Keywords
How to Clean Keyword Lists
A safe cleaning order that removes the mess without destroying meaning.
4 min read
Clean in a fixed order
Order matters. Strip punctuation first, then numbers if you need to, then collapse whitespace, then lowercase, and only then deduplicate. Deduplicating before lowercasing leaves case-variant duplicates behind.
Be careful with stop words
Removing stop words turns 'how to store bread' into 'store bread'. That is fine for tag extraction and harmful for long-tail research, where the question words carry the intent.
Leave stop-word removal off unless you have a specific reason to use it.
Always check the counts
A cleaning pass that removes forty percent of your list has probably done something you did not intend. Compare before and after counts and investigate large drops.
Keep the original
Cleaning is destructive. Keep the raw export somewhere before you start, so a bad rule choice costs you a minute rather than an afternoon.