Filters no longer block on single words; they score patterns. Guarantees, urgency, money claims, “click here”, and anything that promises income or prizes still carry weight. Ordinary words like free and discount matter only in combination.
How filters actually use words
Gmail and Outlook use statistical models trained on billions of messages marked as spam or not. Words are features, weighted by how often they appear in spam versus wanted mail from similar senders. One word rarely decides anything. Several together, from a domain with little reputation, do.
Categories that still carry weight
| Category | Examples | Weight |
|---|---|---|
| Guarantees and risk removal | guaranteed, risk-free, no obligation, money back | Strong |
| Urgency | act now, limited time, last chance, urgent | Strong |
| Income and prizes | earn money, winner, you have been selected, pre-approved | Strong |
| Click instructions | click here, click below | Strong |
| Hype adjectives | amazing, incredible, revolutionary | Mild |
| Commerce words | free, discount, cheap, offer | Mild alone, strong in clusters |
| Sales jargon | leverage, synergy, touch base, circle back | Mild, but hurts replies |
What is not on the list
Normal business vocabulary. Price, contract, meeting, demo, proposal are all fine in a one-to-one email. The spam checker weights words by the categories above rather than treating every list word as equal.
Paste a draft and see whether it lands in the inbox or in spam, with every risky phrase highlighted.
Questions
Does one spam word send my email to spam?
Almost never. Filters score the whole message plus sender reputation. Several triggers on a new domain is the pattern that fails.
Is the word free always bad?
In the body, once, it is a mild signal. In the subject line it is one of the strongest single triggers.