Warming strategy

Start with a small amount of expected mail and increase it only when delivery results support the change. Keep your existing provider available during a migration so urgent mail does not depend on a new domain’s warming allowance.

Plan around the limits

The default schedule is 2 recipients per hour and 10 per day for the first seven days, then 10 per hour and 50 per day until day 14. After that, account limits still apply. An operator override can change the tier or limits; check your account console.

Each domain has its own warming start. See warming tiers for how the relay chooses that date.

First seven days

Choose a small group of recipients who expect your mail. Send within both the hourly and daily limits. The limits apply from the start.

Check delivery results before increasing traffic. A recipient’s request for a login link or password reset takes priority over a bulk campaign. Keep excess traffic on your previous provider or in your application’s queue.

Days seven to fourteen

The default allowance rises to 10 recipients per hour and 50 per day. Increase traffic gradually within those limits. Investigate bounces and complaints before expanding the recipient group.

After day fourteen

The default warming cap ends, but your account ceiling remains. Move more traffic only when your account limits and delivery results allow it. Reaching warmed does not guarantee inbox placement at any provider.

When a submission reaches a limit

Wait until the reported retry time or window reset. Spreading mail across more hours can help with an hourly cap, but it does not raise the daily allowance. Multi-recipient requests count once per recipient, so batching does not avoid a limit.

Keep retries in your application or SMTP client’s normal queue. Do not send the same message through both providers when the first may already have accepted it.

Monitor delivery

Use the account console to check domain status and recent sends. The deliverability endpoint provides 14-day send, bounce, and complaint totals and a daily send series:

GET /member/deliverability?did=did:plc:your-did
Authorization: Bearer atmos_your-api-key

These aggregate totals are different from the 24-hour window used for automatic domain pauses. Poll at a rate suitable for your dashboard, or use webhooks for supported event notifications.

Automatic pauses

The relay checks each domain every 15 minutes over a rolling 24-hour window. With at least 10 messages in that window, it pauses a domain when either:

• At least three messages bounced and the bounce rate is 5% or higher.
• The complaint rate is 0.08% or higher.

These are the defaults wired into the serving Relay release checked on November 10, 2026. They are Comail controls, not mailbox-provider thresholds. The current implementation does not choose different bounce thresholds by account plan.

A paused domain cannot submit new mail through HTTP or SMTP, and its warmup sender stops. The pause applies to that domain; other domains remain subject to their own status and limits.

Respond to a pause

1. Read the pause reason in your account.
2. For bounces, inspect the delivery status codes and remove invalid recipients. A rejection can also indicate authentication or policy problems.
3. For complaints, stop the affected campaign and check consent, content, and message category. Do not classify marketing mail as transactional to avoid controls.
4. Resolve the cause before unpausing the domain through the available account or operator controls. The next evaluation can pause it again if the thresholds still apply.

See bounces and suppression for the related controls. A label such as burst_warming or an operator tier override is separate from a domain pause.

Complete a migration

Verify Comail’s DNS records before moving traffic. Select a recipient group whose volume fits the current limits, then route that group’s messages through Comail once. Increase the group as results and limits allow; a fixed percentage does not work for every sender.

Keep the old provider’s DKIM records while messages signed by it can still be delivered or retried. Follow that provider’s retirement instructions before removing the records. Move the remaining traffic only after you have checked authentication, delivery, bounces, and complaints.

Back to docs