A spam score is a number that estimates how junk-like a message looks, based on a set of rules that examine its headers, content, and links. Filters like SpamAssassin add points for each suspicious trait and subtract a few for healthy ones; cross a threshold and the message gets flagged. Lowering your score means removing the traits that add points: authentication failures, spammy wording, broken HTML, image-heavy layouts, risky links, and a missing unsubscribe.

The score is a proxy, not the final word. Mailbox providers run their own filters and weigh reputation heavily. But a high spam score is a reliable warning that your message has avoidable problems, so it is a good place to start.

## How is a spam score calculated?

Rule-based filters run your message through a long list of checks. Each rule that matches contributes points, positive for risky traits and negative for trustworthy ones. The total decides placement against a threshold. SpamAssassin is the best-known example, and tools that emulate it (such as mail-tester) let you preview a score before you send to real recipients.

The exact rules vary, but they cluster into a few categories worth understanding.

## What raises your spam score?

- **Authentication failures.** Missing or misaligned SPF, DKIM, or DMARC is one of the largest contributors. If receivers cannot verify you, your score climbs fast. Fix this first using [the SPF, DKIM, and DMARC guide](/docs/guides/email/dkim-spf-dmarc).
- **Spammy words and phrasing.** Money language, urgency, and shouting (all caps, rows of exclamation points) match classic spam rules.
- **Bad HTML-to-text ratio.** A message that is almost entirely one big image with little or no real text trips the rule designed to catch content hidden inside graphics.
- **Broken or messy HTML.** Unclosed tags, invalid markup, and missing plain-text alternatives all add points.
- **Link shorteners and mismatched links.** Shortened URLs hide the real destination, and visible text that points somewhere else looks like phishing.
- **No unsubscribe.** Commercial mail without a clear opt-out reads as non-compliant and pushes the score up.

## How do you lower it?

Each cause has a direct fix:

- **Pass authentication.** Get SPF, DKIM, and DMARC all passing and aligned with your From domain. This single step usually removes the biggest chunk of points.
- **Rewrite risky content.** Drop the urgency and money words, use sentence case in subject lines, and write the way you would to a colleague.
- **Balance text and images.** Include real text alongside images and never ship a single-image email. Add meaningful alt text.
- **Send valid HTML with a plain-text part.** Generate a multipart message so every client has clean content to render, and validate your markup.
- **Use full, honest links.** Skip shorteners, link to your own domain, and make sure the visible text matches the URL.
- **Add a clear unsubscribe.** Include a visible link and support one-click list-unsubscribe headers.

## How should you test it?

Send a copy to a seed inbox and run it through a scoring tool before a real campaign. SpamAssassin gives you the raw rule breakdown, and a service like mail-tester wraps it in a readable report that names each triggered rule. Treat the output as a checklist: every line item with a positive score is something you can remove. Re-test after each change so you can see what moved the number.

Keep in mind that a perfect score does not guarantee the inbox. Reputation and engagement still decide a lot, which is why this works best alongside the broader [deliverability best practices](/blog/email-deliverability-best-practices). If a campaign is already landing in spam, the companion post on [why emails go to spam](/blog/why-are-my-emails-going-to-spam) walks through diagnosis.

## FAQ

### What spam score is considered safe?

With SpamAssassin-style tools, lower is better and most setups flag messages above a small positive threshold. Aim to get as close to zero added points as you reasonably can, then focus on reputation and engagement, which the score does not capture.

### Do certain words guarantee the spam folder?

No single word will send you to spam on its own. Rules add small amounts of points, and it is the accumulation that matters. A clean, authenticated message can include a word like "free" without trouble; the same word in an all-caps, image-only email from an unauthenticated domain is a different story.

### Why does my score look fine but mail still goes to spam?

Spam scores measure the message, not your sending history. If your domain or IP has a poor reputation, low engagement, or recent complaints, providers can filter you regardless of a clean score. Fix reputation and list hygiene alongside content.

A low spam score is one input to good delivery, not the whole game. Bird's [deliverability tooling](/products/email/deliverability) covers the reputation and engagement signals that a scoring tool cannot see.