Microsoft 365 Copilot at $30: One Year Later, Where the Enterprise Stands

Microsoft 365 Copilot launched at $30 per user per month with the implicit promise that AI-augmented Office workflows would produce 10–30% productivity gains for the modal knowledge worker. Twelve months in, the adoption pattern has been more disappointing than the marketing implied, the productivity claims have softened in the company's own messaging, and the enterprise customer base is reassessing whether the broad rollout was the right strategy.

We tracked Copilot deployments at four enterprise customers through Q3 2025: a 4,200-employee financial services firm, a 1,800-employee professional services company, an 800-employee SaaS company, and a 320-employee fintech.

Where Copilot earns its seat

Outlook and Teams. The meeting summarization, action-item extraction, and email-draft features in the Microsoft-native communication tools are genuinely useful. For employees with high meeting load and high email volume — typically managers and above — Copilot produces measurable time savings.

The cross-product context is the second strength. Microsoft Graph's integration with the rest of the Microsoft 365 surface produces a "find an answer across my company's data" experience that is genuinely better than what standalone AI assistants can match. For organizations whose institutional knowledge lives in SharePoint, OneDrive, and Teams, this is the most defensible Copilot use case.

The compliance and data-residency story is the third durable advantage. For regulated industries — financial services, healthcare, defense — Copilot's data-residency and compliance commitments are the strongest in the AI-assistant category. This is a real procurement-side advantage that is hard for OpenAI or Anthropic to match.

Where the case breaks down

Word and Excel. The promised productivity gains in the core Office productivity apps have not materialized in the way the marketing implied. Word's Copilot is useful for first-draft writing and approximately neutral for editing. Excel's Copilot remains, two years in, mostly a curiosity rather than a productivity tool. The "ask Copilot to analyze this spreadsheet" workflow produces reasonable answers for simple questions and unreliable answers for the kind of analysis a real Excel user needs.

Adoption is the second issue. Across our test sites, the percentage of seats using Copilot more than weekly, three months post-deployment, ranged from 22% to 41%. The remaining 60–80% of seats were paying for the product without using it. The pattern is consistent enough that the broad-rollout strategy now looks like the wrong one.

The "all-employees-on-Copilot" rollout pattern has produced visible disappointment. The selective deployment pattern is the one that's working.

On the alternatives

ChatGPT Team at $25 per seat and Claude for Business at similar pricing produce comparable value for most general-purpose AI-assistant use cases. The Microsoft Graph integration is the one feature these alternatives cannot match for organizations deeply embedded in the Microsoft 365 ecosystem. For organizations whose data lives in Google Workspace, Notion, or other non-Microsoft systems, the alternatives produce equivalent or better value at lower cost.

We have watched two of our test customers reduce Copilot deployment scope materially in the last six months. The remaining seats are concentrated in Outlook-and-Teams-heavy roles where the use case is strongest. The pattern is becoming the consensus enterprise deployment shape.

What to do

For Microsoft 365 customers evaluating Copilot: deploy selectively. Start with managers, sales operations, and customer-facing teams whose meeting and email load is highest. Pilot for 90 days. Measure usage. Roll out further only where usage justifies it.

For Microsoft to consider: the licensing model has been the largest single source of customer disappointment. Per-user-per-month pricing creates pressure to deploy broadly. A usage-based or selective-tier model would produce better customer experience and, in our reading, better long-term retention.

The verdict

Copilot is real and useful for specific workflows. It is not the universal productivity unlock the marketing implied and the pricing model has produced visible adoption failure. The product earns its 3.5 on the strength of the Outlook-and-Teams use case. The path to a higher rating runs through better packaging, not through better features. We expect Microsoft to revisit the licensing model in 2026; the current deployment pattern is not sustainable.