The most important fact for me is to know whether there were any PO failures ... a PO failure which can cause a CRC error is defined as more than 4 PO errors in 1 ECC block AFAIK.
I do not care about PI that much (might be 32 in 1 ECC block or 280 in 8 ECC blocks) ... because PI errors are unlikely to introduce corrupted data.
If you take it to the limit (like ECMA states), you should do 2 scans at 1x ... 1 scan for the so-called PI-Sum 8 and 1 scan for the PO-1.
Doing so, you still have to think about the fact that e.g. Audiodev (with their highly-acclaimed CATS system) uses a specially-calibrated drive for measuring.