How many samples are enough?

There is no intuitive answer to this, and the intuitive answer is almost always on the wrong side. Sample size follows from the risk you are willing to accept — you first answer "how much escaping defect can we live with", and the number falls out of that. Teams that work backwards pick a comfortable number first and justify it afterwards. The bigger issue is deviation from the plan: taking fewer samples than planned is more dangerous than never writing a plan, because the record implies the plan was followed. SenSight QMS keeps sampling records and their deviations together (§7.3).