When a client comes to me because their emails are landing in spam, I'm looking for one of six things: broken authentication, a damaged sender reputation, weak engagement, a content or link problem, a shared infrastructure issue, or bad list quality. Nearly every case I've worked fits into one of these six.
None of them announce themselves. A domain with a reputation problem and a domain with a content problem look identical in the inbox, both just end up in spam, which is exactly why I don't start by rewriting subject lines or re-checking SPF for the third time. I rule these out in order. Here's each one, what causes it, and how I check for it. If you want the broader method behind how I decide what order to check things in, I've written that up separately in why your emails are going to spam, and why most fixes don't work.
SPF, DKIM, and DMARC prove to Gmail that a message actually came from who it says it came from. If any of the three is missing, misaligned, or set up on the wrong domain, Gmail treats the message as unverified, and unverified mail has a much lower ceiling on inbox placement no matter what else is right.
The version I see most often isn't a missing SPF record. It's a domain where SPF and DKIM both pass, but the From address doesn't match the domain those records are published on, so DMARC alignment fails quietly. A basic checker shows everything green, and it still fails the check that actually matters.
Gmail scores every sending domain and IP on a rolling basis: spam complaint rate, unsubscribe rate, how often recipients mark mail as spam versus actually engage with it. This score is sticky. One bad week doesn't just cost that week. It suppresses placement for weeks afterward while the rolling average works its way back.
Reputation damage almost always traces to one identifiable event. A send to an old or purchased list, a sudden volume spike from a new campaign, a compromised account sending mail nobody asked for. The mail itself can be perfectly authenticated and still get buried, because Gmail is scoring the sender's history, not that message.
Gmail's spam filter is partly trained on how people actually treat your mail: opens, replies, archiving without reading, dragging a message out of spam. Sustained low engagement teaches that model your recipients don't want this mail, and it does that regardless of complaints or authentication.
This is the one I see missed most, because nothing looks broken. The list is just mailed too often, to people who stopped opening months ago, and Gmail has quietly learned to route around them.
Gmail also scans content and structure: link reputation (where your links point, what domain hosts them), image to text ratio, URL shorteners, phrasing patterns tied to mass unsolicited mail. One link to a domain with a bad reputation, even one you don't control, can pull an otherwise clean message into spam.
This is also the cause I see most overhyped in generic advice, since it's the easiest one to write a checklist about. It's real, but it's usually secondary. A message with strong authentication and engagement can survive imperfect content. No amount of avoiding "spam trigger words" fixes a reputation or engagement problem sitting underneath it.
If you send through a shared IP pool or shared sending domain, common on many ESPs' lower tiers, part of your reputation belongs to everyone else on that infrastructure. A neighbor sending to a purchased list can drag down placement for everyone sharing that IP block, and there's no authentication fix for it, because nothing about your setup is actually wrong.
I go into this one in more depth in shared IP pools: when your reputation isn't yours, since it needs its own diagnostic approach.
How your addresses got onto the list matters more than almost anything else here. Purchased lists, scraped addresses, old contacts pulled from a CRM migration all carry a much higher rate of spam traps, invalid addresses, and recipients who never opted in and will report you the moment mail arrives. One send to a contaminated list can undo months of clean sending.
Spam traps are the sharpest version of this. They're addresses maintained specifically to catch senders with poor list hygiene. Hitting even a handful tells Gmail your acquisition practices aren't sound, and that signal gets weighted heavily and remembered for a long time.
All six of these produce the same symptom in your inbox. Mail lands in spam instead of landing where it should. That's the whole reason this is hard to fix on your own: two completely different problems look exactly the same from where you're sitting, so a guess usually lands on the wrong one and burns weeks while the real cause keeps running underneath it.
I check what's cheap and fast first, authentication and reputation scores, before moving to what takes longer to confirm, engagement patterns and list quality. That order is the whole method behind the diagnostic.
Because authentication and reputation are two separate systems. Passing all three proves you are who you say you are. It says nothing about whether recipients actually want your mail. Gmail layers engagement and reputation scoring on top of that, so a fully authenticated sender with weak engagement still ends up in spam.
It depends on the cause. A misconfigured SPF or DMARC record can be fixed in a day, but it takes 24 to 48 hours to propagate and rebuild trust. Reputation damage from spam complaints or a bad list send usually takes 2 to 4 weeks of clean sending to recover, since Gmail scores reputation on a rolling window, not a switch you flip.
Yes, if it generates enough spam complaints or hits enough spam traps. Gmail weighs recent sending behavior heavily, so one send to a stale or purchased list can drop your domain or IP reputation for weeks, even after months of otherwise clean sending.
It shows symptoms, not causes. It reports spam rate, IP and domain reputation, authentication pass rates, and delivery errors, but it won't tell you which one is actually driving your placement problem or how they're interacting. Reading it right means knowing which metric to check first and what a given combination of numbers actually rules out.