Overview
The Claude Code Metrics Dashboard provides visibility into Claude Code development activity, token consumption, cost efficiency, developer adoption, productivity, seat usage and return on investment across engineering teams. It enables organizations to monitor how effectively Claude Code is being used, optimize AI spending, and measure the impact of AI-assisted software development. The dashboard is designed for engineering leaders, platform administrators, and development teams who need to monitor AI coding adoption, usage efficiency, and return on investment. Open it from the left navigation menu: AI Coding Tools → Claude Code. Data comes from the Claude Code integration; see Claude Code for setup. If no Claude Code connection exists, the dashboard shows Claude Code Connection Required with a link to the Integrations page. Each tab opens with a short description. Hover the ⓘ icon next to a tab’s title to see the formulas behind its metrics.Controls and Filters
What is it?
Interactive controls used to customize dashboard data and reporting views.Why is it important?
Allows users to analyze Claude Code metrics for a specific project and across different time periods, and to export what they see.Where does it exist?
Top-right section of the dashboard header. On small screens they are grouped under a Filters button.Features
Project Narrows the dashboard to a single project. Date Range Sets a custom start and end date. Future dates, and dates before the selected project’s first sync, cannot be picked. Quick-select buttons- 7 Days (default)
- 15 Days
- 90 Days
The Cost Savings and ROI tabs are not affected by the date filters. They use only the Project filter and their own fixed calculation windows.
Use Cases
- Compare short-term and long-term trends
- Focus on a single project
- Monitor spending over time
- Share a tab’s data as a report
Navigation Tabs
What is it?
A set of sixteen analytical views that provide deeper insight into Claude Code usage, productivity, efficiency, seats and ROI.Why is it important?
Allows users to drill into specific aspects of Claude Code performance and usage across the organization.Where does it exist?
Directly below the dashboard header.Available Tabs
Executive Overview — high-level KPIs, the developer productivity table and daily active users. Token Consumption — token usage by type, cache hit rate, context waste and token efficiency. Cost Optimization — daily spend, unit costs, savings levers and model downgrade potential. Tool Efficiency — tool acceptance and success rates, by tool and by decision source. Adoption & Usage — user growth, daily and weekly activity, feature adoption and sessions. Productivity — completions per session, requests per active day and tokens per completion. Model Utilization — request and cost share by model, model family trends and latency. Top Performers — developers with the highest productivity scores and efficiency. Low Performers — developers with lower scores, higher retry rates or lower acceptance, for structured coaching. Inactive Users — seat roster health and the list of dormant developers. Waste Detection — the high-waste developer cohort and its token usage. Efficient Devs — the efficient developer cohort and its cache utilization. Operational Efficiency — off-hours spend and hour-of-day patterns. User Productivity — every developer ranked by productivity score. Cost Savings — license cost and the saving available from reclaiming inactive seats. ROI — hours and cost saved against seat investment.Use Cases
- Analyze AI adoption patterns
- Monitor developer engagement
- Identify optimization opportunities
- Improve cost efficiency
- Measure productivity impact
Executive Overview
What is it?
A high-level summary of Claude Code usage across the organization, followed by the developer productivity table, an acceptance-versus-productivity chart and the daily active users trend.Why is it important?
Provides an immediate understanding of Claude Code adoption, usage costs and productivity without requiring detailed analysis.Where does it exist?
First tab of the Claude Code Metrics Dashboard.Metrics Included
Total Tokens Total tokens consumed (input, output, cache read and cache creation) in the selected period. Total Cost Total Claude Code spend in the selected period. Sessions Total number of Claude Code sessions. Tool Acceptance Percentage of tool-use decisions (for example, file edits) that developers accepted. Active Claude Code Developers Number of developers who used Claude Code in the period. Cost / User Total cost divided by the number of distinct users. Completions / $1 Completions generated for every dollar spent. User Productivity table Every developer with their Productivity Score, acceptance rate, completions per session, completions per 1k tokens, lines accepted per active day, total tokens and chats. Sortable and searchable (see User Productivity). Acceptance Rate vs Productivity Score A bubble chart plotting each developer’s acceptance rate against their productivity score. Larger bubbles mean more completions per session. Daily Active Users Unique developers using Claude Code each day.Use Cases
- Monitor Claude Code adoption across teams
- Track AI spending and usage trends
- Measure AI coding effectiveness
- Spot developers who need support
Token Consumption
What is it?
A breakdown of token usage by type and how efficiently context is used.Why is it important?
Tokens drive cost. Seeing where they go shows whether caching is working and how much context is wasted.Where does it exist?
Token Consumption tab.Metrics Included
Total Token Consumption — tokens per day across all users and sessions. Input / Output / Cache Split — each day’s tokens by type, as a percentage of that day’s total. Cache Hit Rate % — share of tokens served from the context cache. Context Waste % — cache-creation tokens as a share of all tokens, shown against a 15% target line. Token Efficiency % — output tokens as a share of all tokens; higher is better.Use Cases
- Track token growth
- Verify that caching reduces cost
- Reduce context waste
Cost Optimization
What is it?
Spend trends, unit economics and estimated savings opportunities.Why is it important?
Shows how spend relates to output and where money can be saved.Where does it exist?
Cost Optimization tab.Metrics Included
Daily Cost vs Completions — daily spend overlaid with daily completion count. Cost Efficiency — Completions per Dollar — completions per dollar over time, against a target of 3 completions per $1. Daily Cost ($) — total spend per day. Cost per User vs Cost per Session — unit cost trends. Off-policy / Misuse Trend — requests flagged as misuse, such as calls to a powerful model tier that return very small outputs. Savings Levers ($) — estimated savings from model downgrade, caching and coaching high-waste users. Model Downgrade Potential — current cost of calls that could move to a cheaper model, against the estimated saving.Use Cases
- Identify high-cost periods
- Track unit cost against targets
- Prioritize savings actions
Tool Efficiency
What is it?
How often developers accept Claude Code’s tool actions and how often those actions succeed.Why is it important?
Low acceptance or success rates point to friction or wasted work.Where does it exist?
Tool Efficiency tab.Metrics Included
Tool Acceptance Trend — acceptance rate over time. Tool Success Rate — successful tool executions as a share of all executions, with a 90% target and a 70% alert line. Acceptance by Tool — acceptance rate and rejected count per tool. Success Rate by Tool — call volume per tool, ranked by success rate. Decision Source Breakdown — total versus accepted decisions by decision source.Use Cases
- Find tools developers reject most
- Detect failing tool executions
Adoption & Usage
What is it?
User growth and engagement over time.Why is it important?
Shows whether Claude Code is being adopted and used consistently.Where does it exist?
Adoption & Usage tab.Metrics Included
Daily Active Users — unique developers each day. New vs Returning Users — daily split between first-time and repeat users. Feature Adoption % — share of active users who used at least one advanced feature. Session Growth — sessions created per day. Weekly Active Users (Rolling) — 7-day rolling active users.Use Cases
- Track rollout progress
- Monitor engagement depth
Productivity
What is it?
Output and depth of Claude Code use.Why is it important?
Measures how much useful output Claude Code produces per session and per token.Where does it exist?
Productivity tab.Metrics Included
Completions per Session — average completions per coding session. Requests per Active Day — requests per developer active day. Tokens per Completion — average tokens per completion, an efficiency proxy.Use Cases
- Measure productivity output
- Compare efficiency over time
Model Utilization
What is it?
How usage and cost are spread across Claude models.Why is it important?
Shows whether expensive models are used where they add value, and how fast each responds.Where does it exist?
Model Utilization tab.Metrics Included
Model Distribution (Requests) — share of requests per model. Model Family Trend — daily request share by model family. Cost Share by Model — each model’s share of total cost. Latency by Model (ms) — P50, P95 and average response time per model.Use Cases
- Right-size model choice
- Spot slow models
Top Performers
What is it?
The developers with the strongest productivity and efficiency.Why is it important?
Identifies practices worth sharing across teams.Where does it exist?
Top Performers tab.Metrics Included
Top 10 — Productivity Score Top 10 — Completions per Session Top 10 — Lowest Cost / Completion (ranked by completions per 1k tokens)Use Cases
- Recognize effective usage
- Identify internal champions
Low Performers
What is it?
Developers with lower scores, higher retry rates or lower acceptance.Why is it important?
Intended for structured coaching, not performance evaluation.Where does it exist?
Low Performers tab.Metrics Included
Bottom 10 — Productivity Score Top 10 — Highest Retry Rate Bottom 10 — Acceptance Rate Bottom 10 — Completions per SessionUse Cases
- Target enablement and training
- Track coaching outcomes
Inactive Users
What is it?
Seat roster health: how many seats are in use and who is dormant.Why is it important?
Unused seats are paid for but deliver nothing.Where does it exist?
Inactive Users tab. Requires an uploaded seat roster.Metrics Included
Total Users, Roster Size (licensed seats), Active Users and Inactive Users. Inactive Developers — table with Name, Email, Productivity Score, Chats, Last Active and Inactive Days. A seat counts as inactive when it is both under-used (fewer than 5 chats or a productivity score below 40) and has had no activity for the number of days set for Claude Code under Settings → Inactive User Settings (30 days by default). A seat that was never used counts as inactive.Use Cases
- Reclaim unused licenses
- Follow up with dormant developers
Waste Detection
What is it?
The cohort of developers classified as high-waste: high token use for few completions.Why is it important?
Highlights where tokens are spent without results.Where does it exist?
Waste Detection tab.Metrics Included
High-Waste User Count Trend — high-waste developers per period. Tokens per Completion — Waste Cohort Completions per Session — Waste Cohort Waste Cohort — Tokens vs CompletionsUse Cases
- Reduce cost leakage
- Target coaching
Efficient Devs
What is it?
The cohort of developers achieving high completions per token and strong cache use.Why is it important?
Shows what efficient usage looks like.Where does it exist?
Efficient Devs tab.Metrics Included
Efficient User Count Trend Completions per 1k Tokens — Efficient Cohort Cache Hit % — Efficient Cohort Efficient vs Waste User Trends — both cohorts compared over time.Use Cases
- Benchmark good practice
- Track whether the efficient cohort grows
Operational Efficiency
What is it?
When Claude Code is used and what it costs by time of day.Why is it important?
Reveals off-hours spend and peak-usage windows.Where does it exist?
Operational Efficiency tab.Metrics Included
Off-Hours vs Business Hours Cost ($) Hourly Cost ($) — cost by hour of day. Hourly Request Volume — requests by hour of day. Cache Hit % by Source SystemUse Cases
- Spot high-burn windows
- Understand usage patterns
User Productivity
What is it?
Every developer ranked by productivity score.Why is it important?
Gives a complete, sortable view of individual Claude Code effectiveness.Where does it exist?
User Productivity tab (the same table also appears on Executive Overview).Metrics Included
Productivity Score — 40% acceptance rate, 35% lines accepted per active day (percentile rank) and 25% completions accepted per token (percentile rank). The weights can be changed under Settings → AI Coding Tool Settings → Claude Settings → Score Weights. Plus Acceptance Rate, Completions / Session, Completions / 1k Tokens, Lines Accepted / Active Day, Total Tokens and Chats. Use Search user… to find a developer.Use Cases
- Compare developers on one scale
- Prepare 1:1 conversations
Cost Savings
What is it?
What Claude Code licenses cost and what reclaiming inactive seats would save.Why is it important?
Turns inactive seats into a concrete figure.Where does it exist?
Cost Savings tab. Requires an uploaded seat roster and uses only the Project filter.Metrics Included
Claude Code — License Cost- Total Seats
- Cost / Seat / Month — from Claude Code’s cost per seat under Settings → Tool Cost Settings
- Monthly Cost
- Annual Cost
- Inactive Seats
- Monthly Saving Potential — inactive seats × cost per seat
- Annual Saving Potential
Use Cases
- Justify license reductions
- Track the cost of unused seats
ROI
What is it?
An estimate of the value Claude Code delivers against its seat cost.Why is it important?
Supports executive reporting on AI investment.Where does it exist?
ROI tab. Requires an uploaded seat roster and uses only the Project filter.Metrics Included
Total Hours Saved — monthly, based on active users × Hours Saved / User / Day. Estimated Cost Saved — monthly hours saved × Developer Hourly Cost. Monthly ROI — monthly savings versus monthly investment (Per Seat Cost × seats). Annual ROI — annual savings versus annual investment. ROI Multiplier — cost saved ÷ investment, with the number of days to break even. The three inputs are set under Settings → ROI Settings → AI Coding Tool Settings → Claude Code Settings (defaults: $25 per seat per month, 0.5 hours saved per user per day, $40 developer hourly cost).Use Cases
- Report return on investment
- Test different savings assumptions