Recap
- Watch five signals weekly: bounce rate, spam complaint rate, inbox-versus-spam placement, blocklist status, and domain reputation.
- Each one has a line you do not cross. Bounce above 2 percent and complaints near 0.1 percent are the two that get domains burned.
- The point of monitoring is early warning. By the time replies dry up, the reputation damage is already weeks old.
- When a signal trips, slow the volume first, then find the cause. Pausing buys you time that apologizing never will.
Deliverability monitoring is the weekly habit of checking whether your email is actually reaching inboxes, before your reply rate tells you it is not. Five signals cover most of it: bounce rate, spam complaint rate, inbox placement from seed tests, blocklist status, and domain reputation. Each has a threshold. Cross it and you act. The rest of this is what the numbers mean and what to do when one moves.
What are the five signals worth watching every week?
These five give you the clearest read on sender health with the least noise. Bounce rate and complaint rate are leading indicators that move first. Seed placement shows where mail lands right now. Blocklist status is a binary alarm. Domain reputation is the slow-moving score the mailbox providers keep on you. Track all five and you rarely get surprised.
| Signal | Healthy | Act when | What it tells you |
|---|---|---|---|
| Hard bounce rate | Under 2 percent | Above 2 to 3 percent | List quality and verification |
| Spam complaint rate | Under 0.1 percent | Approaching 0.1 percent | Fit, copy, and consent |
| Inbox placement | 90 percent or higher | Drops below 80 percent | Where mail actually lands |
| Blocklist status | Not listed | Any listing | A provider flagged you |
| Domain reputation | High or medium | Slips to low | Long-run sender trust |
Why do bounce rate and complaint rate matter most?
Because they move fastest and cost the most. A climbing hard bounce rate means you are mailing dead or invalid addresses, and mailbox providers read that as a sender who does not clean a list. Complaints are worse. Every spam click is a person telling Gmail your mail is unwanted. A handful per thousand is enough to tank placement for everyone on the domain.
The numbers are public. Google asks bulk senders to keep complaints under 0.3 percent and to aim for 0.1 percent. Hard bounces above 2 percent is the line most teams treat as a warning. If either climbs, the cause is almost always upstream: a stale list, weak verification, or a segment that never wanted to hear from you.
How do I know if mail is landing in spam, not just being accepted?
Acceptance is not placement. A provider can accept your message and quietly file it under spam, and your sending tool will report it as delivered. The only honest read is a seed test. You send to a set of monitored addresses across Gmail, Outlook, Yahoo, and a few others, then check where each copy landed.
Run a seed test weekly and after any meaningful change: new domain, new copy, a volume jump. If placement at one provider falls off a cliff while the rest hold, you have a provider-specific problem, often authentication or a reputation dip at that one mailbox host. If it drops everywhere at once, look at content and volume.
How do I check blocklists and domain reputation?
Blocklists are public databases of senders that providers distrust. Tools like MXToolbox check your domain and sending IP against dozens of the major ones in a single lookup. For domain reputation, Google Postmaster Tools shows the reputation Gmail assigns your domain, plus the spam rate Gmail actually sees, which is the closest thing to ground truth you can get.
Set both up before you need them. Postmaster Tools needs a verified domain and a little volume before the graphs populate, so it is useless to register it the day after a problem starts. A weekly blocklist check takes two minutes. A drop from high to medium reputation in Postmaster is your earliest warning that something is off, often before bounces or complaints have moved at all.
What do I do the moment a signal trips?
Slow down first, diagnose second. The instinct is to investigate while you keep sending. That is backwards. Volume is what converts a small problem into a burned domain, so cut it the moment a number crosses a line. Then work the cause.
- Bounces spike. Stop the send. Re-verify the list, drop every invalid address, and look at where the bad addresses came from. A single dirty import is the usual culprit.
- Complaints climb. This is a fit and consent problem, not a tuning one. Tighten targeting, cut the worst-performing segment, and reread your copy as a stranger would. Make sure unsubscribing is one click.
- Placement drops at one provider. Check authentication first: SPF, DKIM, and DMARC all aligned. Confirm reputation at that specific mailbox host before touching content.
- Blocklist hit. Find out which list and why. Stop the behavior that triggered it, then file for delisting. Many temporary lists clear on their own once you go quiet.
- Reputation slips to low. Pull volume back hard, mail only your most engaged recipients for a week or two, and let the score recover before you scale again.
This is also why disciplined senders monitor at the per-mailbox level, not just the domain rollup. One bad inbox can drag a shared domain down, and you want to pull that single mailbox before it poisons the rest. An autonomous operator like LaunchSurface watches these signals on every send and throttles itself when one trips, which is the same discipline, just without a human refreshing dashboards.
How does monitoring fit into a weekly rhythm?
Keep it boring and consistent. Once a week, pull the rollup: bounce rate, complaint rate, a fresh seed test, a blocklist scan, and the Postmaster reputation graph. Five minutes if nothing moved. Write down the numbers so you can see trend, because a single week tells you almost nothing and four weeks tells you everything. The goal is not a perfect score. The goal is to catch the slope early, while a pause and a list scrub can still fix it, instead of after replies have already gone silent and the domain needs months to recover.