Shopping, Performance Max & Feed Management

When Performance Max delivery slips, the product feed underneath it is usually where the answer starts.

Performance Max automates bidding, budget, audiences, and creative across Google's channels from one campaign, which can make the cause of an item's delivery loss difficult to see. We audit the product feed against Google's data specification, connect disapprovals to items losing impressions, and then review the campaign layer above it. Teams running Shopping or Performance Max off a real product feed worth auditing are the fit here. A service business with no Merchant Center feed has nothing for this method to work on.

A clear line from a feed error to its campaign-level symptom, so a fix gets applied at the source where the problem started.

A specialist tracing a disapproved product from a Merchant Center feed error to its Performance Max delivery drop

Some of the 500+ brands we've worked with

See all references
  • Defacto
  • Little Caesars
  • Hisar
  • Bundle
  • Yolcu360

How the work runs

Feed evidence leads the campaign diagnosis

Merchant Center and the live campaign provide the checkpoints for every fix. AI can flag mismatches and track approval timing. It cannot resubmit a feed, edit product data, or change a live Performance Max campaign.

How we hold ourselves to it

  • Audit the feed before adjusting the campaign
  • Treat every disapproval as a diagnosis to work through
  • Respect the platform's own re-crawl and approval timing
  • Investigate what automation leaves unexplained

  1. Audit the feed against the product data spec

    Check attributes, identifiers, images, and price and availability parity with the live site against Google's own product data specification.

    Feed audit against Google's product data specification covering attributes, identifiers, images and price and availability parity.

    AI assist
    Flags feed rows missing required attributes or mismatched against the site.
    Human gate
    Ecommerce or feed owner confirms which flagged issues are real versus expected.
    Owners
    Feed owner + ecommerce lead

  2. Reconcile Merchant Center with Ads-side performance

    Connect specific disapprovals and warnings to the items actually losing impressions in the campaign, with feed and delivery evidence in the same review.

    Reconciliation connecting specific disapprovals and warnings to the items actually losing impressions.

    AI assist
    Matches Merchant Center diagnostics to Ads-side item performance data.
    Human gate
    Paid-search lead validates which disapprovals are worth prioritizing first.
    Owners
    Paid-search lead + feed owner

  3. Fix and resubmit, tracking the approval window

    Correct the feed, resubmit it, and log the expected re-crawl and approval timing. Review begins after the platform has had time to process it.

    Remediation log recording each fix, its resubmission and the expected re-crawl and approval timing.

    AI assist
    Tracks elapsed time against the platform's typical re-crawl and approval window.
    Human gate
    Feed owner approves the corrected data before resubmission.
    Owners
    Feed owner + paid-search lead

  4. Audit Performance Max asset and listing groups

    Check asset groups and listing groups against the now-cleaned feed for gaps, overlaps, or groupings that no longer match the catalog.

    Asset and listing group map checked against the cleaned feed, with gaps, overlaps and stale groupings flagged.

    AI assist
    Compares asset and listing groups against the current feed for mismatches.
    Human gate
    Paid-search lead approves any structural change to asset groups.
    Owners
    Paid-search lead + feed owner

  5. Diagnose automation visibility gaps

    Use the diagnostics and insights the platform does expose to explain what an automated asset group is otherwise hiding about a delivery change.

    Diagnostic note using the exposed insights to explain a delivery change an automated asset group otherwise hides.

    AI assist
    Pulls available Performance Max insights and drafts a plain-language explanation of the pattern.
    Human gate
    Strategy lead validates the explanation before it counts as a finding.
    Owners
    Strategy lead + paid-search lead

  6. Set the recurring feed-health cadence

    Assign an owner and a schedule for checking feed diagnostics before an unresolved error turns into a campaign-level symptom again.

    Feed-diagnostics cadence with a named owner and a carried-forward queue.

    AI assist
    Prepares the recurring review queue from the latest Merchant Center diagnostics.
    Human gate
    Feed owner reviews and closes each cycle.
    Owners
    Feed owner + paid-search lead

What lands with your team

A remediation record tied to the underlying cause

The audit, reconciliation, and remediation record tie a delivery drop to a specific feed row and the evidence around it.

  • Feed health audit

    Every error and disapproval found, its severity, and its reconciled impact on Ads-side performance.

    Accepted when

    Every error and disapproval is listed with its severity and its reconciled effect on Ads-side delivery.

    Cadence: At setup, then recurring

  • Feed and campaign remediation log

    What was fixed, when it was resubmitted, and when it actually cleared. The record shows the fix's actual timeline.

    Accepted when

    Each fix records when it was resubmitted and when it actually cleared, so the real approval timeline is visible.

    Cadence: Every fix and resubmission

  • Performance Max asset and listing group map

    Current structure checked against the cleaned feed, with gaps and stale groupings flagged.

    Accepted when

    The structure is checked against the cleaned feed, and every gap or stale grouping is flagged rather than assumed intentional.

    Cadence: On material feed change

  • Recurring feed-diagnostics cadence

    The review schedule, its owner, and the open queue carried from the last cycle.

    Accepted when

    The schedule names its owner and carries the open queue forward, so an unresolved error cannot go a cycle unseen.

    Cadence: Fixed recurring schedule

Before the work starts

When this work is the right next step

Campaign symptoms can begin with a product-feed problem. Google's product data specification sets out what an item needs: correct identifiers, complete attributes, and a price that matches the live landing page. If any of those are wrong, the item can be disapproved while the campaign layer shows only weaker delivery and no obvious cause. Performance Max makes the problem harder to trace because it automates across channels from one campaign and does not expose which keyword or placement drove a result.

A good fit when

  • Shopping or Performance Max runs on a real product feed with enough SKUs to audit, but feed errors have not been reconciled with item-level delivery.
  • A feed fix has been resubmitted, yet the team is judging campaign delivery before Google's re-crawl and approval window has had time to close.
  • Performance Max delivery has dropped, but nobody can tell whether the cause is in the product feed or campaign layer.
  • Items sit disapproved in Merchant Center for attribute or price mismatches nobody has reconciled with Ads-side performance.
  • A feed fix went live yesterday, and someone's already calling it a failure because nothing has changed yet.
  • Performance Max delivery dropped, and nobody can say whether the feed, the automation, or something else caused it.
  • A price shown in the feed does not match the live landing page, and the mismatch surfaced as a disapproval.

Better handled as other work when

  • The business does not sell physical products through Shopping and has no Merchant Center feed, so this method has no catalog evidence to audit.
  • The expected outcome is confirmation before Google's re-crawl window can finish, so a feed remediation cannot be judged on the requested timeline.
  • Feed or site product data cannot be edited by the team, which means a confirmed attribute or price mismatch cannot be corrected at its source.

People who own a single channel

Paid Search, Paid Social, CRO, and Programmatic each run under a named owner at Zeo. The consultants below are matched to the channel this page is about, so you can see who you'd actually work with.

Tools we use

Tools behind this work

  • Optmyzr

    reads the Performance Max campaign layer once the feed itself is already clean

  • DataFeedWatch

    reshapes the raw feed against the channel's own attribute requirements before it gets submitted

Start with the product feed

Audit the feed behind a Shopping delivery drop

Bring us the feed and the campaign. We'll check it against Google's specification, reconcile disapprovals with delivery, and set a cadence that keeps errors from recurring.
Audit your Shopping feed

The questions we hear most

New or corrected feed data typically needs 24 to 48 hours to be recrawled, and a brand-new feed can take up to three days to approve. We log the fix against that window and wait for the platform to process it before reviewing the result.