GitHub Enterprise is billed per seat, and the seat count has a way of climbing without anyone deciding it should. Members are added for a single pull request and never removed, contractors keep access after a project ends, and outside collaborators quietly become billable. Layered on top sits Copilot, billed separately per seat, frequently assigned to the entire organization rather than to the developers who actually use it. The result is an invoice that reflects everyone who was ever granted access, not the active engineering team. Optimization works the active membership, the Copilot assignment, and the plan tier as three separate questions. Reclaim the inactive and orphaned seats, match Copilot to measured usage rather than blanket entitlement, and confirm the Enterprise tier is justified over Team for the features actually in use. Run before renewal, when the seat count is locked in for the next term, this is where the largest and most durable savings sit.
GitHub seat inflation comes from three distinct places, and each one fills quietly. None of them announce themselves on the invoice as waste, which is exactly why the seat count keeps climbing while the active engineering headcount stays flat.
A developer added for one task who never contributes again, or a team member who moved to a role that does not touch code. Each consumes a billable seat indefinitely. Inactive members are the largest pool in most organizations, and because access is rarely revoked when activity stops, they accumulate term after term until someone measures sign in and contribution against the member list.
Contractors, vendors, and outside collaborators granted access for a project that ended months ago. They still hold billable seats, and they also carry a security exposure that compounds the cost problem. Offboarding rarely reaches GitHub, so these seats persist until a deliberate review of external access removes the ones whose engagement is over.
Copilot is billed per seat on top of the base seat, and it is frequently assigned to the whole organization on the assumption that everyone wants it. Usage telemetry almost always shows a meaningful share of assigned seats with little or no activity. Matching Copilot to developers who actually use it, rather than to everyone with a login, is often the single largest line in the optimization.
Beyond the seat count sits the plan tier. GitHub Enterprise carries a premium over Team for advanced security, governance, and identity features. The question is whether the organization actually uses those features or is paying the Enterprise premium for capabilities sitting idle.
Enterprise is the right tier where the organization genuinely relies on its advanced capabilities: centralized identity and access management, advanced security and code scanning, audit and compliance controls, and enterprise wide policy enforcement. For regulated environments and large engineering organizations with real governance needs, the premium buys capability that is actively in use and worth paying for.
Where the Enterprise only features are not actually used, the premium over Team is pure waste. We test feature usage against the tier and confirm whether Enterprise is earning its price. Smaller engineering groups and organizations without active governance requirements sometimes sit on Enterprise by default when Team would cover their real usage at a materially lower per seat cost.
The correct position cleans the membership, matches Copilot to usage, confirms the tier, and lands the corrected seat count at renewal when the term and price are set. Doing the work after renewal leaves the saving stranded until the next cycle.
We reconcile the full member and collaborator list against sign in, contribution, and Copilot usage telemetry. Inactive members and lingering external collaborators are flagged for reclaim, Copilot is rescoped to the developers who actually use it, and the plan tier is tested against real feature consumption. The output is a clear, evidenced target seat count and Copilot count that reflects the active engineering team rather than everyone who was ever granted access, which on a large organization is consistently a major recurring saving.
Seat counts and pricing are set at renewal, so the cleanup pays back fully only when the corrected numbers go into the renewed term rather than after it. We sequence the optimization to complete before the renewal date, so the negotiated seat commitment reflects the true active population and the Copilot count reflects real usage. We also set the joiner mover leaver process for GitHub so reclaimed seats stay reclaimed and the count does not drift back up over the term, turning a one time clean up into a position that holds through the next year.
The three sources of billable drift, the Copilot usage test, the Enterprise versus Team tier check, and the renewal timing that locks the corrected count into the term. Sent on request.
We reconcile membership against activity, reclaim the inactive and orphaned seats, rescope Copilot to real usage, confirm the Enterprise tier earns its premium, and time it all to land in the renewed term. The result is a GitHub commitment sized to the engineers who actually contribute.