Data cleansing
Cleansing corrects data you already have; enrichment adds data you do not have.
Cleansing, normalization and enrichment
It gets confused with three neighbouring processes, and the distinction matters because the tools are different:
- Extraction turns a product page into structured fields. It creates a record from a source. See product data extraction for how that works.
- Cleansing fixes what is already there: typos, broken encoding, HTML left in a description, prices stored as text.
- Normalization makes existing values consistent:
Blue,blue, andNAVY BLUEbecome one value;42 EU,EU42, and8 UKland in the same size scale. - Enrichment adds what is not there at all. An empty field stays empty until data arrives from outside the record.
Cleansing and normalization work on your own data. Enrichment requires a source. That is the whole problem in one sentence, and it is the part most enrichment software leaves to you.
Normalizing incoming data
Standardize units, value vocabularies, and category paths so that data from multiple sources reads as one dataset. Normalizing the incoming data is part of collection and should be done before delivery. Reconciling it with the vocabularies already used in your own catalog is a separate job, and it belongs to whoever owns the schema.
Full guideProduct Data Enrichment for E-commerce Catalogs: Where the Missing Data Actually Comes From
Turn any website into a product data table
Every product page is a data source.