer for Google Merchant Center

Check a Google Merchant Center product feed before you upload it. Give the actor a feed link, upload the file, or paste it; it returns one row per item listing each problem found against Google's published product data specification, plus a corrected copy of your feed with only safe, mechanical fixes applied and a change log of every value it changed.

It never invents product data: titles, GTINs, brands, prices, currencies and categories are never made up or guessed. Anything that needs your own product information is reported, not changed.

Supported feeds

• Tab-delimited text (.txt / .tsv) and comma-delimited CSV with a header row of attribute names (id, title, image_link or image link, with or without g:). Header names wrapped in single quotes ('id') are rejected with an explanation, because Google would not recognise them.
• XML: RSS 2.0 or Atom 1.0 with the Google namespace xmlns:g="http://base.google.com/ns/1.0". An Atom feed that leaves out the Atom namespace is still checked, with a warning. XML that is not well-formed is rejected with the line and column of the error (for a bare &, which must be written &, the message says so); it is not repaired.
• Values the checker does not fix are written back exactly as they were read, including Windows line endings inside descriptions.
• Plain, .gz, or a .zip holding one feed file. The corrected feed is written in the same format (UTF-8).

What it checks

Rule (dataset rule)	What	Fixed automatically?
missing-required	id, title, description, link, image_link, availability, price present and non-empty	No
too-long	id > 50, title > 150, description > 5,000, brand > 70, mpn > 70, additional_image_link > 2,000 characters	No (never truncated)
duplicate-id	the same id on more than one item	No
availability-value	must be in_stock, out_of_stock, preorder or backorder	Yes for spelling variants ("In Stock", "out of stock", schema.org InStock); others reported
condition-value	must be new, refurbished or used	Yes for case/schema.org variants; others reported
identifier_exists-value	yes or no	Yes for TRUE/FALSE, Y/N
price-format	"15.00 USD" form: number, space, ISO 4217 code	Yes: currency moved after the amount, capitals, decimal comma "12,99" to "12.99", unambiguous thousands separators ("1,299.50", "1.299,00")
price-currency	currency code present and a real ISO 4217 code	No: a symbol such as $ or a missing code is reported, never guessed
price-zero, price-value	price of 0 (allowed only for some contract/subscription goods) or negative	No
price-ambiguous	warning for an amount with three digits after the point, like "1.299 EUR": Google keeps 2 decimal places and rounds the rest (1.299 becomes 1.30), and the point may be meant as a thousands separator	No
sale-price-currency, sale-price-not-lower	sale_price in the same currency as price and lower than it	No
missing-availability-date	availability_date required for preorder, asked for backorder	No
date-format, date-order	ISO 8601 dates (seconds and fractions of a second allowed); sale_price_effective_date start before end	No
gtin-format	digits only, no spaces or dashes; spreadsheet scientific notation (4.00638E+12)	Spaces/dashes removed only when the result is a valid 12-, 13- or 14-digit GTIN (length and check digit); anything else, such as an SKU or a web address in the GTIN column, is reported unchanged; scientific notation reported
gtin-length, gtin-check-digit	8, 12, 13 or 14 digits; GS1 check digit correct	No (a wrong GTIN is never "repaired")
gtin-restricted	restricted-circulation (02, 04, 2x) and coupon (98, 99) prefixes Google rejects	No
gtin-isbn10	ISBN-10 given as GTIN	Converted to ISBN-13 only when the item's google_product_category is Media > Books (784) or below and the ISBN-10 check digit is valid; any other 10-digit value is reported as gtin-length, not changed
missing-identifiers	no GTIN and not both brand and MPN (warning: depends on category)	No
identifier-exists-conflict	identifier_exists is no but a GTIN is given	No
brand-placeholder	brand values such as "N/A", "Generic", "No brand"	No
url-scheme	link / image_link start with http:// or https://	No
url-encoding	spaces and other characters that must be percent-encoded	Yes (percent-encoded)
image-format	image URL ending in a non-accepted format (.svg, .avif, .heic, .pdf, ...); a URL ending in .html or .htm is a warning (some image services serve images from such addresses)	No
category-unknown, category-format	google_product_category is a taxonomy ID or exact path (English, or one of Google's translated taxonomies); not both	Yes for "ID - path" (ID kept) and spacing/capitals of an exact path
missing-namespace-prefix	XML attribute element without the g: prefix (Google ignores it)	Yes (moved into the g: namespace)
whitespace	leading/trailing spaces, repeated spaces (not in description), control characters	Yes
unknown-attribute, extra-columns, encoding, missing-namespace	whole-file problems	Encoding written as UTF-8; others reported

What it does not check

• It does not guarantee approval. Google also reviews landing pages, images (size, watermarks, placeholders), policies, shipping and tax settings and your account; none of that is checked here.
• It does not visit your links or images, compare the feed with your website, or check category-specific rules (apparel size/colour/gender, energy labels, alcohol, and so on).
• It does not check shipping, tax, installment or other group attributes (left unchanged in the corrected feed).
• Google Sheets and Content API sources are not read; export them as a file first.

Input

• Feed URL or file: a direct link to the feed, or upload the file in the input form.
• Or paste the feed: for small feeds.
• Include items with no problems (default off), Maximum file size (default 100 MB) and Maximum dataset rows (default 1,000; 0 = no limit).

Large feeds

The whole feed is read into memory. Our measurements on generated test feeds with defects seeded at known positions (every seeded problem reported, exact counts per rule, nothing else; run on our own machine, 2026-09-25):

Feed	Items	File size	Run time	Peak memory
Tab-delimited text	50,000	38 MB	21 s	405 MB
RSS 2.0 XML	50,000	53 MB	20 s	490 MB
Tab-delimited text	100,000	76 MB	37 s	739 MB

Apify bills dataset writes per row (US$0.005 per 1,000 rows on its published pricing), which dominates the cost of a big feed with many problems: in our private test run on Apify of the 50,000-item text feed with every problem row written to the dataset (27,498 rows) the platform usage was US$0.14, of which US$0.137 was dataset writes and US$0.001 compute. That is why the dataset is capped at 1,000 rows by default; the complete lists are in ISSUES.csv and CHANGES.csv. The same feed with the default cap: 20 s, US$0.0086 platform usage (private run, 2026-09-25).

As a rule of thumb give the run about 10 to 12 times the file size in memory. The default run memory (1 GB) suits feeds up to about 80 MB; raise the run memory for bigger files. Run time on Apify can differ from these figures.

Output

Dataset, one row per item (items with no problem are left out unless you turn them on; at most Maximum dataset rows rows, default 1,000):

{"itemNumber": 3, "id": "SKU-003", "title": "Cast Iron Skillet 26 cm", "status": "error", "errors": 1, "warnings": 0, "fixed": 3,
 "problems": "FIXED availability: Normalised 'out of stock' to 'out_of_stock'.\nFIXED price: Reformatted price to '1299.00 EUR' (converted decimal comma).\nERROR gtin: GTIN '4006381333932' has a wrong check digit. ...",
 "issues": [{"attribute": "gtin", "severity": "error", "rule": "gtin-check-digit", "message": "...", "value": "4006381333932", "fixed": false, "source": "..."}]}

status is error (open errors left), warning, fixed (only safe fixes, now clean) or ok. Whole-file problems appear as item 0.

Key-value store

• CORRECTED_FEED.txt / .csv / .xml: your feed with only the fixes marked FIXED applied. Review CHANGES.csv before uploading it.
• CHANGES.csv: item number, id, attribute, rule, old value, new value for every change.
• ISSUES.csv: every open problem (item number, id, attribute, severity, rule, message, value), never capped.
• REPORT.md: summary, counts by rule and the items with problems.
• OUTPUT: JSON totals.

Sources (accessed 2026-09-25)

• Google Merchant Center Help, Product data specification: https://support.google.com/merchants/answer/7052112 (required attributes, lengths, supported values, price/ISO 4217 format, GTIN rules, brand placeholders, image formats, google_product_category ID-or-path rule).
• Google Merchant Center Help, RSS 2.0 specification: https://support.google.com/merchants/answer/160589 (g: namespace; un-prefixed attributes are ignored).
• Google product taxonomy with IDs (version 2021-09-21 at the time of access): https://www.google.com/basepages/producttype/taxonomy-with-ids.en-US.txt (downloaded when the actor is built, not bundled).
• ISO 4217 currency codes, list one published 2026-09-17: https://www.six-group.com/dam/download/financial-information/data-center/iso-currrency/lists/list-one.xml
• GS1, How to calculate a check digit manually: https://www.gs1.org/services/how-calculate-check-digit-manually

Google can change its specification at any time; each dataset issue names its source so you can check it.

Privacy

Your feed is read inside the run and the results are written to your own Apify storage. Nothing is sent anywhere else, and no links or images in the feed are visited.

Support

Open an issue on the actor's Issues tab. This actor is built and maintained with AI assistance by Madrasco; a human owner can be reached on request through the Issues tab.

Disclaimer

Independent tool, not affiliated with or endorsed by Google. Google, Google Merchant Center and Google Shopping are trademarks of Google LLC, named here only to describe the feeds this tool checks. Passing these checks does not guarantee that Google will approve your products or account.