Why the $30 Figure Is Not the Real Price

Microsoft markets Microsoft 365 Copilot at $30 per user per month. This number appears in budget proposals, vendor presentations, and board-level business cases. It is not the cost of Copilot — it is the cost of the add-on licence, which cannot be used without a qualifying base Microsoft 365 subscription. The true per-user cost is always the base licence plus $30, making the TCO calculation licence-tier dependent and significantly higher than the marketed figure.

On a Microsoft 365 E3 base ($36/user/month), the true per-user cost of Copilot at list price is $66/user/month. On E5 ($57/user/month), it is $87/user/month. At 1,000 users on E3 with Copilot, the annual licence cost is $792,000 before EA discounts. At 10,000 users, it is approximately $7.9 million per year — and that is before implementation costs, training investment, data readiness work, and Copilot Studio consumption.

The E7 Alternative: Better or Worse TCO?

Microsoft 365 E7 at $99/user/month bundles E5, the $30 Copilot add-on, Agent 365 ($15), and Entra Suite ($12) into a single SKU. The M365 SKU stack now runs E1 → E3 → E5 → E7, with E7 as the top tier. For organisations already on E5 that need Copilot, Agent 365, and Entra Suite, E7 at $99 offers better unit economics than purchasing those components separately at list price (which totals approximately $114/user/month). However, for organisations on E3 that only need Copilot, upgrading to E7 adds $63/user/month of capability (E5 security, Agent 365, Entra Suite) that may not be needed or activated. The E7 versus add-on decision must be based on actual capability requirements — not on Microsoft's commercial preference, which is clearly E7 at renewal.

The Hidden Cost Layers: What Is Not in the Price List

Data Readiness: SharePoint Permission Remediation

Before Copilot can be safely deployed at enterprise scale, most organisations need to conduct a SharePoint permission audit and remediation exercise. Copilot accesses content using the user's existing permissions — which means that overly broad access configurations (the "Everyone" permission pattern on SharePoint sites is endemic in enterprise Microsoft 365 tenants) create data governance risk when Copilot is activated. For a 5,000-seat organisation, a comprehensive SharePoint permission audit and remediation typically requires four to eight weeks of internal IT effort or external consultancy, at a cost of £40,000 to £120,000 depending on the complexity of the SharePoint estate. This cost is not in the Copilot price list, but it is a prerequisite for responsible deployment.

Training and Change Management

Copilot adoption does not happen without deliberate training investment. The minimum viable training programme for an enterprise Copilot rollout includes role-specific prompt libraries, 15 to 30-minute onboarding sessions for each user cohort, a champions programme for peer-to-peer adoption support, and ongoing reinforcement for the first 90 days. For a 1,000-user rollout, training and change management investment typically runs at £15,000 to £50,000 depending on the number of distinct job functions and the maturity of the organisation's change management capability. Under-investing in training is the single most common reason for low Copilot adoption, which in turn increases the shelfware cost.

Copilot Studio: Consumption-Based Cost Volatility

Microsoft Copilot Studio — the platform for building custom AI agents — is priced entirely separately from the Copilot add-on. The Studio licence is sold as 25,000 Copilot Credits for $200 per pack per month. Credits are consumed when agents respond to queries or execute actions; the per-interaction credit cost varies based on agent complexity and knowledge sources. A conservative enterprise agent deployment handling 5,000 queries per month from knowledge workers might consume two to three credit packs per month ($400 to $600/month). A customer-facing agent handling 50,000 interactions with complex workflows could consume 20 to 30 packs per month ($4,000 to $6,000/month).

The key TCO risk with Copilot Studio is that consumption scales with adoption success. An agent that becomes genuinely useful will be used heavily, generating credit consumption that grows with the deployment. Organisations that include a fixed Copilot Studio budget in their EA without a consumption baseline are routinely surprised by Studio costs in the second half of their EA term. Pay-as-you-go metering through Azure is available as an alternative, but it eliminates the cost predictability that the EA is intended to provide.

"The most expensive Copilot deployment we have reviewed was a 3,000-user commitment on E5 with full-tenant Copilot. After 12 months, 68% of licences were unused. The annual cost of that shelfware was $1.1M."

The Shelfware Cost: The Largest Hidden Risk

Shelfware — licences that are contractually committed but not actively used — is the highest-probability hidden cost in enterprise Copilot deployments. Based on enterprise deployment patterns observed across the market, 40 to 70% of Copilot licences are inactive at the 12-month mark in organisations that did not run a structured pilot and did not invest in change management before full-tenant commitment. The shelfware cost is not visible in the initial EA proposal — it becomes visible at the next True-Up or renewal when the business case is reviewed and the ROI calculation produces a deeply unflattering result.

The shelfware risk compounds with scale. A 1,000-user commitment with 60% shelfware means 600 unused licences at $30/user/month — $18,000 per month or $216,000 per year in pure waste. Over a three-year EA term, that is $648,000 in unrecovered licence cost from a deployment that appeared to cost only $360,000/year. The EA discount (10 to 15% on Copilot SKUs in a well-negotiated agreement) does not compensate for shelfware at this scale.

Building an Accurate Copilot TCO Model

An accurate Copilot TCO model includes six components: the base licence cost (E1, E3, or E5) for all users receiving Copilot, the Copilot add-on cost ($30/user/month, discounted by the EA rate) for the committed user count, the projected active user rate based on pilot data or industry benchmarks (used to calculate the effective cost per active user), data readiness and governance preparation costs (one-time), training and change management costs (one-time for the initial rollout, recurring for new cohorts), and Copilot Studio capacity costs if agent development is in scope.

The effective cost per active Copilot user — the metric that should drive the business case — is calculated as total annual licence cost divided by monthly active users. At 70% adoption (a strong pilot outcome), the effective cost per active user on E3 is approximately $94/user/month. At 40% adoption (a poor deployment without structured change management), the effective cost per active user rises to $165/user/month. At that level, the productivity savings required to justify the investment are very difficult to achieve across a broad knowledge worker population.

E7 TCO vs. Selective Add-On: A Worked Example

Consider a 2,000-user organisation on Microsoft 365 E5 planning to deploy Copilot to all users. Path A (E7 upgrade) costs $99/user/month × 2,000 users = $198,000/month or $2.376M/year at list price. Path B (E5 plus selective Copilot add-on for 800 high-value users based on pilot data) costs ($57 × 2,000) + ($30 × 800) = $138,000/month or $1.656M/year at list price. Path B saves approximately $720,000 per year at list price — before EA discounts on either path. The difference narrows if E7 is successfully negotiated at 12 to 15% discount, but the selective path consistently saves meaningful budget while reducing shelfware risk to near zero.

The decision changes when Agent 365 and Entra Suite capabilities (included in E7 but not in E5 plus add-on) are genuinely needed at scale. Organisations that plan to build and govern 10 or more Copilot Studio agents, deploy advanced identity governance at enterprise scale, and activate advanced Intune Suite capabilities may find E7 provides better long-term unit economics. The analysis must be done on actual capability needs, not vendor projections of what you will eventually want.

For a complete guide to the Copilot licensing landscape including E7 and SKU tiers, see the Microsoft Copilot Licensing Guide 2026. For independent TCO modelling support, visit our Microsoft licensing advisory services or the Microsoft Knowledge Hub.

Need an independent Copilot TCO model before your next EA renewal?

Our Microsoft EA negotiation specialists build scenario-based models that capture all cost layers.
Request a TCO Review →
FF
Fredrik Filipsson
Co-Founder, Redress Compliance

20+ years in enterprise software licensing. Co-Founder of Redress Compliance, an independent buyer-side advisory firm with 500+ engagements across EMEA and North America. Gartner recognised. Specialises in Microsoft EA, MCA, Copilot TCO modelling, and commercial negotiation strategy.

Connect on LinkedIn →