Perhaps use the Application Insights Availability graph instead? Would need another equivalent for non-web apps. Or perhaps need to make it clear that the MTTR stat should be when the application is completely down, instead of just degraded