Email Deliverability Monitoring: The Four Signals Worth Alerting On

Email deliverability monitoring fails in a specific way: not from too little data, but from too much. The average dashboard shows forty metrics with no hierarchy, so the person meant to watch it either checks everything and burns out inside a month, or checks nothing and finds out placement dropped when replies stopped. A dashboard nobody acts on is worse than no dashboard, because it feels like coverage while providing none.

Real monitoring is narrower and louder. It watches a small set of signals that reliably predict a placement problem, sets a threshold on each, and fires an alert to a named person the moment one crosses. Everything else is a metric you can review weekly without a pager. There are four signals worth waking someone for, and the rest of this piece is those four: what they are, where the line sits, and the first move when the line breaks.

What Monitoring Is, and What It Is Not

An audit is a snapshot: it tells you the sending setup is sound before you launch. We covered that full pre-send pass in the deliverability audit. Monitoring is the moving picture that runs after. Its only job is to tell you, fast, when a known-good setup starts to degrade, so you act while the fix is still small.

That framing matters because it decides what deserves an alert. A signal earns a page only if it predicts a placement drop and if you can do something about it when it fires. A number that is merely interesting, or that you cannot influence, belongs on a weekly review, not on an alert. Hold every candidate metric to that pair of tests and the list shrinks to four.

Signal 1: Domain Reputation

Google Postmaster Tools reports domain reputation as High, Medium, Low, or Bad. This is the closest thing to a direct read on how Gmail views you, and Gmail is where most B2B recipients live, so it leads the list.

Alert when reputation drops a level, not only when it hits the floor. A slide from High to Medium is the early warning, and it is the moment to look, before Medium becomes Low and Low starts routing you to spam. The first response is diagnostic, not cosmetic: check whether volume jumped, whether a new list loaded, or whether complaint rate moved in the same window. Reputation is a lagging summary of the other three signals, so a drop here usually means one of them already broke.

Signal 2: Spam Complaint Rate

This is the hardest line on the board. When a recipient marks your message as spam, that feedback goes straight to the mailbox provider, and it is the single strongest negative signal you can generate. Gmail's published guidance treats 0.3 percent as the ceiling and asks senders to stay below 0.1 percent.

Set the alert well under the ceiling. If complaint rate crosses 0.1 percent, look now; if it approaches 0.3 percent, stop sending on the affected domains and diagnose before you send again. Complaint rate is set by who you send to and what you say, not by infrastructure, so a spike here points at targeting or copy rather than DNS. That is also why an infrastructure-only fix never resolves it. The lever is the list and the message.

Signal 3: Authentication Pass Rate

Postmaster reports the percentage of your mail that passes SPF, DKIM, and DMARC. In a healthy setup this sits at or near 100 percent, so it is one of the cleanest alerts you can set: any sustained drop means something concrete broke.

Alert on any fall below roughly 99 percent that holds across a day. The usual causes are a changed DNS record, a new sending source added without matching authentication, or, less often, someone spoofing your domain. Each has a specific fix, which is what makes this signal so useful: unlike a vague reputation dip, an authentication drop points you at a named record to check. The groundwork behind this number lives in our SPF, DKIM, and DMARC setup; monitoring is how you catch the day it silently stops aligning.

Signal 4: Hard Bounce and Block Rate

The fourth signal is the rate of hard bounces and outright blocks, the SMTP 5xx rejections that say a mailbox provider refused the message. A rising bounce rate means dead addresses in the list; a rising block rate means a provider is actively rejecting you, which is a reputation emergency one step short of the spam folder.

Alert when hard bounce rate crosses roughly 3 percent in a campaign, and treat any pattern of provider blocks as an immediate stop. High bounce is usually a list-hygiene failure, and it feeds the other signals, since providers read repeated sends to dead addresses as either carelessness or list-buying. The bounce rate benchmarks show where the healthy band sits and where the alarm should trip.

The Thresholds, in One Place

What Not to Alert On

Cutting the list is half the discipline. Open rate is the clearest example of a metric to demote: since Apple and others began pre-loading images to protect privacy, opens are inflated by machine fetches and no longer track whether a human saw your mail, let alone whether it hit the inbox. Alerting on it means chasing noise. Review it as a soft trend if you like, but never page on it.

The same goes for any single day of placement data on a low-volume domain, where normal variance looks like a crisis, and for raw send counts, which measure activity rather than health. Monitoring earns its keep by being quiet until something real moves. A system that cries wolf on vanity metrics trains its owner to ignore the one alert that mattered.

The opinion we will stand behind: the number of signals you alert on should be small enough to fit in your head. Four is close to the ceiling. Teams that try to alert on twenty end up alerting on zero, because alert fatigue is a real failure mode and a silenced pager protects nobody.

Where This Leaves You

Deliverability monitoring is a filter problem before it is a data problem. The signals that matter have been the same for years: reputation, complaints, authentication, bounces. What changes account to account is whether someone drew a line on each one and agreed to act when it broke. Draw those four lines, hand them to a named owner, and you have monitoring. Skip that and you have a dashboard, which is a different thing wearing the same clothes. For the recurring routine these alerts plug into, the deliverability checklist is built to be run by one owner, week after week.

Want Someone Watching These Four for You?

We run deliverability monitoring as a managed function inside our outbound system: these four signals, named thresholds, and a real person who acts when one moves, not a login we hand you and wish you luck with. Every domain and inbox stays registered in your name.

Book your free pilot →