Email sending domain warm up: changing domains without wrecking the list
Reputation is attached to the domain, not to you. Move it and you start again, so the first question is whether you have to.
A sending domain carries years of history, and that history belongs to the domain, not to your brand. Swap it and none of it comes with you. Most of the warm-up jobs we get asked about were caused by a decision nobody had to make.
The cheapest warm up is the one you never have to do
Mailbox providers score the domain that signs the mail. That score gets built over months of people opening, ignoring, deleting and complaining, and it stays with the domain when you leave it behind.
The two reasons brands give for moving are a rebrand and a new agency wanting its own subdomain. Neither is a reason on its own. The second one is a flag: an agency that insists on moving your sending domain in week one is optimising for its own convenience.
The four reasons that do justify the risk
- The shop's own domain has genuinely changed, so the from address no longer matches anything the customer recognises
- You're sending on a shared platform domain and have never had one of your own
- Marketing and transactional mail share a domain, and one is dragging the other down
- The current domain is burnt: sustained complaints, blocks at one of the big providers, or a spell of bought traffic
Everything else is cosmetic, and cosmetic is expensive here. A move costs you weeks of reduced volume, plus a spell where you can't tell whether a bad campaign was bad or simply undelivered.
What has to be true in DNS on the day you start
Warming an unauthenticated domain is warming nothing. Klaviyo generates the records, somebody with access to the DNS has to publish them, and that person is usually not in the meeting where the decision got made.
- The dedicated sending domain verified in the platform, and the records live and not pending
- DKIM signing on the new domain, checked against a real send and not a preview
- SPF and DMARC aligned with the new domain, not the old one
- A from address on the new domain that a customer would recognise on sight
- The new domain registered in Google Postmaster Tools before the first send, because it cannot show you yesterday
Do all five before anything goes out. A warm-up that starts on a half-published record teaches the mailbox providers exactly the wrong thing, and you can't take a fortnight of that back.
Small, engaged, and slower than you want to go
One change at a time. Volume climbs or the audience widens, never both in the same week, so when something moves you know which decision moved it.
| Stage | Who you send to | What you are watching |
|---|---|---|
| First sends | People who opened or clicked in the last month | Bounces and complaints, and nothing else |
| Second | Add the last quarter's openers | Whether bounces stay flat while volume climbs |
| Third | Add the rest of the engaged segment | Domain reputation in Postmaster Tools |
| Full volume | The list you would normally mail | The same three things, for another fortnight |
Resist the urge to catch up. The commonest failure we see is a brand that runs three careful stages, then mails the whole file in week four because a promotion landed. That spends the reputation the first three weeks bought.
What to watch, and what to ignore
Bounce rate and complaint rate are the two that matter, because they're the two the mailbox providers are looking at as well. Watch them per send, not as a monthly average. An average hides the day it went wrong.
We won't pretend an open rate is a hard number. It's soft now and we wouldn't run a programme on it. During a warm-up it's still useful in one narrow way: on the same list, sent the same week, a collapse says the mail isn't arriving, not that the writing got worse.
- Ignore a single bad send and act on two in a row
- Ignore the total opens and read the trend on the segment you're warming
- Don't add a discount to force engagement, because you're teaching the list a habit you can't afford
- Don't send to the unengaged to make the volume look normal
Two mistakes that get made at the end
The first is switching the old domain off the week the new one goes live. Leave it authenticated and quiet. A domain that stops signing mail while its records vanish looks abandoned, in exactly the way spam infrastructure looks abandoned.
The second is the calendar. Nobody should warm a domain into Black Friday, a product launch or the run-up to Christmas, and every year somebody does. Finish a clear month before the peak, or leave it until January and send the peak from the domain that already works.
Theo Tziapouras
Founder and strategy at Engage Commerce, the ecommerce agency for 7 and 8 figure DTC brands.
FAQs
How long does it take to warm a new sending domain?
Two to four weeks on a healthy list, and longer if bounces or complaints move while you're doing it. The variable isn't the calendar, it's how quickly your engaged segment gives the mailbox providers something good to look at. Rush the last stage and you undo the first three.
Do I need a dedicated sending domain at all?
If you send marketing email regularly, yes. Sending on a platform's shared domain pools your reputation with strangers, and you can't fix anything they do. Your own subdomain costs nothing and gives you something you can actually repair when it goes wrong.
Should I use a subdomain or the root domain?
A subdomain, for marketing mail. It keeps your bulk sending away from the address your staff email customers from, so a bad month for the newsletter doesn't follow your invoices and order confirmations into the spam folder. Keep transactional mail on its own subdomain too.
Can I warm a domain while still sending from the old one?
Yes, and it's the safer way to do it. Run the new domain on the engaged segment while everything else carries on as normal, then move the rest across in stages. Splitting a single campaign across two domains is the one version to avoid, because neither result tells you anything.
What happens if I skip the warm up?
Sometimes nothing, on a small list with genuinely engaged people. On a list of any size, the first full-volume send from an unknown domain is exactly what a spam run looks like, and the filters treat it that way. Recovering from that takes longer than the warm-up would have.
Slow is the whole method
Deliverability gets earned by being boring for long enough, and a warm-up is that idea squeezed into a month. Send less than you want to, to people who want it, and let the volume climb on evidence, not on the calendar. If the domain is already in trouble, the Rebel Aromas rebuild is what the fix looked like on a real account, and it started here rather than with a campaign.
Deliverability and UK consent
Getting into the inbox, and staying inside UK GDPR and PECR while you do it.
Your email open rate dropped suddenly: what to check, and in what order
Opens are a soft number now. A collapse still means something, and it is almost never the subject line.
Read it
The WhatsApp 24 hour window is the real reason your broadcasts keep failing
Free-form messages live inside a window the customer opens. Everything outside it is a template Meta has already judged, and keeps judging.
Read it
Your email list doesn't carry SMS marketing consent, and assuming it does will get you fined
A phone number on file is not permission to text it. SMS needs its own recorded yes, and the smaller list is the point.
Read itWould you rather this was just handled?
Bring your Klaviyo account and the thing annoying you most. We will tell you what we would fix first, on the call, before you spend anything.
Book a call with our founder.
We're all about relationships built on trust, mutual respect and a shared vision for success. If that sounds like your vibe, let's make some waves together 🌊
