Gmail decides placement from three inputs that are scored separately: whether it can verify you sent the message, what your sending history has earned, and how Gmail users behave when your mail arrives. Passing the first one is a prerequisite. It is not a result.
That separation is the part senders get wrong most often. I regularly look at domains where every authentication check is green, the sender is certain the setup is correct, and mail is still filtered, because authentication was never the thing being scored. It was the thing being checked before scoring began.
Gmail verifies your identity through SPF, DKIM and DMARC, then scores your domain and IP on sending history, then weighs how recipients actually treat the message. Authentication is pass or fail. Reputation and engagement are continuous scores built over time, and they are what decide placement once you have passed.
Google states the entry requirements directly in its sender guidelines. Senders delivering more than 5,000 messages a day to personal Gmail accounts must authenticate with SPF and DKIM, publish a DMARC policy, align the From header domain with either the SPF domain or the DKIM domain, offer one-click unsubscribe on subscribed mail, and keep the spam rate reported in Postmaster Tools below 0.3%.
Every one of those is a floor. Meeting all of them gets your mail considered on its merits, and nothing more. The scoring that follows is where placement is actually won or lost.
Google's published threshold is 0.3%, measured in Postmaster Tools, and its guidance is to stay below that number rather than to treat it as a target. Reaching it is described as a level at which filtering can be expected. In practice the number behaves as a ceiling you should never approach, not a line you sit under.
There is a mechanical detail in how that rate is computed that catches people out. Spam rate is user-reported spam measured against messages that reached the inbox, not against everything you sent. As more of your mail gets filtered to spam, the denominator shrinks. Your reported rate can rise while the absolute number of complaints falls, because the recipients still receiving you in the inbox are a smaller and increasingly engaged group being measured against the same complaints.
That is why a spam rate climbing during a placement problem is not automatically evidence of a worsening list. Sometimes it is the arithmetic of the problem you already have.
Postmaster Tools reports domain reputation and IP reputation in four bands: bad, low, medium and high. The bands are computed over a rolling window rather than from your last send, which is why a single clean campaign does not lift a damaged domain, and why one bad send keeps costing you after it is over.
Across 50 domains I monitor in Google Postmaster Tools, the reported reputation band changed 13 times in 703 consecutive readings over a seven-week window in 2026. A band is a slow-moving state. One day's reading tells you very little, and two weeks of them tell you almost everything.
The practical consequence is that reputation is worth checking early even though you cannot fix it quickly. Knowing on day one that you are looking at a multi-week recovery changes what you do in week one.
Because each provider scores you on its own recipients. Your Gmail reputation is built from how Gmail users treat your mail, and Microsoft runs separate signals with separate thresholds. Divergence between providers is normal and it is diagnostically useful, since it usually points at one audience rather than at the message.
When a sender is clean at Microsoft and filtered at Gmail, I stop looking at the template. The message is identical in both places. What differs is the population receiving it, so the cause is in how that specific population was acquired and how it has behaved since.
Not per message. Gmail shows the recipient a short banner naming a general category, and it does not expose a per-message score to the sender. Postmaster Tools reports aggregate signals for your domain instead: reputation bands, spam rate, authentication pass rates, encryption and delivery errors.
This is the limit that makes generic advice unreliable. Nobody can tell you why one message was filtered. What you can do is read the aggregate signals in an order that rules causes out, which is the method I use across every case and have written up in why most deliverability fixes do not work.
Postmaster reports symptoms. It shows a reputation band without telling you which send moved it, a spam rate without telling you which segment produced it, and authentication pass rates that can read as healthy while alignment is quietly failing on the subset of mail that matters.
Reading it usefully means bringing your own send calendar to it, then lining up each movement against what you actually sent that week. The graph on its own is a record that something happened. The calendar next to it is what turns that into a cause. If you want the six specific causes underneath Gmail placement problems, they are in the 6 most common causes of Gmail spam placement.
No. Reputation and spam rate only populate once a domain sends enough volume to Gmail for Google to report it without exposing individual users. Low-volume senders often see a verified domain with empty graphs, which means insufficient data rather than a clean record. It is a common and expensive misreading.
Only if a shared pool you do not control is the cause. A dedicated IP starts with no sending history, so it builds reputation from zero, and at low volume it can perform worse than a well-run shared pool. Gmail also weighs domain reputation heavily, and changing IP does not reset that.
Weeks of consistent clean sending, not days, because the bands are computed on a rolling window. The recovery is driven by the ratio of good sending to bad inside that window, so continuing to mail disengaged segments during recovery extends it. Reducing volume to engaged recipients only is what shortens it.
No. Gmail scores aggregate behaviour across recipients, and actions taken in your own mailbox carry no weight for your domain. The signals that move reputation are the ones produced by the people you send to, which is why buying opens or seeding engagement from accounts you control does not survive contact with the filter.