图片出处:https://newsletter.getdx.com/p/how-18-companies-measure-ai-impact-in-engineering
一、指标明细

GitHub
- AI time savings:人工智能节省的时间
- PR throughput:拉取请求吞吐量
- GitHub Copilot CSAT:GitHub Copilot 客户满意度
- Change Failure Rate:变更失败率
- Engaged users:参与用户
- Innovation rate: feature expenses/all engineering expenses:创新率:功能支出/所有工程支出
- Developer experience, e.g. code maintainability, focus time, overall satisfaction:开发者体验,例如代码可维护性、专注时间、整体满意度
- % of code by AI (drafted & submitted) & Acceptance rate:人工智能编写(草拟和提交)的代码占比及接受率
- Weekly active users (WAU):周活跃用户数
- Usage intensity:使用强度
- Change throughput:变更吞吐量
- Active coding time, investigation time, & review time per change:每次变更的活跃编码时间、调查时间和审查时间
- Change cycle time:变更周期时间
- User sentiment and trust in AI:用户对人工智能的看法和信任度
- Perceived productivity:感知生产力
- Developer friction:开发者摩擦
Dropbox
- Daily active users (DAU) / weekly active users (WAU) for AI tools:人工智能工具的日活跃用户数/周活跃用户数
- AI tool CSAT, PR throughput (comparison by AI usage):人工智能工具客户满意度、拉取请求吞吐量(按人工智能使用情况比较)
- Time saved per engineer per week:每位工程师每周节省的时间
- Change fail percentage:变更失败百分比
- AI spend total and per developer:人工智能总支出和每位开发者支出
GLASSDOOR
- DAU/WAU of AI tools:人工智能工具的日活跃用户数/周活跃用户数
- DORA metrics:DORA 指标
- % capacity worked per AI tool (actual AI tool usage as percentage of its maximum potential):每个人工智能工具的工作容量百分比(实际人工智能工具使用量占其最大潜力的百分比)
- % code coverage:代码覆盖率
- Number of bugs:漏洞数量
- Number of A/B tests per month:每月 A/B 测试数量
webflow
- DAU/WA

最低0.47元/天 解锁文章

被折叠的 条评论
为什么被折叠?



