What are heuristic rules in spam filtering?

You've probably heard that certain words or phrases can "trigger spam filters." That's heuristic filtering at work. A heuristic rule is a predefined check that looks for a single characteristic commonly associated with spam. On its own, one flag rarely does much. But they stack.

Each matching rule adds points to your spam score. Get enough points and you're in the junk folder. The threshold varies by filter, but the logic is consistent: more red flags, more risk.

Here's what heuristic rules actually check for:

• Text patterns like excessive capitalization ("FREE MONEY NOW"), multiple exclamation marks, or phrases historically linked to scams ("act now", "no obligation", "winner").
• HTML tricks like invisible text, font color set to match the background, or content designed to hide from filters while showing something else to readers.
• Structural signals like a wildly imbalanced image-to-text ratio, no plain-text version, or missing headers that legitimate mail servers always send.
• Formatting habits like using red or green colored font, too many different fonts in one message, or a subject line in all caps.

SpamAssassin is the most well-known open-source filter built on this approach. It runs hundreds of heuristic tests and assigns each a weight based on how reliably that pattern predicts spam. A single weak signal might add 0.1 points. A really bad one adds 3 or 4. Most filters treat anything above 5 as spam.

Heuristics were the foundation of filtering long before machine learning entered the picture. They're still useful today because they catch pattern-based tricks that ML models don't always flag on their own. The two approaches work together in most modern filters.

For legitimate senders, the practical takeaway is simple. Write naturally. Don't use sales-y buzzwords, shouting caps, or gimmicky HTML. If your email looks like something a real person wrote to another real person, most heuristic checks will pass without a problem. The rules exist to catch emails that don't look like that.

If you want to see how your subject line holds up before you hit send, our free subject line tester checks for common patterns that heuristic filters flag. Worth a quick look before a big send ;)