> ## Documentation Index
> Fetch the complete documentation index at: https://doc.entelligence.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Metrics glossary

> Definitions and interpretation guidance for Code Review, Agent Insights, Code Reliability, Model Router, Team Insights, and Deployment Health metrics.

The current UI label and tooltip are authoritative. This glossary explains how to use the metric, not a guaranteed implementation formula.

## Code Review

| Metric                     | Interpretation                                                            |
| -------------------------- | ------------------------------------------------------------------------- |
| Total PRs                  | Pull requests in the selected scope and period                            |
| Time to first review       | Time from eligible PR state to first counted review                       |
| Time to merge / cycle time | Time from the selected start event to merge                               |
| Merge rate                 | Share of eligible PRs merged in the period                                |
| Comment acceptance         | Addressed or accepted review feedback according to current classification |
| Comments per PR            | Counted review comments divided by eligible PRs                           |
| PR velocity                | Pull-request throughput over time                                         |
| Churn                      | Code changed again within the product's displayed churn window            |
| Reverted PR                | Pull request classified as reverted                                       |
| Confidence score           | 1–5 merge-confidence signal derived from review findings and state        |

## Agent Insights

| Metric            | Interpretation                                                                |
| ----------------- | ----------------------------------------------------------------------------- |
| Adoption          | Connected members with qualifying activity                                    |
| Efficiency        | Computed session-behavior signal; not developer performance                   |
| Retry rate        | Share or frequency of repeated attempts                                       |
| Context health    | Signals such as full-file reads, compaction, interrupts, and retries          |
| Spend             | Reported or estimated model cost for covered activity                         |
| Tokens per dollar | Token volume relative to displayed spend                                      |
| Run rate          | Projected spend based on current period activity                              |
| Untracked spend   | Spend not attributed to a mapped project                                      |
| Outcome           | Classified feature, fix/rework, test/review, or exploration/overhead activity |
| Adoption stage    | Getting started, Developing, Proficient, or Pioneering coaching view          |

## Code Reliability

| Metric              | Interpretation                                              |
| ------------------- | ----------------------------------------------------------- |
| Detection           | Production event ingested from a mapped source              |
| Pattern / EKU       | Recurring behavior derived from related detections          |
| Cluster             | Related patterns grouped for investigation                  |
| Confidence          | Strength of the derived pattern relationship                |
| Severity            | Product-assigned impact or priority signal                  |
| Active / dormant    | Pattern lifecycle based on current activity                 |
| Regression          | Matching behavior observed after a monitored fix            |
| Audit score / grade | Summary of the current observability-audit rubric           |
| P0–P3               | Recommended work priority, not incident or exploit severity |

## Model Router

| Metric             | Interpretation                                        |
| ------------------ | ----------------------------------------------------- |
| Baseline spend     | Cost of the comparison model/path used by the product |
| Routed spend       | Cost attributed to routed sessions                    |
| Savings            | Baseline cost minus routed cost                       |
| Projected spend    | Forecast, not guaranteed invoice                      |
| Cached tokens      | Tokens reported as served from provider cache         |
| Model distribution | Share of traffic sent to displayed models             |

## Delivery

Deployment frequency, lead time, change failure rate, and time to restore depend on source events, environment classification, repository mapping, and period. Confirm Entelligence's displayed definitions before comparing with another DORA implementation.

## Next step

Use [Code Review analytics](/code-review/analytics), [Agent Insights metrics](/agent-insights/metrics), or [Deployment Health](/team-insights/deployment-health).