How feed health protects sales

Read diagnostics first#

Open catalogue diagnostics before touching campaigns: item rejection counts, fetch status, availability warnings and event match quality tell the truth faster than ROAS does. Sort by revenue impact, not by error count, so hero faults lead the queue. Meta own catalogue ads guide explains the feed requirements behind the warnings.

Record a baseline while healthy: item counts, rejection rate, match rate and fetch time. Baselines turn later dips into measurable incidents. Catalogue setup sets the healthy starting point.

Share diagnostics with trading and developers, not just media. Most feed faults start outside marketing. Meta Ads hub keeps the shared language.

Availability faults first#

Fix availability mismatches before anything cosmetic: in stock items marked out, and out of stock items marked available. The first starves winners, the second sells absence and teaches distrust. Refresh stock daily minimum, hourly for fast movers, weekends included. Product sets can quarantine uncertain stock while fixes land.

Map every stock state in the shop to a catalogue value explicitly; unmapped states default badly. Preorders, backorders and made to order each need a deliberate policy. Meta placement and feed specs show how availability renders.

Alert on availability swings, because sellouts masquerade as performance dips. A sudden ROAS fall with falling available items is stock truth, not auction failure.

Price, currency and tax faults#

Reconcile feed prices with landing page prices including currency, tax and delivery logic. Mismatches trigger rejections and shopper complaints in equal measure. Keep one price source per market and publish from it everywhere. Multi-country catalogues isolate currency faults before they spread.

Handle sale pricing with start and end discipline: stale sale flags teach shoppers to distrust every later badge. Dynamic overlays stamp prices automatically, which makes source discipline matter more, not less.

Confirm decimals, rounding and price qualifiers per market. Small formatting faults reject large numbers of items silently.

Images and links that break#

Repair broken images and links in revenue order: heroes first, margin second, long tail last. Use absolute URLs on fast hosting, keep minimum resolution honest, and never hotlink temporary campaign assets. Check links deep to product pages, never to retired URLs or generic categories. Collections UX depends on landing continuity.

Keep image backgrounds consistent and colour accurate so matching starts with recognition. Lifestyle covers can sell while pack shots carry the grid. Feed creative serves both.

Revalidate after every site migration, CDN change or image optimisation pass. Performance work breaks absolute paths routinely.

Match-rate drift#

Watch event match rate weekly as the vital sign of ID health. Sudden drops mean scheme changes, usually from a platform update nobody announced. Compare Pixel content IDs against catalogue IDs on new products first, since onboarding gaps grow with range growth. Shopify wiring shows the alignment, and Pixel setup covers the event side.

Benchmark match rate against catalogue size monthly: growth that outpaces matching means new SKUs arrive unmatched. Fix onboarding before scaling spend behind them. Standard events define the expected payload.

Log every ID change with date and owner. Mystery dips become known causes once history is written down.

The weekly routine#

Run a thirty minute weekly check: fetch status, rejection delta, availability swing, price mismatch sample, hero image and link sample, match-rate trend. Fix heroes same day, queue the tail, and re-preview product sets after changes. Seasonal peaks deserve twice weekly passes.

Escalate repeated fetch failures to hosting and platform owners with timestamps; intermittent faults need logs, not opinions. Keep a backup feed export where the business allows it.

Book a pre-peak audit before the season that pays salaries. Our team reviews diagnostics with trading calendars in hand.

Frequently asked questions#