A delivery score
you can defend.
predictDeliveryScore writes delivery_score_signals from four components: identity_strength, freshness, platform_health, and click_id_present. The score explains why an event is likely to match, deliver, and optimize.
Built from the product surface, not brochure claims.
What the team actually gets.
Identity strength
Scores whether email, phone, external_id, fbp, fbc, and platform click IDs are present and usable.
Freshness
Penalizes stale events that arrive outside useful optimization windows.
Platform health
Uses delivery attempts, errors, and latency so the score reflects current destination conditions.
Click ID presence
Separates browser identifiers from click identifiers so paid media teams can see exactly what is missing.
How it compares to ordinary tracking work.
- identity_strength
- component score
- freshness
- event latency
- platform_health
- delivery-backed
- click_id_present
- fbc/ttclid/gclid
- score evidence
- delivery_score_signals
- identity_strength
- EMQ symptom
- freshness
- not shown
- platform_health
- platform UI
- click_id_present
- tag debug
- score evidence
- black box
- identity_strength
- manual
- freshness
- not shown
- platform_health
- connector status
- click_id_present
- partial
- score evidence
- dashboard only
Real merchant-shaped cases and measurable signals.
The references an operator can inspect.
Where this matters in production.
Meta EMQ dropped but events were still delivered.
delivery_score_signals showed fbc_presence_rate, not delivery status, was the problem.
LinkedIn leads arrived 20 minutes late after queue congestion.
Freshness score isolated the lag and prevented a false identity investigation.
One regional store had poor Google Ads match quality.
Click ID component showed gclid was stripped only on the EU checkout.