Just like the Twitter sentiment, there might be other indicators of user satisfaction that can be gathered automatically. Continuously collecting the number of support calls or emails per minute and detecting a certain spike can be a clear indicator of a user issue. Using machine learning and system integrations this can be harnessed for automated responses or signaling a user to the results.

Adopting practices like this can save minutes or hours detecting production issues. Taking user feedback and making decisions based on that sentiment can go even further. This is called hypothesis-driven development, which is discussed next.