License Optimization Based on Actual Usage, Not Assumptions

The engine doesn't start from a spreadsheet of who was assigned what license three years ago. It starts from what is actually happening inside the platform today, across twelve data points:

• Assigned licenses — license type and tier currently held by every user, cross-referenced against role.
• Feature usage — which modules, objects, and workflows each user actually touches inside the platform.
• Login history — frequency and recency of access, not just whether an account exists.
• Permission sets — every permission set and profile assignment, checked against the access the user's role requires.
• Role hierarchy — whether reporting-line and approval access matches actual organizational structure.
• Object access — record-level and object-level usage patterns per user and per team.
• API usage — call volume and patterns per integration, connected app, and individual user.
• Inactive accounts — users with no login activity within a defined window.
• Duplicate users — the same person provisioned twice across business units, sandboxes, or merged systems.
• Premium-feature usage — whether users on premium-tier licenses are actually using the features that justify the premium tier.
• Contract commitments — minimums, tier structures, and bundled SKUs already locked in.
• Renewal milestones — upcoming renewal and notice-period dates, so recommendations land before a deadline, not after.

Every engagement follows the same disciplined sequence — the same twelve steps whether the platform is Salesforce, ServiceNow, or NetSuite:

• 1. Import license and user data from platform admin APIs, your identity provider, and existing contract or order-form documentation.
• 2. Map every license type to the users currently assigned to it, by department and by role.
• 3. Cross-reference login history against assignment to separate active users from dormant seats.
• 4. Analyze feature and permission-set usage per user, not just per license category.
• 5. Assess object and API access patterns against the license tier each pattern actually requires.
• 6. Identify inactive, duplicate, and orphaned accounts that are consuming paid seats without corresponding usage.
• 7. Compare premium-feature usage against premium-tier license spend to catch tier mismatches in both directions.
• 8. Score every user as over-licensed, correctly licensed, or under-licensed relative to their actual activity.
• 9. Model department-level and company-level license mix scenarios to see the cost impact of alternative configurations.
• 10. Validate every proposed change against your vendor agreement, contract terms, and platform rules before it's presented.
• 11. Present recommendations for approval — your FDE and your team review and sign off; nothing is executed automatically.
• 12. Execute approved changes and track realized savings against the original model, quarter over quarter.

The engine produces documentation built for finance, IT, and platform admins to act on — not a slide deck:

• License utilization dashboard — live view of assigned vs. active vs. idle licenses across every connected platform.
• User-level recommendations — a specific action (keep, downgrade, remove, reassign) for every individual user, with the usage evidence behind it.
• Department-level cost report — license spend and utilization rolled up by team, so budget owners see their own exposure.
• Renewal preparation report — the exact position to bring into your next renewal negotiation, tied to contract milestones.
• Inactive-user report — every dormant account, flagged with last-login date and recommended disposition.
• Portal and automation opportunity reports — cases where an approved external experience or workflow automation could replace a full internal license.
• Three-year savings model — projected cost trajectory under the optimized license mix vs. the current run rate, as an estimate to validate against your actual agreement.
• Governance checklist — the recurring cadence and controls needed to keep the license mix aligned with usage after the initial pass.

Under a traditional time-and-materials engagement, there's no structural reason for anyone to go looking for licenses you don't need — the incentive runs the other way. Renewals get processed on autopilot: same tier, same seat count, same assumptions as last year, because nobody on the billing clock is rewarded for shrinking the invoice.

PartnerMCP's License Optimization agent is not paid by the hour spent investigating. It's one part of an AI Delivery Engine whose job is to find where the license mix and actual usage have drifted apart, and to bring your Cost Analysis and User Utilization agents' findings into a single, validated recommendation — before your next renewal, not after you've already re-signed.

License decisions carry real operational risk if they're wrong — a downgraded user who actually needed platform access, a removed integration account that breaks a nightly sync. That's why every recommendation from the License Optimization Engine passes through validation against your vendor agreement, security requirements, and platform rules before it's presented, and through your team's explicit approval before anything is executed.

Your dedicated Forward Deployed Engineer owns that review — reading the usage data alongside the Discovery and Architecture agents' findings, confirming the contract language supports each recommended change, and sequencing execution so operations are never disrupted mid-cycle.