---
title: "AI activity dashboard: All (Beta)"
canonical: "https://support.appfire.com/space/FD/3073311013/AI%20activity%20dashboard%3A%20All%20(Beta)"
format: markdown
---
> Macro (aura-html)
The All report consolidates data from the available AI activity dashboard reports, which currently include Copilot, Cursor, Claude, and Windsurf. This AI tool report is designed for organizational leaders seeking to measure return on investment and optimize resource allocation across their AI technology stack.
Visit the [AI metrics](https://appfire.atlassian.net/wiki/spaces/FD/pages/3073966122) page to learn more.
## Permissions
To view the All report, ensure the users are assigned a role that includes this report. We recommend confirming on the *Role management* page in Flow **Settings** that this report has been successfully added to any roles you want to grant access to.
Visit the [Role management](https://appfire.atlassian.net/wiki/spaces/FD/pages/1802207285) page to learn more.
## Filters
You can filter the All report by:
- Teams
- Nested teams
- Date range
- AI tools data extends back 12 months for better long-term adoption visibility
- >90 days is all monthly
- <90 days remains bi-weekly
- Champions data is limited to the most recent 30 days and is not filtered by the selected team or date range.
## Adoption and impact
This dashboard provides engineering leaders with a bird's-eye view of your GenAI investment. Instead of just showing if developers are using AI, it bridges the gap between tool adoption and engineering business value.
A higher score reflects a more successful rollout. For easy interpretation, the score is color-coded by position on a 0–100 scale, with **Excellent **(blue), **Good** (light blue), **Fair** (yellow), and **Poor** (orange) categories.
### Adoption score
This section answers the question: *Are our developers actually incorporating AI into their daily habits?*
- **Seat utilization:** The ratio of active users to total purchased licenses. It directly monitors the ROI of your software spend. If this is low, you are wasting money on shelfware. It signals to leaders that they need to run training sessions, check whether developers don't know they have access, or reallocate unused seats to eager teams.
- **Code acceptance rate:** The percentage of AI-generated code suggestions that developers actually keep. It measures the relevance and quality of the AI tool within your specific codebase. A very low percentage means the AI is generating noise or bad code, forcing developers to constantly delete or reject suggestions, which actually hurts developer flow.
- **Feature adoption:** Tracks the depth of usage beyond the inline code autocomplete (for example, using chat features, unit test generation, or documentation tools). It shows whether engineers are treating AI merely as a glorified spellchecker or using it as a true paired-programming assistant.
The chart also surfaces absolute inventory figures: **Total current seats**, **Active users**, and **Seats never used**.
### Impact score
This section answers the critical executive-level question: *Is this AI tool making our delivery faster, better, and more predictable?* It maps AI usage directly against traditional DORA and engineering metrics.
- **Speed - Cycle time, Time to merge:** Measures the total elapsed time from a developer's first commit to the PR's merge. This is the ultimate operational efficiency metric. It proves or disproves the core marketing pitch of AI coding tools—that they speed up shipping cycles.
Click **Details** to view stats for Cycle time and Time to merge metrics. You can also access the [Team Health Insights](https://appfire.atlassian.net/wiki/spaces/FD/pages/1802141869) report using the built-in **View on THI** shortcut button next to the Cycle time and Time to merge.
- **Quality - Efficiency, Unreviewed PRs:** Tracks code health and compliance flags (for example, bypassing standard code review processes). Speed is dangerous if it comes at the expense of technical debt or buggy code. It protects the organization against code bloating. If AI usage spikes speed but quality drops (for example, due to an influx of unreviewed PRs or broken builds), leadership can step in and adjust code review policies.
Click **Details** to view stats for Efficiency and Unreviewed PRs metrics, and launch the THI report.
- **Throughput - Tickets Completed**,** PRs Merged:** The volume of output moving through the pipeline. It tracks the engineering organization's total capacity and velocity. It helps leaders see if AI is helping teams handle a larger volume of work without increasing burnout or headcount.
Click **Details** to view stats for Tickets completed and PRs completed metrics, and launch the THI report.
## Adoption
The **Adoption** tab consists of the following dashboards:
- Tool adoption ROI (cards)
- Total active users
- Active user over time
- Team adoption breakdown
- Weekly usage performance trends ROI (cards)
- Usage metrics overview
- Quality & chat engagement
### Tool adoption ROI (cards)
The **Tool adoption** dashboard focuses entirely on the utilization, distribution, and lifecycle of all your GenAI tools. While the Adoption and Impact tabs focus on AI results, this one focuses on your developers' behavior.
These cards break down usage for each vendor, tracking **Active Users** and **Inactive Users** and their respective trends. The cards highlight license waste immediately. If you have a high number of inactive users on an expensive tool like Cursor, you can instantly reassign or cancel those seats to optimize your SaaS budget.
They also show that AI is not a single monolith. Different developers use different tools for different tasks (for example, Cursor for a dedicated IDE, Claude for complex architectural chat, Copilot for inline autocomplete).
#### Total active users
The Total active users chart displays the absolute number of assigned users versus active users per tool alongside a percentage calculation. It identifies the saturation rate and real developer preference across your entire ecosystem.
This chart helps you make data-driven decisions when contracts are up for renewal. If Claude has an 84.4% active rate while Copilot sits at 81.5% with far fewer total users, it tells leadership where organic demand is strongest. Moreover, it reveals where adoption has stalled. If a tool's active percentage is low, it may signal that the team needs better onboarding or that the tool isn't a good fit for your specific codebase.
#### Active users over time
This graph maps engagement fluctuations over a multi-month timeline, showing how user numbers change week-over-week or month-over-month. It separates temporary hype from sustained habit.
If you ran a company-wide workshop on Claude in December '23, this chart lets you see whether users stayed active in February '24 or if the initial curiosity wore off. Also, if you see a steady downward trend for a specific tool over three months, it serves as an early warning that the team is encountering friction or finding the tool less valuable.
To filter the chart by a specific AI tool, select or deselect its checkbox at the bottom of the page.
### Team adoption breakdown
The **Team adoption** table provides a granular, tabular view of AI usage across the organization. While the previous charts showed high-level trends and overall percentages, this screen is the operational ledger that maps specific AI behaviors to individual contributors and teams.
This table serves as an operational management tool to track the spread of generative AI tools across your engineering organization. Note that you can sort each column in the A-Z and Z-A order.
### Users view
This micro perspective drills down into the habits and engagement levels of individual contributors across the organization.
- **User **&** Active member:** Identifies the exact developer and whether they are currently utilizing their assigned AI seats. It bridges the gap between anonymous data and actual team behavior and tells you exactly who is stuck. For example, seeing that a specific user is marked as Inactive with no activity data allows a team lead to step in directly to offer help, remove technical blockers, or reallocate that license if it is not needed.
- **Adoption**: Colored indicator dots mapped to an internal rating scale: **Excellent **(blue), **Good** (light blue), **Fair** (yellow), and **Poor** (orange). Displays the exact percentage of an individual’s eligible work time or sessions utilizing the AI tool. It helps you separate your true power users (for example, 81% adoption) from casual dabblers (53%) and non-adopters (0%). This makes it easy to find internal mentors to pair with struggling developers.
- **Last activity date:** The exact timestamp of the developer's last AI session. It differentiates between a developer who tried the tool once three weeks ago and someone who is using it daily. This column helps track habit formation. If a user is active every day and suddenly stops, it can signal a shift in their project type (for example, moving from heavy coding to architecture/planning) or a drop-off in tool satisfaction.
- **Most used model:** Identifies the specific vendor and the exact LLM version that the developer relies on most. It reveals organic user preference at the grass-roots level. Engineering teams are not monoliths. If your senior developers are heavily favoring Claude for complex logic while junior devs prefer Cursor for autocomplete, this data guides your future procurement strategy and helps you buy the right mix of licenses or point to the user who could benefit from advanced prompt engineering training.
### Teams view
This view shifts the focus from individual contributor tracking to cross-team comparison and hierarchical auditing. It allows engineering leaders to focus on organizational structures rather than on names.
- **Team:** Identifies the specific functional group, squad, or business unit and establishes accountability. It moves AI metrics out of a generic company-wide pool and ties them directly to existing organizational hierarchies, allowing managers to see which engineering or business leads are successfully driving tool adoption.
- **Most used tool **& **Most used model**: Surfaces the preferred AI platforms and underlying LLM engines chosen collectively by that specific team to reveal workflow specialization. Different teams need different tools (for example, a backend team might heavily favor a specific code-generation model, while a documentation team favors another). This helps procurement understand if they need to negotiate multi-tool enterprise bundles or consolidate around a single standard.
- **Active AI members**: Instead of showing single active/inactive statuses, this column displays the active-to-total ratio for the total seats assigned to that specific team (for example, 18/23 or 30/35). It measures team-wide alignment and participation density. It instantly flags under-provisioned teams that need more seats, or over-provisioned teams where expensive licenses sit idle and waste budget.
- **Adoption**: Colored indicator dots mapped to an internal rating scale: **Excellent **(blue), **Good** (light blue), **Fair** (yellow), and **Poor** (orange). It provides instant visual triage for massive organizations (this specific list spans 2,203 entries). With this column, a leader doesn’t have to manually analyze the math across dozens of rows. They can simply scroll down the list, ignore the blue healthy rows, and stop precisely on the orange dots to identify which teams require immediate management attention, training resources, or license de-provisioning.
- **Last activity**: Shows the most recent timestamp showing when anyone on that team last triggered an AI action to track the momentum. It distinguishes between teams that tried the tool once months ago and abandoned it, versus teams that are actively using it as part of their current sprint cycles.
### Weekly usage performance trends ROI (cards)
This dashboard features several interactive charts that visualize AI tool usage trends and activity patterns.
The top metrics provide an overview of your team's AI tools usage: the aggregated performance score measures balanced engagement; User engagement shows the percentage of members actively using AI tools; and Code acceptance rate tracks how often AI-suggested code is adopted.
Click **View graphs** to open dashboards.
#### Usage Metrics Overview
This chart evaluates the active adoption of all AI tools across your engineering teams and measures whether they deliver the tangible value and reliability you expect.
The Usage metrics overview dashboard visualizes three data points related to AI-generated code:
- Suggestions
- Acceptances
- Chat active users
Select/Unselect the metric you want to see on your chart. You can also select the **Exclude weekends** to avoid the sharp dips in your data.
To inspect details for a specific date, mouse over the chart line for the respective metric.
#### Quality & Chat Engagement
This dashboard assesses whether AI has become a trusted, daily part of your engineers' workflow for both writing and reasoning about code.
The dashboard visualizes four metrics:
- Acceptance rate
- Chat prompts
- [Thoroughly Reviewed PRs](https://appfire.atlassian.net/wiki/spaces/FD/pages/1802338547)
- [Impact](https://appfire.atlassian.net/wiki/spaces/FD/pages/1801945528)
Select/Unselect the metric you want to see on your chart. You can also select the **Exclude weekends** to avoid the sharp dips in your data.
To inspect details for a specific date, mouse over the chart line for the respective metric.
## Impact
The **Impact** tab consists of the following dashboards:
- Impact metric cards
- Impact card details
- Team metric breakdown
### Impact metrics
The **Impact metrics** dashboard acts as the core data engine for proving the value of your AI investment. While the Adoption tab focuses on who is using the tools, this dashboard directly isolates the cause and effect of AI by providing side-by-side A/B testing data across six core engineering metrics.
Every single metric card contains two bars: blue (**All AI**) and grey (**Without AI**), evaluating cohorts within the exact same team scope and timeframe.
The cards isolate variables and remove external noise (such as holiday code freezes or sprint-planning overhead) because both groups are working under the same organizational conditions. They provide definitive proof of value or friction.
For example, in Time to Merge, the All AI cohort is at 18.7 hours, while the No AI cohort is at 21.7 hours. This gives a VP hard, data-backed evidence to justify expanding the AI tool license budget to the CFO.
The data is grouped into distinct dimensions of engineering health:
- **PRs Completed: **Tracks total volume and splits the raw workload baseline between the two cohorts. It shows the distribution of labor. If, for example, 81 PRs were done with AI vs. 64 without AI, it proves that the AI tool has deeply penetrated the code production pipeline.
- **Time to Merge:** The clock time from PR creation to merge. It highlights peer-review bottlenecks. A sharp overall downward trend indicates that AI is successfully helping teams streamline and accelerate the code review process.
- **Cycle Time:** The total time from the first commit to production deployment. It shows macro velocity. A massive drop for the AI cohort proves that the technology is successfully shortening the overall time-to-market.
- **Iterated PRs:** The percentage of PRs that require additional commits or rework after being opened for review. It measures code maturity. The lower AI cohort compared to the non-AI cohort shows that AI helps developers submit more accurate, higher-quality code on the first pass, leading to fewer back-and-forth review loops.
- **Tickets completed**: The total volume of issues moving through the pipeline. It shows total capacity. An overall upward trend helps leaders assess whether AI is helping teams handle a greater volume of work without increasing burnout or headcount.
- **Backflow Rate:** The percentage of tickets that move backward in the workflow (for example, from QA back to In Progress due to bugs or failed requirements). It guards against fake speed. If AI makes teams faster but backflow increases, they are just shipping bugs faster. If, for example, the AI group's backflow rate is lower, it proves the speed gains are structurally sound and sustainable.
When you click any card, a dedicated macro view expands at the bottom.
The stacked bar chart or a line chart (depending on the metric) breaks down the volume over time, showing the ratio of all work with all AI tools (dark blue) and No AI work (grey) for each interval. Hovering reveals specific tooltips for precise tracking.
### Team metric breakdown
This **Team Metric Breakdown** table serves as a matrix ledger that combines structural team performance data with statistical benchmarks. While previous dashboards isolated *either* adoption numbers or high-level charts, this view maps multi-column impact metrics side by side across your specific users and team sub-structures.
Each major metric category (for example, PRs Completed, Time to Merge, and Iterated PRs) is split into three granular data points per row: **All**, **All AI**, and **Without AI**.
The table provides an immediate baseline comparison on a single horizontal axis. You can immediately see a team's total average alongside exactly how their AI-assisted work compares to their unassisted work.
With such an overview, you can avoid false conclusions based on overall averages alone.
For example, the average Time to merge is 108 hours across teams. However, looking closer, work with AI finishes significantly faster (91.2 hours) compared to work without AI (129.6 hours). This clearly attributes the efficiency gain directly to the use of AI tools rather than to general process changes.
## Champions
The **Champions Dashboard** helps engineering leaders identify their top AI tool adopters, understand what makes them effective, and use those insights to drive broader adoption across their teams.
The Champions tab consists of the following sections.
#### KPI summary (cards)
The top of the Champions tab displays the core metrics related to your team’s AI tools adoption:
- **Total champions** = Total users with 5+ active days within the date range using AI tools.
- **Power users (Coach pool)** = A total count of users scoring 70+ (out of 100) who can help coach others.
- **Average score** = Mean score across all ranked users (expressed on the 0-100 scale). Score represents a weighted percentile that combines consistency, code output, agent usage, completions, and engagement.
- **Total agent edits** = Sum of all accepted autonomous agent code edits across all ranked users.
- **Total AI lines** = Sum of all AI-generated lines of code accepted into the codebase across all ranked users.
#### Champions: Top power users across all tools ranked by score
The leaderboard ranks every eligible user by their overall Champion score. On this board, you can:
- Sort by any column to find champions in specific areas (for example, sort by Agent Edits to find your top agent users).
- Search by name, email, or login to find specific users.
- See which AI tool a specific team member uses.
The leaderboard consists of the following columns:
- **Score**: A weighted percentile (expressed on the 0-100 scale) combining consistency, code output, agent usage, completions, and engagement.
- **Active days**: Days a listed user actively used Cursor.
- **AI lines**: AI-generated lines of code accepted into the codebase.
- **Agent edits**: Autonomous agent code edits accepted by the user.
- **Total requests**: All AI interactions.
#### Category Champions
> 📝 Some categories have eligibility thresholds (50+). In such cases, the threshold is indicated with a tooltip.
The Category Champions dashboard identifies the top performers across each specific dimension of Cursor usage. This view highlights the five most proficient users within each category:
- **Agent champion**: A user who accepts the most autonomous agent code edits. They lead in integrating large-scale, agent-driven modifications into the codebase.
- **Code champion**: A user who accepts the most Cursor-generated lines. They incorporate the most AI-suggested code into their active files.
- **Most consistent**: A user who maintains the most active days using Cursor. They demonstrate the most consistent, daily engagement with this AI tool.
- **Tab ace**: A user who achieves the highest tab completion accept rate. They show the most efficient alignment with predictive suggestions (requires 50+ suggestions shown).
- **Chat champion**: A user who initiates the most chat/ask mode interactions. They lead the team in leveraging conversational AI for explanations and troubleshooting.
- **Composer champion**: A user who initiates the most composer/edit mode interactions. They frequently use advanced editing modes to refactor or build complex code blocks.
#### Power Users vs Others
The Power Users vs Others dashboard provides a comparative analysis of the top 10% of users (power users) and the remaining users in your organization. For each metric, you can see:
- The average value for power users vs the rest.
- A color-coded multiplier showing how much more productive power users are in each metric:
- If the multiplier is 1.5x or higher, it turns **green**.
- If the multiplier is below 1.5x, it turns **gray**.
> 📝 The multiplier is determined by dividing the power user average by the average of all other users. A 2.0x multiplier indicates that power users achieve double the average output for that specific metric.
#### Adoption Tiers
The Adoption Tiers dashboard provides a color-coded visual distribution of your team members based on their overall Champion score.
The Champion score is a weighted percentile ranging from 0 to 100, calculated by aggregating performance across five distinct dimensions of AI tool usage.
| **Dimension** | **Weight** | **What does it capture?** |
| --- | --- | --- |
| **Consistency** | 25% | How regularly the user engages with AI tools (active days). Heavily weighted because sustained usage drives real workflow change. |
| **AI Code Output** | 25% | Volume of AI-generated code accepted into the codebase (lines added). Measures productive output, not just activity. |
| **Agentic Mastery** | 20% | Use of autonomous agent workflows (agent edits accepted). Reflects the adoption of the most advanced AI capabilities. |
| **Completion Adoption** | 20% | Use of inline/tab completions. The bread-and-butter of AI-assisted coding. |
| **Engagement** | 10% | Total interactions (chat, composer, agent requests). Captures the breadth of AI tool exploration. |
> ℹ️ Why percentiles?
> ℹ️
> ℹ️ The score compares each user against their peers, not against an absolute standard. A score of 80 means the user is in the 80th percentile across all dimensions. This makes the score meaningful regardless of team size or total usage volume.
The color-coded distribution is as follows:
- **Power user **(green): Champion score = 70+. These are your top AI adopters—individuals who consistently use AI tools, generate a high volume of code, and leverage complex agent workflows. Because of their deep engagement and proficiency, they are ideal candidates to coach others and drive team-wide adoption.
- **Active **(blue): Champion score = 40-69. These are regular users with steady habits and significant growth potential. While they demonstrate consistent tool usage, they have yet to fully leverage the most advanced AI capabilities or complex agent workflows.
- **Light** (orange): Champion score = 10-39. These are occasional users who have begun integrating AI tools into their workflow but have yet to establish consistent usage habits or a predictable cadence.
- **Minimal** (purple): Champion score: 0-9. These are users with minimal engagement with AI tools despite having access to them. They may require targeted onboarding support or assistance with integrating AI into their specific workflows to see its full value.
## How to use data from the Champions dashboard
### Identify coaching opportunities
- **Assess the coach pool:** Monitor the Power Users count to identify individuals naturally suited for peer mentorship.
- **Identify specialists:** Use the Category Champions section to find niche experts—such as an Agent Champion—who can demonstrate specific advanced workflows to the broader team.
- **Quantify adoption impact:** Use the Power Users vs Others multipliers to provide concrete evidence of AI value (for example, Power users accept 2x as many agent edits as the average user).
### Track adoption progress
- **Analyze distribution trends:** Monitor the Adoption Tiers over time; a healthy trajectory shows a steady migration of users from Minimal or Light tiers into Active and Power User status.
- **Evaluate team growth:** Use the Average Score KPI card as a high-level pulse check on overall organizational improvement.
- **Measure tool penetration:** Compare the Ranked Users count to your total licensed seats to determine the deployment's true reach.
### Target training investments
- **Focus on high-potential users:** Users in the Light tier are your primary training targets, as they have already begun using the tools but require support to build consistent habits.
- **Address onboarding gaps:** Minimal-tier users may require different interventions, such as foundational onboarding, workflow integration, or the removal of technical blockers.
- **Spot skill deficiencies:** Identify categories with low qualification rates; for example, if the Agent Champion slot remains empty, it indicates a specific need for agent-led workflow training.
## Champions dashboard: Limitations
| **Item** | **Description** |
| --- | --- |
| **Date range** | Champions data is capped to the most recent 30 days for performance. A note appears when the selected range exceeds this. |
| **Minimum data** | Users need 5+ active days to appear. At least 3 eligible users are needed for rankings. |
| **Tab Accept Rate** | Only shown when 50+ completions were shown. This avoids misleading rates from small samples. |
| **Score is relative** | Scores are percentile-based. A user's score can change even without changing their behavior if peers change. |
## Cost analysis
The **Cost analysis** tab is designed around visibility, budget accountability, and financial predictability for Claude across the entire organization.
### Cost analysis
The **Cost analysis** table translates variable AI token consumption and seat metrics into predictable financial reporting, preventing budget surprises and enabling proactive capacity planning.
By surfacing unit economics like average spend per user and isolating overage costs, it allows leadership to accurately measure true ROI and optimize software vendor allocation.
Ultimately, the dashboard transforms raw usage data into actionable business insights that help procurement teams maximize tier efficiency and justify tool expansions.
> 📝 Cost values are based on usage data and may not reflect your exact billing.
> 📝
> 📝 Use the **Usage-based billing** toggle next to Claude to switch between usage data-based and billing-based cost analysis.
- **Total cost** / **Token spend** / **Total spend**: By aggregating macro spend data and token consumption metrics, the dashboard effectively helps prevent budget surprises. Generative AI introduces variable-cost structures into engineering budgets that have historically relied on flat SaaS licensing. The cost analysis breakdown translates raw LLM metrics into predictable financial reporting, allowing finance and engineering leaders to track total consumption trends at a glance.
- **Average per user / Cost per seat**: To help organizations measure the true ROI justification of AI platforms, the cost analysis dashboard normalizes broad expenditure into unit economics, such as average spend per user and monthly cost per seat. To evaluate whether an AI tool truly pays for itself, leadership must know exactly what it costs the business to equip an individual developer. This capability allows decision-makers to weigh the average cost per developer directly against efficiency gains—like hours saved—to build an accurate, data-backed business case for seat renewals or expansions.
The **Cost per seat per month** is blank by default. Enter the costs applicable to your Claude and Cursor plans manually.
- **Overage cost**: By isolating and surfacing dedicated overage cost metrics, the platform identifies waste and tier-optimization opportunities. High overage costs indicate that a team has outgrown its current subscription tier or that a few power users are driving up variable expenses. Seeing this number clearly helps procurement teams determine precisely when it is more cost-effective to upgrade to a higher enterprise tier than to absorb pay-as-you-go penalties.
- **Financial trends**: Financial trend mapping provides color-coded directional indicators for costs and token utilization, enabling proactive capacity planning. Instead of reacting to a massive end-of-quarter invoice, these trend lines provide early warnings of exponential spend growth. This allows infrastructure and engineering leaders to adjust usage guardrails, implement prompt-engineering best practices, or dynamically reforecast budgets before overages compound.