Cursor’s 18% and GitHub Copilot’s 68% may look like a head-to-head result. They count different groups, though, and neither tells a team lead which tool its own developers actually use. A new guide sets out a roughly 75-minute audit to measure that, and what each active developer costs.
Watch Desk analysis
What happened
Shattered.io’s guide joins a staff roster with seat and activity exports from Cursor and Copilot. Its small Python project reports assigned and active share, utilisation, overlap, idle spend, orphan seats and cost per active developer. It runs locally, uses standard-library Python and requires no paid service, according to the guide.
The method fixes a shared denominator and activity window before looking at the results: use the same headcount, a chosen as-of date and the same 30-day definition of active for both tools. The guide also advises keeping employee-identifying CSVs out of shared repositories, checking with HR or a privacy officer before processing individual activity data, and using a read-only token. Read the audit guide.
Why it matters
A survey can describe respondents; licence counts can describe seats. Neither, on its own, tells a company whether its own developers are using a tool or what that active use costs. The guide offers a practical way to compare those measures without pretending that a vendor’s customer count or a survey percentage answers the renewal question.
There is a catch: the result depends on having suitable exports and a current staff roster, and individual activity data raises privacy questions. The guide recommends sorting those questions before collecting the numbers, not after the spreadsheet has acquired a life of its own.
Our read
This is a useful antidote to market-share charts doing more work than their denominators allow. Teams choosing between tools can compare their own assigned seats, recent activity and costs on consistent terms. But the findings will describe that team, not settle which product is best for everyone.
What to watch
- Whether teams define the active-use window consistently before comparing tools.
- Whether their exports include the activity and cost fields the audit needs.
- Whether privacy rules permit the use of individual activity data for licence management.
Discussion spark: When deciding which coding assistant to renew, should teams prioritise active use and cost per active developer, or weigh capability and developer preference more heavily?
Sources and evidence
- Cursor vs Copilot Seat Audit: 12 Steps, 75 Min – shattered.io (19 November 2026, 01:48 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.