Although I have never been an auditor, and have never audited inventory, auditors rely on two things, samples, and statistics. Statistics does a remarkable job of interpreting the whole based on the sample. Statistics can tell you that if you sample 10 inventory items and they match the records, what is the likelihood that the total 100 SKUs also match. They allow a small margin of error. So, essentially, they pick random samples that is sufficient for the total size of the inventory SKUs that gets to a statistical margin of error that is small enough to make them comfortable.
What Steve said is correct. The inventory audits I have been involved in often wound up counting way too many samples.
You have to check both sides of the coin. Some items you count will be to see if they agree with what is reported. You then have to flip that to test if what is reported actually exists.
You will also have to test if the values are correct. For example is the cost of the item equal to the value per unit used to determine total value. And is the math correct. For example 100 units at 25 per unit equals 2500. These latter items can be done before or after the actual physical count.