Case study 02
Building a feed audit and alerting system that turns hours of manual work into minutes
A scoring-based audit system for Google and Microsoft Shopping feeds — giving agencies and enterprise advertisers visibility into feed health that previously required spreadsheets, formulae, and hours of hunting.
50+↑ adoption
Agencies and in-house teams actively using
Hours→ mins
Manual audit time eliminated for enterprise inventories
100K+
Products auditable per account without spreadsheets
Solo PM
Google Shopping
Microsoft Ads
Merchant Center
Enterprise & agency focus
The problem
Feed health is the silent killer of Shopping campaign performance
For Shopping and Performance Max campaigns, the merchant feed is the foundation everything is built on. A poorly optimised feed — missing attributes, weak product titles, incorrect categorisation — directly suppresses campaign performance, often without any visible signal in the ad platform itself.
The standard workflow was manual: download the feed, open a spreadsheet, write formulae across multiple sheets, and hunt for the specific issues relevant to your business. For agencies managing multiple clients, or enterprise advertisers with inventories exceeding 100,000 products, this process took hours — and was often skipped entirely.
Merchant Center surfaces issues at the product level, but doesn't tell you which issues matter most for your specific business, or where to focus attention first.
What I built
A two-layer audit system built around presence and quality
Layer 1
Attribute presence audit
Checks whether products have a defined value for each feed attribute. Users configure which attributes are critical for their industry. The audit scores coverage and flags gaps accordingly.
Layer 2
Qualitative efficiency audit
Checks the quality of high-impact attributes — primarily product titles and descriptions. Validates optimal length, keyword presence, and structural best practices.
Across both layers, I defined a scoring system based on domain knowledge — weighting issues by their likely impact on campaign performance. Fix the highest-scored issues first, see the most improvement.
Key UX decision: surface products first, not issues. Merchant Center groups results by issue type — which makes it hard to assess a specific product's overall health. We inverted this.
Alerting layer
Three alerts covering disapprovals, stock, and performance drops
Product disapprovalsAlerts with the specific disapproval reason from Merchant Center, translated into plain language with recommended next steps.
Out-of-stock productsFlags products that have gone out of stock and are no longer eligible to trigger ads, so advertisers can act before wasting budget on unavailable inventory.
Top performer drop-offMonitors highest-performing products and alerts when one falls out of its usual performance range — catching silent degradation with no platform-level signal.
Impact
A workflow that didn't exist before — now used by 50+ teams
Enterprise advertisers with 100,000+ product inventories can now run a complete feed health audit in minutes, with a prioritised action list. The top performer alert goes further — it's the difference between finding out a flagship product degraded in a monthly review versus catching it the same week it happened.