Inside Our $70M Cold Email Infrastructure Setup

Most people blame their copy when a campaign flatlines.

But the email actually never got read.

It landed in a spam folder nobody opens, or a filter dropped it before delivery.

We have pushed 10 million+ sends through the setup below. Here is every piece of it.

---

The buying list:

Before any of the technical work, you are shopping.

What gets purchased and roughly what it runs:

• Secondary domains, meaning 167 of them for a 10,000/day operation at 3 mailboxes each
• 500 mailboxes, provisioned through ScaledMail at $3 a Google inbox against $8 retail
• A sequencer, meaning EmailBison for anything at real volume
• Verification credits, meaning Million Verifier at roughly $0.0004 a check
• Backup providers, meaning Sending.ac for Azure inboxes and ZapMail alongside

Total lands around $1,500 a month.

Do the CAC maths against your own ACV before you spend any of it.

At a $45,000 average deal, that number is a ROUNDING error.

At a $2,000 deal with a 10% close rate, the model needs different inputs.

---

Buying and configuring the domains:

Your main domain stays out of this ENTIRELY.

Cold volume never touches the asset your existing customers email.

The purchase process:

• Have Claude generate 200 variations on your brand name
• Bulk search them on Porkbun, 100 at a time
• Take whatever is available, then generate more variations for the gaps
• Skip registrars with auto renewal traps

On extensions, .com costs the most of the common options.

.xyz and .info perform fine and cut the bill meaningfully at this domain count.

2 rules on the names themselves.

Keep them close enough to your brand that a curious prospect finds something coherent.

Skip numbers, hyphens, and stuffed keywords, since those patterns are already associated with bulk senders.

Then have every secondary forward to your main site.

A domain sitting on a blank page is its own signal.

---

Provisioning and authentication:

Each mailbox requires SPF, DKIM, and DMARC records set CORRECTLY.

Handling that manually across 167 domains means roughly a week of clicking before a single email exists,

and one misconfigured record silently ruins that domain's placement for the life of the campaign.

ScaledMail covers the provisioning end to end:

• Google and Microsoft secondary inboxes
• Every DNS and authentication record configured
• Monitoring across 75,000+ inboxes on their infrastructure
• Direct upload of the accounts into the sequencer

Leave DMARC on none through the ramp period.

Enforcing strict rules on a domain nobody has sent from yet causes FAILURES you have no reputation to absorb.

Spread the load across providers deliberately.

A single provider setup inherits every outage and policy change that provider has.

We weight roughly 80% Google, 20% Microsoft, adjusted to wherever the buyers actually are.

An Outlook heavy vertical moves that ratio.

Match sender to recipient provider where the volume justifies segmenting, since Google to Google and Microsoft to Microsoft lifts placement measurably.

---

The ramp period:

New mailboxes get flagged if they open at full throttle.

3 to 4 weeks of warming before anything real goes out.

The ramp discipline:

• Begin at a trickle per mailbox and climb weekly
• Stagger the start dates across the batch, since 100 mailboxes activating on the same Monday is a visible pattern
• Keep warmup traffic running permanently underneath live campaigns
• Distribute sends across working hours with irregular gaps

Per mailbox volume stays low forever. The daily total comes from mailbox count.

500 mailboxes at conservative volume is an operation.

50 mailboxes pushed to compensate is a reputation problem with a countdown on it.

The permanent warmup matters because it generates positive engagement continuously, which offsets what cold sending costs you.

---

Data quality as an infrastructure problem:

Every technical decision above gets UNDONE by a dirty list.

The pipeline that runs before any campaign launches:

• Waterfall enrichment, since 1 provider covers 50-60% of a list and stacked providers reach 85%+
• Million Verifier across the full file
• Bounce Ban as a second pass wherever bounce tolerance is tight
• Role based addresses stripped out, meaning info@, sales@, support@ and similar
• Deduplication against your entire contact history

Bounce rate ceiling is 1%.

Anything above that, the campaign pauses and the file goes back through verification.

Catch-alls are the specific trap.

They report as deliverable and bounce anyway, which is why two verifiers run on high-stakes files.

Suppression gets its own discipline:

• Anyone who has replied, in either direction
• Active clients and anyone mid-deal
• Every opt-out request
• Domains already exhausted in previous campaigns

Skip this and the same prospect gets hit by three separate campaigns off three separate domains, and a warm contact turns into a complaint.

---

Campaign configuration:

Settings that hold at volume:

• Open tracking disabled, since the pixel damages placement and produces nothing actionable
• Plain text only
• No unsubscribe link on cold sequences
• Randomised start times, given that every sequencer defaults to 9am
• Monday through Friday
• Auto replies counted in the stats as a placement signal
• Sequence starts maxed, since the first touch to a fresh contact beats every follow-up

Spintax does real work here.

Identical bodies across thousands of recipients is a detectable pattern, so 3 to 5 variants per line rotating through the send keeps it clean.

---

Watching it and replacing what breaks:

Degradation happens quietly, so the checks run weekly at the individual mailbox level.

What gets reviewed:

• Bounce rate by campaign
• Reply rate by mailbox, since silence usually means spam placement
• Blacklist status across every domain
• Seed inbox results from live campaigns across both providers

Any mailbox with placement issues leaves rotation that day.

Any blacklisted domain gets retired instead of rehabilitated.

Reading the failure signals:

• Every campaign dropping simultaneously points at infrastructure, not messaging
• Bounces climbing on a verified file means your data source degraded
• One mailbox producing nothing for a week is almost certainly filtered
• Auto replies vanishing means you stopped reaching real inboxes

Build the replacement cycle in from day one.

Domains burn, mailboxes wear out, and providers change their rules without telling anyone.

A setup with domains already warming, spare mailboxes ready, and 3 providers in rotation absorbs all of that without the send volume ever dropping.

That is the difference between an operation and a campaign.

---

- Christian

If you're running GTM/outbound yourself, we teach all of this in the community.

50+ modules, 3 live calls a week, and direct access to the operators running this at scale.

**Join up @ gtmelites.io**

OR, if you want an in-house GTM team running your lists, infrastructure, sending, replies, and your entire market contacted every 90 days,

**DM me "COLD" or book a call @ cal.link/chris-article**

30,000+ meetings booked and $70M in pipeline across 100+ B2B companies.