I find I often have an integration test that is failing, and then I start adding unit tests until I find the error. Especially when the failure is not amenable

I find I often have an integration test that is failing, and then I start adding unit tests until I find the error. Especially when the failure is not amenable to investigation by a debugger.