Product feed audit for online stores
Run a product feed audit to find errors in product IDs, prices, stock, and links. Compare listings with product pages and prioritize fixes.
Free app. Runs on your Mac with the Claude or Codex plan you already have.
What a product feed audit is
A product feed audit checks the file that sends your store's product data to a shopping platform. Check every row against the platform's rules and compare selected rows with their landing pages. The result is a list of product feed errors, the items affected, and the exact changes you can make. Start with problems that can keep an item from being eligible, such as duplicate IDs or a price that differs from the landing page.
A feed row usually represents something a shopper can buy. Product variants are versions of the same item, such as a pair of shoes in different sizes. Each buyable version needs consistent identity, price, and availability. Feed validation tests the submitted data. Shopping results and account reports provide evidence about where listings appear.
What it does for the business
- Restore eligibility for affected itemsResolve errors that can lead to refusals before spending time on wording or image polish. If a single export rule causes the same fault across a category, one correction at the source may resolve many rows.
- Send buyers to the item they clickedMatching links, prices, and stock help shoppers find the offer they saw. Use the audit to catch a link that opens the default size or a listing that still shows yesterday's sale.
- Send fixes to the right personAn exact field and source row let a merchandiser repair an item or a developer correct the export mapping. The next export shows whether the fix took effect.
- Decide where to investigate visibilityAfter resolving data faults, use account reports and dated observations to identify eligible items whose visibility needs a closer look.
Product feed audit, step by step
Identify the feed, market, and sellable items
A feed check starts with knowing which item and market each row represents.
What to do
- In the shop admin, open the shopping channel or the app that creates the export. Download the feed or copy its link. Save the original together with the export date.
- Record the destination platform, target country, language, currency, and field specification for that data source. Use the export and site to resolve unclear settings, and label assumptions.
- Map the supplier fields to the destination fields. For example, confirm which source column becomes the item ID and which becomes the group ID that ties related sizes or colors together.
- Keep a row reference from the original file. Read identifiers as text to preserve leading zeros, and confirm what a real variant looks like before writing any rule for duplicate IDs.
Why
Field requirements depend on the destination and the item. A valid currency in one market can be wrong in another. A shared group ID can be correct across many sizes, while a shared item ID can cause one size to overwrite another. Confirm the row's identity before proposing a change.
Check the fields on every row
Validate every row before judging the size of a feed problem.
What to do
- Check item IDs for duplicates and verify that related variants have the group and distinguishing attributes their destination requires. Review duplicate rows in context to see which sellable items they represent.
- Flag empty required fields and invalid values, such as an availability label outside the accepted list. Test price format, currency, sale dates, and whether sale price exceeds regular price.
- Check product and image links for formatting, the expected domain, placeholder images, and links that take different variants to the same selection. Review titles for excessive length, promotional claims, and missing distinguishing details.
- Validate identifier length and check digits where the format defines them. A GTIN is a Global Trade Item Number. Compare repeated codes with the product records and flag missing codes where the product calls for one.
Why
A script applies the same rules to the entire export and shows which rows share a problem. Shared faults point to the field mappings worth checking first. The original value and row reference help the feed owner trace the fault.
Open the landing pages and check shopping evidence
Compare each offer with the exact page and variant a shopper reaches.
What to do
- Start with flagged rows. Open the submitted link and confirm which variant is selected. Then review a spread of categories and prices to look for discrepancies that field checks cannot detect.
- Compare the visible price, currency, and stock with the feed and with structured data, the machine-readable product fields on the page. Record the variant, displayed values, and date of each check.
- Read merchant account reports or look up selected products where you can observe their shopping results. Record what appeared, along with the item and the context of the observation. Mark missing observation evidence as unknown.
Why
A row can contain a valid price, a valid currency, and a working link while still taking the shopper to a different offer. A visible price mismatch can lead to refusal. Account reports can supply item status when a shopping observation has limited scope.
Propose the correction and identify its source
A finding should tell the owner what to change and where to change it.
What to do
- Write the source row, item ID, field, original value, fault, and proposed value. Rank fixes by risk, with potential refusals first, followed by issues that may reduce visibility and cosmetic problems.
- For a duplicate ID, identify the store's stable reference for each variant. Check how an ID change would affect an existing listing and whether the group ID correctly ties the variants together.
- For a price or stock mismatch, check where the store keeps the working value. Use the correct buyer-facing value as the starting point, and verify the store record if the page itself looks wrong. Name the place to repair the source or mapping.
- Leave unknown values open and say what evidence would resolve them. Ask for the manufacturer code from the packaging or supplier records, or the correct price from the person who manages that market.
Why
Comparing current and proposed values shows which problems come from a stale record or a rule shared by many items. Repair that rule to prevent the next export from repeating the fault.
Verify and hand off the fixes
Every finding should trace back to a row and any page or report that supports it.
What to do
- Recheck findings against the original rows, landing pages, and reports. Reconcile issue counts with affected row counts, since one item can have several faults.
- Run your checks again on the proposed values. Keep unresolved values visible and confirm that identifiers, prices, and stock changes have supporting evidence.
- Hand off ranked edits, affected items, and examples of repeated faults. Include the row count, pages checked, and assumptions.
Why
The handoff shows where to start and where you need more evidence, such as a category with few page checks or a market with no account report.
Common mistakes
- Using the same item ID across sizesGive each buyable variant a stable item ID and use the group ID to show which variants belong together.
- Checking the default variantFollow the exact submitted link and compare the size or color the shopper actually lands on.
- Correcting the export by handFix the product record or export rule so the next scheduled feed carries the correction.
- Filling missing codes from a similar productVerify the identifier against the manufacturer or supplier record for that exact item.
- Treating a missing observation as zero visibilityRecord the limits of your observation and use an account report when you need item-level status.
What a good result looks like
- The exact export, destination specification, and market are recorded.
- Every source row was checked and the reviewed pages are listed.
- Every finding identifies the field, affected items, and reason for its priority.
- The proposed values are validated and unknown merchant facts stay open.
- Shopping observations have a source and a recorded scope.
And then what? Using the result
- Assign refusal risks to whoever owns the product record or feed mapping, and give them the affected row references.
- Resolve merchant decisions about genuine item identifiers, prices, and inventory before enabling the corresponding changes.
- Generate a fresh feed after making repairs and recheck the affected variants and their landing pages.
- Submit the corrected feed through your normal workflow and check the account's processing and item status after it updates.
When to run it
- Before opening a new shopping channelCheck how the new destination expects prices, variant attributes, and identifiers. You can repair the export mapping before the first submission.
- After a change to the store or feed appA migration can change IDs, links, or field names. Compare fresh output with the old export and check where the new links land.
- After a sale starts or endsCheck the exported prices and sale windows against the storefront, especially when different markets use their own prices or schedules.
- When the account flags affected itemsExport the item issues and match them to source rows. Start with the largest groups of repeated refusal risks.
Questions people ask
How do you do a product feed audit?
Identify the destination and market, validate every row, compare selected items with their landing pages, and write a ranked list of exact corrections. Check account reports for visibility evidence.
What should I export for a Google Merchant Center feed audit?
Get a copy of the submitted feed and the item issues from the account's Needs attention view. Record the data source and market and keep the product links so you can match each issue to the right variant.
What causes a feed price mismatch?
A feed price mismatch occurs when the submitted price or currency differs from the offer on the linked product page. Check the selected variant, the market, and the sale schedule before deciding which source needs repair.
What should I do if a product has no GTIN?
Confirm with the manufacturer or supplier whether the exact item has an assigned GTIN. If there is none, follow the destination rules for products without that identifier and use other supported identifiers where applicable.
Why is an approved product missing from shopping results?
Even for eligible items, shopping results depend on the query, market, and timing. Record those details and compare the observation with the item's account status.
How do you test a product feed correction?
Run your validation rules on the proposed rows and verify the corrected fields against their source records. After the change reaches the live export, open links for affected variants and confirm that the new values carry through.
More jobs in E-commerce
Check product feeds and shopping results, on your own Mac.Free app. Runs on your Mac with the Claude or Codex plan you already have. Your files stay on your computer.
Download for Mac