Why Your Cold Email Lands in Spam (and the 7 Infrastructure Fixes That Got Us to 98.5% Inbox Placement)

You wrote a good email. You sent it to a clean list. And it still landed in spam, or worse, it never showed up anywhere you can see. So you rewrote the subject line, swapped a few words, and nothing changed. That is the loop most founders get stuck in, and it is the wrong loop, because the problem is almost never the words.

Cold email deliverability is an infrastructure problem. The mailbox provider decides where your message goes based on how your sending domain, your IP reputation, and your authentication records behave, long before it reads a sentence. We send cold email at volume: 2.5M+ messages across client campaigns, and we average 98.5% inbox placement versus the roughly 60% that is typical on shared infrastructure. Here is what actually moves that number.

The real reason: shared infrastructure poisons your reputation

Most cold email tools put you on shared or recycled sending infrastructure. You inherit the reputation of every other sender on that pool, including the spammers. When one of them gets flagged, your placement drops with theirs, and you never find out why. That is the mechanism behind the ~60% placement most cold senders quietly accept as normal.

The fix is not a setting. It is owning the infrastructure. We run dedicated sending domains separate from your primary domain, so a deliverability problem never touches your real company email. Across Google, Microsoft, and Azure, we stand up 52 warmed mailboxes per client, spread the volume thin across them, and keep per-mailbox send rates low enough that no single inbox looks like a machine. Reputation you control is reputation you can protect. See how we build the sending infrastructure for the full setup.

Authentication: SPF, DKIM, and DMARC are pass or fail, not optional

If your SPF, DKIM, and DMARC records are missing or misconfigured, Google and Microsoft now treat you as suspicious by default. This is the single most common cause of cold email going to spam, and it is invisible unless you check the raw headers. A record that is 90% right is a record that fails.

Every dedicated domain we set up gets all three configured and verified before a single email goes out, with DMARC moving to an enforcement policy once the domain is warm. We test placement from real mailboxes against live seed accounts rather than trusting a dashboard score. The result is the bounce range we hold across campaigns: 0.15% to 0.9%, well under the threshold where providers start throttling you.

Warmup, volume discipline, and list hygiene

A brand new domain that sends 200 cold emails on day one looks exactly like a spammer to a mailbox provider. New domains need weeks of gradual, organic-looking activity before they carry real volume. We warm all 52 mailboxes before campaigns start, then ramp send volume on a schedule instead of flooding from a cold start.

Volume discipline only works if the list is clean. Decayed Apollo data is a top source of bounces we see in audits: contacts who left the company, role changes, dead inboxes. A high bounce rate is a direct signal to providers that you are not a careful sender, and it tanks placement for everyone on the domain. We verify and waterfall-enrich every contact through Clay so bounces stay in that 0.15% to 0.9% band. Bad data is a deliverability problem before it is a targeting problem.

The 7 fixes, in order

If you want the short version, here is the sequence we run for every client. Each one is an infrastructure change, not a copy tweak.

• Move off shared infrastructure onto dedicated sending domains, separate from your primary domain.
• Configure and verify SPF, DKIM, and DMARC on every sending domain, then enforce DMARC once warm.
• Spread volume across many mailboxes (we run 52) so no single inbox spikes.
• Warm every mailbox for weeks before sending, then ramp volume on a schedule.
• Verify and enrich the list so bounce rate stays under 1%.
• Run real seed-inbox placement tests across Google, Microsoft, and Azure, not just a dashboard score.
• Monitor reputation continuously and pull any domain that starts to slip before it drags the rest down.

Run all seven and inbox placement stops being a mystery. You can pressure-test your copy separately with our free spam words checker, but fix the infrastructure first.

What 98.5% placement actually changes

Placement is not a vanity metric. It is the multiplier on every other number in your campaign. At ~60% placement, four out of every ten good prospects never see your message, so your reply rate is capped before anyone reads a word. At 98.5%, the email gets seen, and the copy and targeting get to do their job.

That is the difference behind results like ATI, where 78K emails on owned infrastructure produced a 37% positive reply rate and $300K+ CAD in pipeline. You can see more of these in our case studies. The whole system, dedicated domains, 52 warmed mailboxes, enrichment, and monitoring, is what we build and hand over in the 3-month outbound pilot. At day 90 you own all of it, so the deliverability you are paying to fix does not walk out the door when the engagement ends.