Signal health agent
Error logs, read for you.
Error logs are usually read only after something has already gone wrong. The signal health agent reads them continuously, groups related failures, identifies the probable cause, and proposes a fix with the evidence attached.
Triage · signal health agent
2 open clusters
purchase rejected - order_value arriving as string
prod · 4.2% of prod purchase events · first seen 3d ago
Probable cause. Checkout service serialises currency as a formatted string after the v8.2 deploy.
Proposed fix. Cast to number at the call site, or widen the property type and re-activate the definition.
heartbeat lost - stage helper silent
stage · all stage click context · first seen 6h ago
Probable cause. Helper 1.3.9 is pinned on stage; the origin allowlist changed with the new preview domain.
Proposed fix. Add the preview origin to the app's base URL list and upgrade the helper to 1.4.2.
Illustrative data
What it covers
Every point where signal can be lost.
Rejected events
Schema violations and unknown events, grouped by cause rather than listed one by one.
Lost heartbeats
Environments where the browser helper has stopped reporting, or reports from the wrong origin.
Throttling
Traffic approaching or exceeding your plan's limits, before it affects reporting.
Delivery failures
Conversions that Meta or Google rejected or could not match.
Expired connections
Ad account tokens that have expired or lost permissions.
Network API changes
Breakage caused by changes on the ad networks' side.
Triage
A cause and a fix, not a wall of log lines.
Each cluster shows the affected app and environment, when it started, how much data it touches, the probable cause, and a proposed fix. The raw events are one click away.
Fixes that change configuration are proposals. They go through the same approval and audit path as every other change in Quantiv.
- Clusters ranked by how much signal they cost
- Probable cause stated in plain language
- Proposed fix with links to the failing payloads
Triage · signal health agent
2 open clusters
purchase rejected - order_value arriving as string
prod · 4.2% of prod purchase events · first seen 3d ago
Probable cause. Checkout service serialises currency as a formatted string after the v8.2 deploy.
Proposed fix. Cast to number at the call site, or widen the property type and re-activate the definition.
heartbeat lost - stage helper silent
stage · all stage click context · first seen 6h ago
Probable cause. Helper 1.3.9 is pinned on stage; the origin allowlist changed with the new preview domain.
Proposed fix. Add the preview origin to the app's base URL list and upgrade the helper to 1.4.2.
Illustrative data
Why it works
You cannot triage a black box.
Validation results, helper heartbeats, and the delivery ledger are first-class records in Quantiv. The agent reads them directly, which is why it can say what went wrong and where.
Delivery ledger
last 4 deliveries
| Order | Destination | Matched on | Status |
|---|---|---|---|
| ord_88213 | Meta CAPI | email · fbc | delivered |
| ord_88213 | Google EC | email · gclid | delivered |
| ord_88207 | Meta CAPI | deduplicated | |
| ord_88199 | Google EC | - | retrying |
Every delivery is recorded, deduplicated against the pixel, and retried until it succeeds or is explicitly dropped.
Illustrative data
Questions
Common questions.
Does it change anything automatically?
Not without approval. It proposes fixes. Applying a configuration change follows the same review and audit path as a change made by a person.
How are we notified?
Open clusters appear in your workspace, and administrators are alerted when a cluster affects production data.
See it on your own numbers.
A two-week pilot runs your conversions through Quantiv alongside what you report today. The difference between the two is the business case.