Variation names are limited to 255 characters and must be unique within a product
BASE variation names have a 255-character limit, but length is only half of the problem. The same variation name cannot be duplicated inside the same product.
A short pair such as “Red / Red” can therefore be invalid even though neither value is long.
New-product rows have no Product ID, so product name becomes a useful clue
Existing products can be grouped by Product ID. New rows have a blank ID, which meant an ID-only checker could miss duplicate variation names across rows of the same new product.
The checker now falls back to product name when Product ID is blank, comparing rows that appear to belong to the same new product.
A correct variation name does not replace Variation ID checks
Existing variation updates depend on Variation ID. Even perfectly unique names can still create the wrong operation if an ID is removed or mismatched.
When IDs are part of the job, run the BASE Product ID / Variation ID checker first.
Grouping new rows by product name is still a preflight approximation
CSV text alone cannot perfectly distinguish every intentional case, such as deliberately similar product names. Use the result as a way to narrow likely duplicates and compare them with the source data; BASE remains the final authority during import.