Correlation

Correlation says how much two metrics move up and down together; a number between minus one and plus one showing the direction and strength of that co-movement.

A positive correlation means two metrics rise together, negative means one rises as the other falls, and near zero means no relationship is seen. Its practical value is finding leads: if view depth and return rate correlate strongly, view depth might be a leading indicator of loyalty. But it has two big traps. First and most important: correlation is not causation; two metrics can move together because both are effects of a third factor, not because one produces the other. Second, correlation sees only a linear relationship; two metrics can have a strong but curved relationship and still show a near-zero correlation. So correlation is the start of a question, not a final answer.