Three AI Metrics Your CFO Already Has but Nobody Is Reading

Executive Summary

• Every mid-market company paying for Microsoft 365 Copilot, Salesforce Einstein, Google Workspace Gemini, or Zoom AI Companion already has usage reports sitting in admin consoles that nobody pulls. These reports reveal adoption rates, active users by department, feature engagement, and licensing utilization — data that turns a vague “is AI working?” into a concrete cost decision.
• The adoption gap is real: Worklytics benchmarks show medium-sized companies (101-1,000 employees) average 61% active seat utilization for AI tools, meaning 39% of purchased licenses generate zero return (Worklytics 2025 Enterprise AI Adoption Benchmarks). At $30/user/month for Microsoft 365 Copilot, a 300-person company with 100 licensed seats and 61% utilization is burning $14,040/year on seats nobody touches.
• Zylo’s 2026 SaaS Management Index (40 million licenses, $75 billion in spend analyzed) finds 36% of enterprise software licenses go unused and 78% of IT leaders encountered unexpected AI-related charges in the past 12 months. The median SaaS spend per employee: $9,455. AI features embedded in existing tools are inflating that number whether or not anyone uses them.
• Three admin console reports — Microsoft 365 Copilot usage, Google Workspace Gemini activity, and Salesforce Agentforce analytics — already exist in tools the company pays for today. The CFO who pulls these reports Monday morning has an immediate cost, adoption, and governance decision to make. No new software required.

The Reports That Already Exist (and Where to Find Them)

Most companies tracking AI adoption are building new dashboards, hiring analysts, or buying third-party tools. They do not need to. The four most common SaaS platforms in mid-market companies already generate AI usage data that answers the CFO’s three fundamental questions: who is using it, how much, and is it worth what we pay?

Microsoft 365 Copilot Usage Report

Where: Microsoft 365 Admin Center > Reports > Usage > Microsoft 365 Copilot tab.

What it shows: Enabled users vs. active users, adoption rate by app (Word, Excel, PowerPoint, Outlook, Teams), retention over 7/30/90/180-day windows, and — since the July 2025 admin center update — departmental segmentation by Entra ID attributes (department, office, job title, manager). The Copilot Dashboard in Viva Insights adds external benchmarks comparing adoption against companies of similar size and industry (Microsoft Learn, updated February 2026).

What to look for: The gap between “enabled” and “active.” Microsoft disclosed 15 million Copilot seats purchased globally as of January 2026 — but that represents only 3.3% of the 450 million Microsoft 365 commercial installed base (Microsoft FY26 Q2 earnings, January 2026). Recon Analytics’ survey (n=150,000+ U.S. respondents, January 2026) found a 35.8% conversion rate from access to active usage. The enterprise average active seat utilization for medium-sized companies is 61% (Worklytics 2025 Benchmarks). If your number is lower, you are paying for shelfware.

Cost math: At $30/user/month, 100 Copilot seats cost $36,000/year. At 61% utilization, $14,040 is wasted. At the 35.8% active usage rate Recon Analytics found, $23,112 is wasted. Microsoft is raising Copilot pricing — some enterprise tiers are moving toward $60/user/month effective mid-2026 (WebProNews, March 2026). The cost of not checking this report doubles.

Google Workspace Gemini Usage Report

Where: Google Admin Console > Menu > Generative AI > Gemini Reports > Org-level usage (or User-level usage for individual tracking).

What it shows: Active Gemini users, usage per Workspace app (Gmail, Docs, Sheets, Slides, Chat), last-used timestamps per user, and — as of the February 16, 2026 rollout — feature-specific usage and threshold reports showing when users hit AI usage caps (Google Workspace Updates Blog, February 2026). Audit logs available since July 2025 provide prompt-level detail for security review.

What to look for: Usage concentration. Gemini usage tends to cluster in one or two apps (typically Gmail and Docs) while going unused in Sheets, Slides, and Chat. The threshold reports are particularly valuable: they show which users are hitting their Gemini usage limits — the employees getting real value — and which have never triggered a single AI interaction. The gap between these groups tells the CIO where training is working and where it is not.

Salesforce Agentforce Analytics Dashboard

Where: Salesforce Dashboard tab > Agentforce (Default) folder. Setup > search “Einstein Audit, Analytics, and Monitoring Setup” to enable full logging (prompts, masking events, toxicity scores, user feedback routed to Data Cloud).

What it shows: Adoption rates, actions used, average interactions per user, success rates, and user feedback. The Spring 2026 release adds agent-specific event logging through Setup > Agentforce Agents (Salesforce Help, Spring '26 Release Notes). Agentforce uses a consumption-based pricing model, making usage data directly tied to cost — unlike flat per-seat licensing.

What to look for: The gap between the Salesforce AI features included in your existing license and the ones you actually activated and use. Many mid-market companies on Enterprise Edition have Einstein features they have never configured. The Agentforce analytics dashboard shows whether the AI features your sales team was promised are delivering measurable pipeline impact — or sitting dormant.

Zoom AI Companion Dashboard

Where: Zoom Web Portal > Dashboard > AI Companion tab. Available on Business plans and above.

What it shows: Meeting summary usage (hosts and meetings), chat compose events, smart recording adoption, voicemail prioritization, and whiteboard AI activity. Filterable by department, group, and timeframe (daily, weekly, monthly). Identifies the top 10 most active users. Exportable to CSV (Zoom Support, 2026).

What to look for: The gap between what Zoom AI Companion costs (included free on Pro and above; $12/user/month for Custom AI Companion add-on) and what it delivers. AI Companion’s base features are bundled — the cost is hidden in the platform license. The dashboard reveals whether anyone is using the features the company is implicitly paying for, and whether the custom add-on (if purchased) justifies the additional spend.

Why Nobody Is Pulling These Reports

The data exists. The reports are free. The admin consoles are already accessible. Three patterns explain why mid-market companies leave this intelligence untouched.

The wrong person owns the question. The CFO asks “is AI worth what we spend?” The CIO manages the admin consoles. The CHRO wants to know about adoption. None of these three are in the habit of running reports from each other’s systems. Zylo’s 2026 SaaS Management Index finds that business units control 81% of SaaS spend while IT directly manages only 15%. The person who pays is not the person who has the data.

The reports exist but nobody is assigned to read them. Mid-market IT teams of 8-16 people are already fully allocated to infrastructure, security, and support. The Flexera 2026 IT Priorities Report finds 94% of IT leaders are integrating AI into their stacks, but the Foundry State of the CIO Survey (2025) shows the CIO at a 300-person company is already spending 75% of their time on AI/ML initiatives alongside cybersecurity, infrastructure, and vendor management. Running a monthly Copilot usage report is a 10-minute task nobody has 10 minutes for.

The data is uncomfortable. A CFO who discovers 12% Copilot adoption on 100% seat licensing now faces a political problem. Someone approved that purchase. Departments that championed AI adoption look bad if the numbers are low. The 78% of IT leaders who encountered unexpected AI charges in the past year (Zylo 2026) did not set out to ignore the data — they set out to avoid the conversation the data forces.

Key Data Points

What This Means for Your Organization

This is not a research exercise. This is a 15-minute task.

Have your IT team pull three reports Monday morning: the Microsoft 365 Copilot usage report, the Google Workspace Gemini usage report, and your Salesforce Agentforce analytics dashboard (substitute whichever platforms you actually run). Ask one question of each: what percentage of licensed users are active in the last 30 days? Compare that number against the Worklytics benchmark of 61% for medium-sized companies. If you are below that line, you are paying for shelfware. If you are above it, you have a proof point for your next budget conversation.

The math is straightforward. A 300-person company licensing 100 Copilot seats at $30/user/month spends $36,000/year. If 39% of those seats are inactive (the benchmark gap), that is $14,040/year of recoverable spend — or redirectable investment into training, workflow redesign, or expanding AI to departments where adoption is actually working. With Microsoft raising Copilot pricing to $60/user/month for some tiers, the cost of ignoring this report doubles by year-end.

The harder question is what to do with the answer. Low adoption does not mean AI failed — it means the deployment skipped the adoption work that separates the 5% from the 95%. The success metrics card defines what to measure after the pilot. This card tells you what you can measure right now, before the pilot even starts, using reports you already own. If the gap between what you are reading here and what your organization is doing raises questions, I am happy to work through the specifics — brandon@brandonsneider.com.

Sources

1. Worklytics 2025 Enterprise AI Adoption Benchmarks — Medium-company (101-1,000 employees) active seat utilization at 61%, with industry breakdowns. Independent analytics platform. Credibility: Moderate-high — primary data from enterprise deployments, methodology not fully public. https://www.worklytics.co/resources/benchmark-copilot-gemini-adoption-2025-enterprise-averages-dashboard

2. Zylo 2026 SaaS Management Index — 40 million licenses under management, $75 billion in spend analyzed. 36% unused licenses, $9,455 median SaaS spend per employee, 78% of IT leaders encountered unexpected AI charges. Credibility: High — large dataset, annual methodology, vendor-neutral (SaaS management platform, not an AI vendor). https://zylo.com/reports/2026-saas-management-index/ and https://www.prweb.com/releases/zylos-2026-saas-management-index-finds-ai-native-app-adoption-is-surging-with-chatgpt-now-the-most-expensed-app-302673438.html

3. Microsoft FY26 Q2 Earnings — 15 million Copilot seats purchased, 3.3% of 450 million M365 commercial installed base. January 2026. Credibility: High — SEC-filed financial disclosure. https://www.techradar.com/pro/barely-any-microsoft-365-users-are-actually-paying-for-copilot-despite-microsoft-ceo-satya-nadella-claiming-it-is-a-true-daily-habit

4. Recon Analytics — 35.8% Copilot conversion rate (access to active use), n=150,000+ U.S. respondents, January 2026. Copilot accuracy NPS at -19.8. Credibility: High — large sample, independent research firm, telecom/tech focus. https://www.stackmatix.com/blog/copilot-market-adoption-trends

5. Microsoft Learn: Copilot Usage Reports — Admin center reporting documentation, departmental segmentation, Viva Insights Copilot Dashboard with external benchmarks. Updated 2026. Credibility: High — vendor primary documentation. https://learn.microsoft.com/en-us/microsoft-365/admin/activity-reports/microsoft-365-copilot-usage and https://learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-reports-for-admins

6. Google Workspace Updates Blog — Gemini feature usage and threshold reports rollout, February 16, 2026. Credibility: High — vendor primary documentation. https://workspaceupdates.googleblog.com/2026/02/view-gemini-feature-usage-and-threshold.html

7. Google Workspace Admin Console: Gemini Reports — Org-level and user-level usage reporting, audit logs since July 2025. Credibility: High — vendor primary documentation. https://knowledge.workspace.google.com/admin/gemini/review-gemini-usage-in-your-organization

8. Salesforce Help: Agentforce Analytics — Dashboard access, Einstein audit setup, agent-specific event logging, Spring '26 release. Credibility: High — vendor primary documentation. https://help.salesforce.com/s/articleView?id=sf.copilot_analytics.htm

9. Zoom Support: AI Companion Dashboard — KPIs, department filtering, feature-level adoption tracking. Credibility: High — vendor primary documentation. https://support.zoom.com/hc/en/article?id=zm_kb&sysparm_article=KB0074179

10. Worklytics: Tracking Microsoft Copilot Usage by Department — July 2025 admin center update details, departmental segmentation via Entra ID attributes. Credibility: Moderate-high — third-party analytics vendor, primary data from integrations. https://www.worklytics.co/resources/tracking-microsoft-copilot-usage-by-department-july-2025-admin-center-update

11. Flexera 2026 IT Priorities Report — 94% of IT leaders integrating AI; 80% increasing AI spending. Credibility: High — established IT management research firm, annual survey. Referenced via findings/it-capacity-ai-ownership-card.md

12. Foundry State of the CIO Survey (2025) — CIO time allocation on AI/ML initiatives at 75%. Credibility: High — established media/research company, annual survey. Referenced via findings/it-capacity-ai-ownership-card.md

Brandon Sneider | brandon@brandonsneider.com March 2026