Copilot sells in three paid tiers and bills per user per month. Read the Business and Enterprise split, the real cost of idle seats, and the levers before you renew.
GitHub Copilot is billed per provisioned seat, yet it only pays per active seat, and that gap is where most enterprise overspend hides.
GitHub Copilot sells in three paid tiers: Copilot Business, Copilot Enterprise, and the seat add ons inside larger GitHub plans. Business and Enterprise are both billed per active user per month. The published list rates sit on the GitHub Copilot plans page.
Business is the floor for organizations. Enterprise layers on knowledge bases, pull request summaries, and a Copilot that can reason over your indexed repositories. The gap between them is large, so the tier choice is the first lever.
Business gives the IDE completion, chat, and organization policy controls. Enterprise adds repository indexing and the GitHub.com Copilot experience. The Copilot subscription docs list the exact feature split.
GitHub Copilot paid tiers, list rate and fit (2026)
| Tier | List per user per month | Best fit |
|---|---|---|
| Copilot Business | $19 | Most engineering orgs needing policy controls |
| Copilot Enterprise | $39 | Teams that will measurably use repo knowledge bases |
| Add on seat | Billed in GitHub plan | Orgs already on Enterprise Cloud |
The headline rate is per seat, but the real number is per active seat. A 2,000 seat Enterprise order at list runs $936,000 a year. If a third of seats are idle, you are paying roughly $300,000 for no output.
GitHub bills Business and Enterprise through the same per user model described in its Copilot billing documentation. There is no usage meter on the standard seat, so an unused seat costs the same as a heavy one.
Pull the seat assignment and activity data from the organization settings every month. A seat with zero accepted suggestions in 30 days is a candidate for reclaim. Track acceptance rate, active days, and chat use, not just logins.
Buy Business first and earn Enterprise. Most policy controls a security team wants live in Business. Pilot Enterprise on the teams that will use knowledge bases, prove the lift, then expand. Tie price to measured active seats, not provisioned seats.
White Paper ยท Microsoft
The Microsoft Renewal Playbook
How Copilot and M365 inflate an EA, and how to renew on your usage. Read it free.
The standard pitch is that Copilot Enterprise pays for itself because the repository knowledge base lifts every developer. We disagree. In roughly 7 of 10 Enterprise rollouts we reviewed, the knowledge base was enabled estate wide with no measurement, and acceptance rates were statistically indistinguishable from Business seats. The 73 percent premium bought a feature almost nobody used. The buyer side move is to license Business as the default, run Enterprise as a measured pilot on two or three teams, and only expand the premium tier where the activity data proves a real lift over Business.
Source: Redress Compliance advisory engagement file, 2024 to 2025.
Copilot is priced per provisioned seat, but it only pays per active seat. The whole negotiation lives in that gap.
Set content exclusions before the first seat goes live. Decide which repositories Copilot may read, log the policy, and review it quarterly. Governance is cheaper to install at the start than to retrofit after a security review flags it.
Most teams see a measurable acceptance rate within the first two weeks. If a cohort shows near zero acceptance after a month, the problem is fit or training, not the tool. Pilot, measure, then expand.
No. Start with the engineers writing code daily, not occasional contributors. The GitHub Copilot product page lists supported editors, so confirm your stack before assigning broad seats.
Bring the active seat report and a benchmark to the table. Re price against measured active users, drop idle seats, and stage any tier increase to proven adoption rather than a vendor forecast.
GitHub Copilot Business lists at $19 per user per month and Copilot Enterprise at $39 per user per month. Both bill per assigned seat with no usage meter, so the published rate is the floor, not an average.
Enterprise adds repository knowledge bases and pull request summaries on top of everything in Business. Business already includes IDE completions, chat, content exclusions, policy controls, and audit logs that most security teams require.
Copilot charges by assigned seat, not by usage. An idle seat costs the same as a heavy user, which is why idle seat reclaim is the largest single saving on most enterprise orders.
Export the seat assignment and activity report from organization settings each month. Track accepted suggestions, active days, and acceptance rate, then flag any seat with zero accepted suggestions in 30 days for reclaim.
Only where teams measurably use the knowledge base. The Enterprise tier is a 73 percent premium over Business, and in most rollouts the repository indexing produced no measurable acceptance lift over Business seats.
Yes. Seats are managed in organization settings and can be unassigned at the billing cycle. Pulling idle seats is the fastest way to align spend with the developers actually writing code with Copilot.
Default to Business and earn Enterprise. Pilot Enterprise on two or three teams, prove the lift on activity data, and expand the premium tier only where it beats Business on acceptance rate.
Reprice against active seats rather than provisioned seats. Ask for quarterly active seat reconciliation, ramped pricing as adoption climbs, and an annual term so you can drop the tier if usage stalls.
Business and Enterprise tier math, the idle seat problem, active seat truing, and the renewal levers that cut an over provisioned Copilot estate.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement leaders running the next renewal cycle.