How do you investigate whether an intermittent automation failure is caused by the product, the test script, or the network?
