Scoring model
01How risk is calculated
- Independent symbolic rules fire with weights (lexicon, subject, formatting, links, HTML, structure, phishing cues, compliance).
- Density and combo multipliers raise stacked risk (e.g. spam phrases + ALL CAPS + many tracking links).
- Limited ham credits reduce raw points — hard-capped so “clean” tokens cannot erase a critical stack.
- Raw points map to a 0–100 risk index and an SA-like 0–10 curve (steep for marketers who want hard gates).
Risk pillars
02What gets inspected
- Lexicon — spam / hard-sell phrases
- Subject — length, punctuation, deceptive patterns
- Formatting — caps density, exclamation spam, symbol noise
- Links — count, domain diversity, shortener / tracking hygiene
- HTML — image-heavy ratio, markup risk patterns
- Structure · phishing cues · compliance — missing unsubscribe / legal patterns where relevant
What this does not mean
03Explicit limitations
- Does not check SPF, DKIM, DMARC, blacklists, or IP/domain reputation.
- Does not predict Gmail / Outlook placement for your domain.
- Does not send test messages or query live filters.
- A low risk score means content rules look cleaner — not that the send will land in Primary.