Two variables moving together does not mean one causes the other. The association may run the other way, be driven by a hidden third variable (a confounder), or be pure coincidence. Establishing causation requires more than a correlation — ideally a controlled experiment or careful control for confounders. This is the statistical root of why Observed success hides its real causes.
Related: Survival is not evidence of a cause