Data cleansing
Data cleansing means correcting the values already in a product file, such as mixed units, misspellings, placeholders and duplicate rows.
- Data cleaning
- Data scrubbing
What it means
Data cleansing fixes what is already there. In a supplier file that means a colour spelt three ways, mixed centimetres and millimetres, a cell reading N/A, a damaged barcode or a product listed twice.
It pairs with product data enrichment, which adds what is missing, and on a supplier file it comes first. A width written as 59.5cm in one row and 595mm in the next is one width written twice. Only once the two agree can you see that a third row with no width is the real gap.
Most of the work is normalisation: folding every spelling onto one canonical value, putting each measure in one unit, splitting compound values and emptying placeholder values. Keep the supplier’s original file untouched, so every change can be traced.
Think of it like this
Sorting a pantry before writing the shopping list is the same job. Until the three jars of one spice, each labelled differently, sit side by side, you cannot tell what you have run out of.
For example
A Finish column holds Steel/Inox, s/steel and SS for the same finish, and an Overflow column says N/A. After cleansing, all three finishes read Stainless steel. The N/A becomes an empty cell, a gap that can be counted and filled rather than a filter option.
Why it matters
Not to be confused with
Data cleansing is not data validation. Validation refuses or flags a value that breaks a rule but repairs nothing, so a wrong value of the right type still passes.
Sources
- Google Merchant Center Help. Colour [color].