Anomaly Triage
Investigate a metric anomaly from data the user supplies by confirming it is real before explaining it.
Sample result
How it works
What Anomaly Triage does
Checks measurement and instrumentation first, sizes the move against a robust same-weekday baseline, localizes it by segment (splitting a rate change into mix and rate effects), lines it up against a timeline of releases, campaigns, outages and calendar events, ranks competing hypotheses with supporting and refuting evidence, and orders the cheapest discriminating next checks with a severity and escalation call.
Use it when
Someone asks why a metric spiked or dropped, whether a change is real or noise, which segment drove it, or what to check next.
Use something else for
A recurring weekly review, a deep funnel step diagnosis, or a live outage.
Inputs and results
What goes in, what comes back
- You give
A short brief describing what you need
Add a specialist method to the work.
- You get
Anomaly triage report, hypothesis table, segment decomposition and escalation note
The method guides the work with your files and context.
- You check
The sources and open questions
Review the output against your brief.
Questions and answers
Questions about Anomaly Triage
Updated Oct 20, 2026.
What does Anomaly Triage do?
Investigate a metric anomaly from data the user supplies by confirming it is real before explaining it. Checks measurement and instrumentation first, sizes the move against a robust same-weekday baseline, localizes it by segment (splitting a rate change into mix and rate effects), lines it up against a timeline of releases, campaigns, outages and calendar events, ranks competing hypotheses with supporting and refuting evidence, and orders the cheapest discriminating next checks with a severity and escalation call. Use when someone asks why a metric spiked or dropped, whether a change is real or noise, which segment drove it, or what to check next. Do not use for a recurring weekly review (use weekly-analytics-insight-report), a deep funnel step diagnosis (use diagnose-conversion-funnel), or a live outage (use coordinate-production-incident).
Who is Anomaly Triage for?
It is designed for product analysts, growth and marketing teams, data owners and operators on call for metrics.
What do I need to use it?
Nothing extra is needed. Sign in to Relay to run it with your own files and context.
What does Anomaly Triage produce?
Anomaly triage report, hypothesis table, segment decomposition and escalation note. Results come back with their sources and any open questions for you to check before anything is used.
How much does Anomaly Triage cost?
It is free to use. Paid model usage, if any, is shown before a run.
Can an AI agent use it?
Yes. Agents can read its public definition through the Relay API and run it with a person’s authorization. See the For Agents page for how to connect.
Details
Trust, access and technical details
Trust and access
- Visibility
- Public: anyone can view
- Status
- Published and maintained
- Pricing
- Free
- cmvgt6uxm0027…
- Oct 20, 2026
- 0 signals
- Commit pinned
Security review is a static package assessment, not a guarantee of safety or a behavioral qualification.
Use it from code or an agent
Read the public definition, then check cost and inputs with a preflight before a run.
Open definitionPOST /api/v1/runs/preflight
How agents connect to Relay →Tags and source
- Repo
- rose-digital/relay
- License
- MIT
Connections
Where Anomaly Triage fits
Have a task this would help with?
Describe it. Relay plans the work, picks the right playbook and shows the cost before anything runs.