Multi-domain sending: a survivor's guide for 2026
Inbox providers got smarter this year. Here is what is still working, what stopped working in spring, and what we expect by summer.
Deliverability is the part of outbound that everyone outsources to a vendor and quietly hopes never to think about. That worked for about a decade. It has gotten harder in the past eighteen months. The providers got smarter. The bar for landing in primary moved. The price for a single misstep got steeper. If you are running cold across three or more domains, the rules of the game have changed enough that the playbook from 2023 will get you spam-foldered by Easter.
What changed
Two things. First, the major inbox providers now weight engagement signals much more heavily than they did a year ago. Opens are noise after Mail Privacy Protection, so they have moved to dwell time, scroll, link clicks, and reply weight. Second, they got much better at clustering domains by ownership. Sending from rep-a@finch.co and rep-a@finchwork.co with the same content is no longer a clever trick. It is a clustered violation.
What still works
- Truly distinct domains, registered in different windows, with different DNS providers, with believable brand stories on the landing page
- Warmup periods of three full weeks per mailbox before any real volume, with engagement that actually looks human
- Send volumes under 40 per mailbox per day, with stop-the-bleed logic if bounce rate crosses 3 percent in a 24 hour window
- Content that varies enough across recipients that the providers cannot fingerprint a template
What stopped working
Spintax. Aggressive tokenization. Generic merge fields. Any sequence where 80 percent of the email is identical across recipients. The clustering models are good enough now to detect that pattern across hundreds of mailboxes. The penalty is not subtle. We watched one customer go from 11 percent reply rate to 1.4 percent in nine days because they pushed a tokenized variant across six domains at once.
If your tool calls something a 'variant' but the underlying body is 90 percent the same, you are running spintax with extra steps. The providers know.
What we expect by summer
We expect Microsoft to push the same engagement-weighting Google rolled out in Q3, which will move the cutoff for primary placement another few points higher. We expect a wave of consolidation among the warmup vendors, since the cheap pools that recycled the same 800 mailboxes are getting flagged. And we expect that AI-written email, on its own, will not move the needle. It is the floor now, not the ceiling.
How we built around it
In Blacksmith we rotate sends across distinct identities, throttle per mailbox, and use the signal context (permit, expansion, leadership change) to actually generate different bodies, not template variants. The body is genuinely different per account because the underlying reason for the email is different per account. That is the trick. Specificity is the deliverability strategy, not just the sales strategy.
“Deliverability used to be a vendor decision. It is now a content decision.”
If this resonated, it'll feel familiar in the product.
Try Blacksmith against your real territory for 14 days. No card, no metered AI credits, no surprises.
Summit seats are limited and allocated to qualifying founding members. Perks subject to final terms.