Symptoms, recommendations, and actions combine to give alerts. In the following figure, we can see where they all fit when it comes to alerts:

Symptoms are made up of a single condition. Conditions can be based on metric conditions, such as when the CPU usage is greater than 90 percent, or property conditions, such as when the DRS/HA is fully automated.
Symptoms can be used just to highlight an issue that directly affects the score of the associated badge, such as Health. Alternately, a symptom can be a part of a greater problem, which in this case is an alert.
Any symptom that is triggered will appear as an issue within the vROps UI under the related badge and object.