Set up the records first and verify they work. An SPF record listing your sender, DKIM signing with a 2048-bit key, and a DMARC record at p=none with a reporting address. Send one message and read its headers to confirm all three pass before the ramp begins.
Order recipients by engagement and work down the list. Week one goes to people who opened something last month, and those opens are the positive signal the whole exercise is meant to generate. A warmup aimed at cold prospects generates the opposite.
A workable schedule doubles weekly: 50 a day in week one, 100 in week two, 200 in week three, continuing to target volume. Spread the daily total across providers in proportion to your list rather than sending to one provider at a time.
Treat the metrics as a brake on the schedule. Bounces above 2% or complaints above 0.1% mean hold at the current volume or step back, and continuing to plan through rising complaints defeats the point of ramping at all.
A subdomain warms faster than a fresh registration, because the organizational domain already has standing. Sending marketing from news.yourdomain.com also keeps a bad campaign away from your password resets.
The doubling schedule, written out to week ten
Write the weekly doubling out in full and you'll learn something. Starting at 50 a day, the weeks run 50, 100, 200, 400, 800, 1,600, 3,200 and 6,400. So at the end of eight weeks you're sending 6,400 messages a day.
If your target is 5,000 a day, that's fine. For a target of 20,000 a day it isn't, because pure doubling from 50 doesn't pass 20,000 until week ten. The four-to-eight-week window assumes a modest target or a faster pace.
You can go faster if your numbers allow it. When bounces stay under 2% and complaints stay under 0.1% for two full weeks running, some senders choose to step up twice a week. Let the metrics grant that, and don't plan it in advance. Size the early weeks to the engaged people you really have, too. If only 3,000 people opened last month, there's no point scheduling 6,400 engaged sends.
Five habits that stall a warmup
Sending once a week is the first. Providers learn from steady daily traffic, and a domain that sends 5,000 on Tuesday and nothing for six days looks erratic. Spread the same volume across weekdays.
Skipping the list check is another. A bounce during a ramp weighs more than the same bounce from an established sender. Verify the segments you'll use in the first month before day one.
Changing the From address partway through resets some of what you've built. Pick the address and the subdomain before you start, and leave them alone.
Mixing streams causes trouble too. If receipts and password resets share the new domain with marketing, a rough marketing week can delay the mail your customers are waiting for.
Watch the link domain as well. If your platform's shared tracking domain appears in every link, part of your reputation rides on strangers. Set up your own tracking subdomain in week one.
Related questions
How long does domain warming take?
Four to eight weeks to reach meaningful volume. A subdomain of an established domain moves faster than a new registration.
What has to be in place before starting?
SPF, DKIM with a 2048-bit key, and a DMARC record. Verify all three pass on a test message before the first day.
Who should receive the first sends?
Your most engaged recipients, in order. The exercise records their behavior, so start with people who open and reply.
Should I use a warmup tool?
They work less well than they did. The reciprocal-reply pattern is detectable and providers have acted against networks built on it.