The percentage is not the dose
The National Weather Service defines probability of precipitation as the chance that at least 0.01 inch will fall at a point during a stated period. That threshold is meteorologically measurable, but agronomically tiny. A high probability of a trace is not the same thing as a useful irrigation event.
Replace one forecast with three gates
A watering decision improves when you separate occurrence, amount and storage. First: will rain happen? Second: is the likely amount enough to cross the root-zone threshold? Third: can this soil store it, or will it run off? Weather apps usually foreground only the first gate.
Use the value-of-information rule
Do not check soil on every dry day. Check it when the forecast sits near your action threshold—roughly when watering now and waiting both look plausible. A 30-second finger, trowel or moisture check is most valuable precisely when it can reverse the decision.
The counterintuitive move
Watering before forecast rain can be rational when containers are already stressed, the forecast window is late, or only a trace is likely. Waiting can be rational at a lower rain probability when the root zone already holds enough water. Probability alone cannot settle either case.