Inbox providers are getting much better at scoring intent, not just spam words

Filtering moved past keyword lists years ago, but the gap is widening fast. Providers increasingly model *behavior*: do recipients open, reply, and not delete-without-reading; does the sending domain behave consistently; is the mail authenticated and warmed.

What's actually changing

• Content tricks (avoiding "free," obfuscating links) matter less every quarter. Behavior signals dominate.
• Engagement is the dominant lever: a domain whose mail gets replies and is rarely marked spam earns inbox placement; one with silence and complaints loses it — independent of copy quality.
• Authentication (SPF/DKIM/DMARC) and consistent sending patterns are increasingly table-stakes gates, not optimizations.

Why it matters for outreach

• Chasing "spam words" is largely wasted effort now. The leverage is in list quality and relevance (which drive replies) and infrastructure hygiene (which earns trust).
• Auto-stopping on bounce/reply/unsubscribe isn't just courtesy — it's reputation protection, because complaint and bounce rates feed directly into placement.
• Sending less to a better-fit list now reliably beats sending more to a broad one, because the broad send depresses the engagement score that governs every future send.

The takeaway

You don't beat modern filtering by writing around it. You beat it by being mail people actually want — measured by their behavior, enforced by your infrastructure.