Cold Email Infrastructure: Domains, Inboxes & Warmup

Cold email infrastructure is everything under the message: the domains you send from, the inboxes on them, how volume is distributed, and the order it all gets warmed up in. It matters because mailbox providers score reputation largely per sending domain — one over-eager campaign can push a domain into the spam folder for weeks, and there is no support ticket that un-burns it quickly (Google’s own sender guidelines tie mitigation eligibility to sustained good behavior).

The design goal is therefore not maximum volume — it is containment and headroom: keep outreach reputation isolated from the email your business runs on, and keep enough warmed capacity that losing one domain does not stop the program.

You have two ways to separate outreach from your primary domain, and the right answer depends on how much isolation you want:

Our default recommendation for cold outreach is the secondary domain: complete isolation is worth $12 a year. Buy a believable close variant (get-, try-, a .co), set a redirect to your real site so anyone who types it lands somewhere legitimate, and authenticate it fully — the exact records are in the SPF, DKIM & DMARC guide. Note that Google counts bulk-sender volume across a primary domain and its subdomains together (per its sender FAQ) — one more reason true separation uses a separate registration.

Subdomains still have a place: warm mail streams you want visibly yours (product notifications, newsletters) and the tracking/booking layer. Inside BusinessMCP, one verified subdomain like link.acme.com serves the tracker, booking pages, and widgets.

Size the stack with arithmetic. Decide your target daily volume, divide by a conservative per-domain ceiling, and that is your domain count. We enforce 200 emails a day per sending domain across all campaigns — a deliberately conservative ceiling — so the capacity math looks like this:

On inboxes: practitioners commonly keep individual inboxes to a few dozen sends a day so no single mailbox looks like a machine gun — commonly cited ranges sit around 20–50 a day per inbox, though no provider publishes an official number. Two or three inboxes per domain (a founder, an SDR, a generic-but-human firstname@) comfortably fill a 200-a-day domain ceiling without any one account spiking.

Resist the temptation to over-build. A seed-stage team doing targeted outbound rarely needs more than one or two sending domains — most reply-rate problems are targeting problems, and no amount of infrastructure fixes a bad list. If your plan needs ten domains to hit volume, the plan is usually the problem; our warm outbound playbook is the higher-leverage fix.

Warmup is where infrastructure planning meets the calendar. A new domain needs two to four weeks of ramping before it carries real volume (the concrete schedule lives in the deliverability guide — we will not repeat it here), so the sequencing rule is simple: start every clock as early as possible.

If you run multiple domains, stagger their ramps a week or two apart rather than warming all of them at once — you get earlier capacity from the first domain and a cleaner read on whether your targeting is producing complaints before the full fleet is exposed. Domain age also helps: an outreach domain registered this morning is a weaker starting point than one that has quietly existed (with DNS and a redirect) for a month.

Inbox rotation — spreading a campaign’s sends across several inboxes and domains — is standard practice in dedicated cold-email tools, and the principle is sound: no single sender identity carries the whole volume. What matters more than rotation mechanics is the aggregate cap: three campaigns quietly stacking 3 × 150 sends on one domain is the classic self-inflicted burn.

This is the part BusinessMCP manages for you rather than hands you knobs for: the 200-a-day ceiling is enforced per sending domain across every campaign combined, cold domains get an automatic warmup ramp, autonomous sends only go to addresses verified as valid (catch-all and unknown addresses are held for review), and any hard bounce or complaint suppresses the contact permanently. A sending-health panel tracks each domain’s bounce and complaint rates against green/amber/red bands and alerts you before a domain drifts into the danger zone.

Infrastructure is the cheap part of outbound. Typical list prices as of 2026 — hedged, since providers reprice:

The real cost in outbound is never the stack — it is burning a warmed domain and losing the weeks of ramp time, or worse, splashing reputation damage onto your primary domain. Spend accordingly: conservative ceilings, verified addresses, and targeting quality are the actual investments. Next steps: Google & Yahoo’s sender rules for the thresholds this stack must respect, and the deliverability checker to verify each new domain in seconds.