How to Warm Up a New Domain Email Address Without Landing in Spam

To warm up a domain email address, you send a small and predictable volume of mail at first, then increase it gradually while watching how receivers treat it. Reputation is built per sending domain and per IP address, and it is earned by consistent volume, clean lists, and authentication that passes. The schedule below is the one I use, and it assumes you control DNS for the domain and can read your own logs.

Before any of that matters, three DNS records have to be right, because filters weight authentication heavily when they have no history for a domain. Publish SPF as a TXT record that authorises your sending hosts, publish a DKIM key and sign every message, and publish a DMARC record in monitor mode so you can see what passes. Start with a policy of p=none and a reporting address you actually read:

_dmarc.example.com.  TXT  "v=DMARC1; p=none; rua=mailto:[email protected]; adkim=s; aspf=s"

Then check that the outside world sees the same thing you published. Resolver caches are not your friend during a cutover, so query a public resolver rather than the one in your laptop's config:

dig +short TXT _dmarc.example.com @1.1.1.1
dig +short TXT example.com @1.1.1.1
dig +short TXT selector1._domainkey.example.com @1.1.1.1

You want to see your DMARC policy, your SPF record, and the DKIM public key. If any of them come back empty, fix that before sending a single message. A warmup on an unauthenticated domain just teaches receivers to distrust you faster.

How to warm up a domain email address: the schedule

Treat warmup as a ramp, not a switch. The shape I use is a low starting volume, a steady daily increase, and a hard cap on how much you send per hour. The exact numbers depend on your list and your provider, so treat the figures below as a pattern to scale, not a prescription. What matters is that the curve is smooth and that you never spike.

Days one to three: send only to addresses you know are engaged, meaning people who signed up recently or who have opened your mail before. Keep it to a handful of messages per hour. Days four to ten: roughly double the daily volume every two or three days, still throttled per hour. After that, keep climbing until you reach your normal steady-state volume, then hold it flat. Flat is good. Sudden jumps are what trigger throttling.

If you are sending transactional mail such as password resets and receipts, the recipients are by definition engaged, which makes warmup easier. If you are sending marketing mail, split the two streams onto separate subdomains so a marketing complaint cannot poison the reputation of your transactional mail. billing.example.com and news.example.com are separate reputations.

Set the rate limit in your MTA

Most MTAs let you cap outbound rate directly, which is better than trusting your application to pace itself. In Postfix, for example, you can throttle per destination domain with the smtp_destination_rate_delay parameter in main.cf:

smtp_destination_rate_delay = 2s
smtp_destination_concurrency_limit = 2
smtp_extra_recipient_limit = 10

That combination spaces deliveries to a given destination and limits how many connections you open at once. Reload with postfix reload and confirm the setting took:

postconf -n | grep smtp_destination
smtp_destination_concurrency_limit = 2
smtp_destination_rate_delay = 2s

If you send through an API rather than your own MTA, the same principle applies: use the provider's rate controls and set them conservatively. Your application should still queue and retry, because a failed send during warmup is a message you have to send again later, which breaks the smooth curve.

What to watch while you ramp

Warmup without measurement is just guessing. The signals that matter are bounces, complaints, and deferrals. A hard bounce means the address does not exist, and a high bounce rate early tells receivers your list is dirty. Suppress hard bounces immediately and never retry them. Complaints are the strongest negative signal there is, so make unsubscribing a single click and honour it instantly.

Deferrals are the most useful early warning. When a receiver is unsure about you, it often answers with a temporary rejection such as a 4xx code and a message like "try again later". That is the receiver telling you to slow down. Read the response text your MTA logs and adjust the ramp accordingly. If you see sustained 4xx replies, halve your volume and hold it there for a few days rather than pushing through.

Also watch your own DMARC aggregate reports. They show which IPs and which authentication mechanisms passed, and they will reveal a misconfigured relay before it becomes a deliverability problem. Keep p=none until the reports show consistent alignment across all your sending sources, then move to p=quarantine.

Keep a seed list and check placement

A seed list is a set of mailboxes you control across the major consumer and business mail providers. Send to them as part of every batch and check where the messages land. If seed messages start arriving in the spam folder, stop increasing volume and investigate. The seed list is not a substitute for real engagement data, but it is a fast tripwire.

Keep the content of early messages plain and consistent. A new domain sending image-heavy HTML with shortened links and a strange reply-to address looks like the pattern filters associate with abuse. Use a real reply-to on your own domain, keep links on the same domain where possible, and avoid link shorteners during warmup.

A repeatable checklist

Confirm SPF, DKIM and DMARC resolve correctly from a public resolver, then send a test message and read the full headers to confirm the signature validates. Verify that reverse DNS on your sending IP points to a hostname in your domain. Set the outbound rate limits in your MTA or API client and confirm them with a config dump. Load a seed list and send the first batch only to engaged recipients. Record daily volume, bounce count, complaint count and deferral count in a spreadsheet so the ramp is visible. Increase on the schedule only if the previous step was clean. Hold at steady state once you reach it.

If you are migrating an existing domain rather than starting fresh, do not reset the reputation by switching IPs and domains at the same time. Move one variable at a time, and give each change several days of steady volume before the next.

Start with DNS and authentication, because everything else depends on them. Then set your rate limits before your first send, keep a seed list running from day one, and let the deferral counts tell you when to slow down. Build the ramp into your deployment process so that provisioning a new sending domain automatically provisions the throttle as well, and you will not have to remember the schedule by hand.

Related articles

Subscribe to our newsletter

Get the latest hosting tips, performance insights, and industry news.