Skip to content
Get MCP AdsGet MCP Ads, home
Start free
View as Markdown
Deep contractGoogle Ads

How to audit product feed health across ad platforms

One pass over the product feeds behind Google Shopping, Meta catalogue, TikTok Shop and Pinterest catalogues: disapprovals, gaps, and the revenue trapped behind them.

IntentFind the products the platforms refuse to serve, and rank feed fixes by the revenue they unlock.

6
Steps
6
Tools
4
Rules
3
Failure modes

When to use it

Use it when

  • Before any peak season: feed rejects discovered in November are margin lost.
  • Shopping spend flat while inventory grew.
  • After a catalogue migration, re-platforming or price update run.

Do not use it when

  • Accounts without catalogue-based campaigns anywhere: nothing here applies.
  • Mid-crawl: right after a large feed push, platforms need hours to re-evaluate; auditing too early reads churn as breakage.

What it needs first

Preconditions

  • At least one catalogue-backed campaign type live.
  • Feed refresh cadence known, so freshness findings mean something.

Inputs

scope
Feeds or catalogues to inspect.Default: every connected platform with a catalogue.

Procedure

In order. Every tool named is one the gateway ships, and links to its reference.

  1. Read product-level status and performance on Google Shopping.

  2. Read the Meta catalogue: products, availability, price mismatches.

  3. Read Pinterest catalogue diagnostics and inventory coverage.

  4. Read TikTok shop catalogue diagnostics where connected.

  5. Cross the four: products rejected everywhere, rejected somewhere, and serving everywhere.

  6. Rank fixes by the trailing revenue of the affected products.

Decision rules

The observable condition, what it lets you conclude, and what takes the conclusion back.

  • When

    A product family is disapproved on one platform but serves on the others

    Conclude

    Platform-specific attribute problem; quote that platform's diagnostic verbatim.

  • When

    Top-sellers by trailing revenue appear in any rejected list

    Conclude

    Lead with them: one unblocked top-seller usually outweighs a hundred long-tail fixes.

  • When

    Rejections concentrate in one feed attribute (GTIN, price, availability)

    Conclude

    One upstream fix, not per-product edits; name the attribute and the count.

  • When

    Rejected products carry no trailing revenue anywhere

    Conclude

    Deprioritise explicitly; a clean feed is not the goal, served revenue is.

Evidence every conclusion must carry

  • Every finding names the product count, the attribute, the platform diagnostic and the trailing revenue affected.
  • Freshness claims cite the feed timestamp, not an assumption.

Where the agent stops

  • Feed fixes happen in the merchant tools upstream; the skill delivers the ranked work order.
  • No campaign change follows automatically from a feed finding.

How it goes wrong quietly

The cases where the analysis is wrong and still looks right. Read them before trusting a number.

  • Out-of-stock read as broken: availability churn is normal retail; compare against the account's baseline reject rate.
  • Price mismatch findings during an active promo window are the promo, not the feed.
  • Platforms cache feed verdicts on their own clocks; the same product can legitimately differ between platforms for a day.

What the answer contains

01Blocked revenue, ranked
Products or families, trailing revenue, platform, diagnostic.
02One-fix clusters
Attribute-level fixes that unblock many products at once.
03Serving coverage
Share of catalogue serving per platform.