They use different data, attribution models, windows, identity signals and definitions. Disagreement is normal. The job is to understand what each system is measuring and create a consistent decision framework rather than trying to force every platform to report the same number.
