Overview
Every LuMay agent ships with a measurement plane - containment rate, escalation rate, connector success rate, sentiment, and cost per outcome are instrumented from day one. The sections below define each metric precisely, including its denominator, so a number from a pilot can be compared correctly to a baseline rather than taken at face value.
Built-in dashboard signals
Containment
Resolved without transferConnector Success
System actions completedCost per Outcome
Unit economics over timeSentiment
Experience quality signalDesign principle
Show the dashboard as a feedback loop - LuMay as a learning operating system, not just a reporting tool.
Built-In KPIs
| KPI | Definition (numerator / denominator) | Why it matters |
|---|---|---|
| Containment Rate | Calls resolved without human transfer ÷ total calls handled | The core automation metric - measures whether the agent replaced human effort |
| Escalation Rate | Interactions transferred to a person ÷ total interactions | Highlights gaps in the conversation flow and knowledge base |
| Connector Success Rate | Tool calls completed successfully ÷ total tool calls attempted | Connector failures silently degrade agent quality if not tracked |
| Outbound Answer Rate | Outbound calls answered by a live person ÷ outbound calls placed | Key efficiency metric for outbound campaign agents |
| Sentiment Distribution | Positive / neutral / negative split across call sessions | A quality signal - deteriorating sentiment often precedes rising escalation |
A baseline matters: a 70% containment rate is only meaningful relative to what your current process resolves without escalation today. Establish that baseline before the pilot, not after.
What Isn't Measured Yet
The KPIs above are available from day one using built-in instrumentation. The following require a separate integration and are not built-in features today:
- CSAT - requires a post-call survey integration; sentiment scoring is the closest available signal today.
- NPS - requires a survey tool integration; tracked as a custom KPI once configured.
- Appointment conversion rate - available only once the agent is configured with a scheduling connector and the outcome is recorded.
Illustrative Example
The numbers below are a clearly hypothetical, illustrative dataset for explaining how to read the metrics together - they are not real customer results.
| Metric | Hypothetical week 1 | Hypothetical week 4 |
|---|---|---|
| Containment rate | 52% | 68% |
| Escalation rate | 48% | 32% |
| Connector success rate | 91% | 97% |
In this hypothetical, rising containment and falling escalation over four weeks would suggest the flow and knowledge base improved - but a cost-per-outcome reduction by itself is a cost-savings signal, not a claim of realised ROI, until it's weighed against the actual cost of running and maintaining the agent over the same period.
FAQ
Is cost savings the same as ROI?
No - cost savings is one input to an ROI calculation; ROI also requires the cost of implementing and operating the agent, which this page does not estimate for you.
Can we see per-call traces, not just aggregate metrics?
Yes - every call session produces a complete trace with timing at every hop, viewable in the tracing backend.
Does a high containment rate always mean good quality?
Not on its own - cross-check it against sentiment distribution and escalation rate to rule out cases where calls are "contained" but poorly resolved.
Can we add our own custom KPI?
Yes - custom KPIs can be configured when the relevant connector or data source is available (e.g. NPS once a survey tool is integrated).
Who sets the baseline for comparison?
You do - the platform doesn't assume a baseline; establishing your pre-pilot numbers is a step your team owns before the pilot starts.
Related
- Architecture
- Deployment
- Voice Engine - the source of per-call traces, sentiment, and containment data
- UK AI Platform hub - all 8 component pages