CSV calculations depend on the meaning of the rows as much as the arithmetic. Use the stated headers, one compatible currency and unit basis, and unique identifiers where required. Keep dates and periods aligned. A blank observation is not automatically zero demand, and a zero-actual period makes some percentage errors undefined. Preserve the source extract so unexpected results can be investigated.

Read a classification as a decision aid

ABC describes value concentration; XYZ describes variability under the selected measure and thresholds. Neither alone establishes criticality or a replenishment policy. Supplier-spend concentration describes the entered spend distribution, not the availability of alternative suppliers. Compare assumptions, inspect boundary cases, and distinguish zero-demand groups from low-variability active demand. Export classifications with their thresholds so the same letters mean the same thing in the next review.