Monthly churn among Cursor Pro subscribers rose from 4% to 7% after a usage-pool change. How would you find the cause?

Introduction

Cursor Pro costs $20 a month and includes a usage pool billed at model API rates, with Auto mode for lighter usage. Pricing changes in AI tools are sensitive: in July 2025, Cursor's move from request-based limits to usage-based pricing caused surprise charges, public complaints, an apology and refunds. In this scenario, Cursor has changed how the Pro usage pool works, and monthly churn among Pro subscribers has risen from 4% to 7%.

A three-point rise in churn on a plan with millions of subscribers is a serious revenue problem. My job is to find out whether the pricing change caused it, which users are leaving and why, and what to do.

Step 1

Clarifying Questions (3 minutes)

• Looking at the timing, what exactly changed in the usage pool, and when? I'll assume the included pool was rebalanced so premium models consume it faster, while Auto mode stayed generous, and the change went live on the first of the month.

• Considering measurement, is churn measured as cancellations, failed renewals, or downgrades to Hobby? I'll assume it includes all three, so I'll split them.

• Thinking about segments, is the rise across all regions and user types, or concentrated? I don't know yet; that's part of the analysis.

• Considering the market, did any competitor launch or change pricing in the same window? I'll check, since GitHub Copilot, Claude Code, OpenAI Codex and Windsurf all compete for the same developers.