Deliverability

# Deliverability

Deliverability is the probability that an outbound email lands in the recipient's inbox rather than spam, junk, or a quarantine. It is the most under-appreciated number in outbound. Reply rate gets the attention because it shows up in dashboards, but deliverability sits underneath it and silently caps how high reply rate can go. A campaign with 30 percent inbox placement is functionally a different campaign from one with 95 percent placement, and the visible metrics will not tell you which one you are running unless you look.

The mechanics are straightforward to describe and difficult to control. When a sending server delivers a message, the receiving server scores it based on the sender's domain reputation, IP reputation, authentication records, content patterns, prior engagement from this recipient, and whatever proprietary signals the provider has trained their filter on. The message is then placed in the inbox, the promotions tab, the spam folder, or quarantined entirely. The decision is opaque from the sender's side, and the only way to know what happened is to seed the campaign with test inboxes you control.

Why it matters

Everything else in outbound is downstream of deliverability. A flawless ICP, a hand-written message, and a perfectly tuned sequence cadence cannot overcome a campaign that lands in spam. The recipient never sees it. The reply rate drops to nearly zero. The team blames the message and rewrites it, which does nothing, because the message never arrived.

Deliverability is also the constraint that determines how aggressive a sender can be. With 95 percent inbox placement, a team can run reasonable volume on a clean list and expect normal reply rates. With 60 percent placement, the same campaign loses a third of its audience invisibly. With 30 percent, the campaign is effectively wasted spend. Most outbound problems that look like "bad list" or "bad message" are actually deliverability problems in disguise.

The defenders on the other side, Google's filter, Microsoft's filter, Proofpoint, Mimecast, the others, have been investing aggressively in detection since the 2023 LLM-driven cold email wave. They have access to engagement data the sender does not, and they update their models faster than senders can adapt. A campaign that landed in the inbox last quarter may land in spam this quarter without any change on the sender's side.

This is the structural reason most experienced operators run dedicated cold sending infrastructure separate from their primary brand domain. A deliverability fire on the cold domain does not then take down the company's transactional and team email.

Common pitfalls

The first pitfall is sending from the primary domain. A founder sends a few thousand cold messages from their company.com address, the domain gets flagged, and now their hiring emails, customer support, and investor updates are all going to spam. The recovery is slow and the damage spreads. The fix is to register secondary domains for cold sending and route all warm communication through the primary.

The second is skipping authentication. SPF, DKIM, and DMARC are not optional. Major providers actively reject or downgrade unauthenticated mail in 2026, particularly after Google's bulk-sender rules tightened in 2024. The records take an hour to configure and the protection lasts forever.

The third is sending too fast on a cold domain. A new domain has no reputation, which means the receiving providers default to suspicion. Sending fifty messages from a brand-new domain on day one is a fast track to spam folders across the board. The fix is email warmup, gradually ramping volume over four to eight weeks while the domain accumulates engagement signal.

The fourth is using spammy content patterns. Excessive links, image-heavy messages, all-caps subject lines, unsubscribe-from-our-amazing-newsletter language. Filters are aggressive on this. A plain-text-looking message with one link and one human-sounding subject performs better than a styled HTML email almost every time.

How it shows up in agent-driven workflows

In Coil specifically, deliverability remains outside the product boundary. Coil can operate bounded sources and reviewed recipes. A configured SmartLead connection supports a separate direct handoff, while an admin-published webhook automation records its own delivery result. Neither path gives Coil mailbox warmup, inbox placement, sequence scheduling, or downstream reply telemetry. The external sender owns those controls.

Coil does not currently ingest placement, bounce, warmup, or reply telemetry from senders, and its webhook automations do not automatically throttle, pause, or rotate campaigns. Operators review those signals and controls in the sending system.

Related concepts

Email warmup is the prerequisite for any new sending domain. Without it, deliverability is permanently capped.

Cold outreach is the activity whose ceiling is set by deliverability. A team that fixes its deliverability typically sees a step-change in reply rate without touching the message.

Reply rate is the visible KPI that hides deliverability problems. If reply rate is dropping and you cannot find a message-level cause, check placement.

Sequence cadence interacts with deliverability, too many sends to the same prospect in a short window can trigger the filter at the provider level.

For more on how Coil splits deliverability work from list and message work, see the Coil methodology pillar.