Why This Document Exists
Most B2B cold email in 2026 still reads like it was written in 2019. Average reply rates sit at one to two percent, and almost half the volume lands in spam before the recipient ever sees it.
That is not a template problem. It is a psychology problem.
If you hire us to run your pipeline, every first-touch message your prospects receive passes five rules, each mapped to a specific psychological primitive from the behavioral economics literature. Each rule is enforced by a template gate that rewrites the message — or kills it — if the rule is not met.
This page documents the five rules and the reasoning behind them.
Log vs. Logic
A typical vendor will show you what they wrote:
"Here is the cold email template we use. Just fill in the blanks."
That is a log. It tells you nothing about why the 73rd word kills the reply rate of the 72nd, or why a single hedge word changes a message from accusatory to collaborative.
We tell you what the message must enforce:
"Every first-touch email is under 90 words, begins with a specific signal drawn from the prospect's public work in the last 90 days, and ends with a single asynchronous CTA that costs the recipient less than 10 seconds."
The five rules below describe that enforcement.
Three Failure Patterns Behind Every Dead Sequence
Before the rules, the anti-patterns. Nearly every cold email that fails fails for one of these three reasons, and each maps to a broken psychological contract.
Pattern 1 — No signal, only pitch. The message opens with a value proposition. The prospect has no prior context, no reason to care, and no evidence that the sender has done the homework. Cialdini's reciprocity primitive never fires because nothing has been given first.
Pattern 2 — Generic pain, assumed. The message names a pain ("struggling with pipeline?") without grounding it in something the prospect said, wrote, or shipped. This triggers the machine-response failure Nass documented: the recipient treats the sender as software, and software does not earn replies.
Pattern 3 — Too many CTAs, each expensive. "Book a 30-minute demo" is a 30-minute ask from a stranger. The prospect's decision cost outweighs any possible upside. Berger's social-currency principle runs in reverse: the ask is high-status for the sender and low-status for the receiver.
The Sniper Formula exists to violate all three patterns deliberately.
Rule 1 — Signal Mirror
The opening sentence must reference a specific, verifiable piece of the prospect's public work from the last 90 days.
"I saw the $0M+ counter still on your homepage" is a signal. "I saw you work in marketing" is not.
The primitive is attention earned through specificity. When a stranger cites something only someone who actually looked at the prospect's work would know, the message crosses the bar from template to research. Signal-first openers consistently test three to five times higher reply rate than generic openers in public B2B benchmarks (Woodpecker 2024, Ebsta State of Outbound 2024).
The template gate requires Tier 1 evidence — something the prospect personally published or shipped, with a URL. If it cannot be cited, the message is killed.
Rule 2 — Pain Translation (Hedged)
After the signal, translate it into a probable working pain — with a hedge word.
"which probably means Monday mornings are spent reviewing SDR output instead of closing" is a translation. "which means you have a revenue problem" is not.
The primitive is cognitive empathy under uncertainty. Humans trust senders who show the work of reasoning rather than those who declare conclusions. The hedge word — "probably," "if so," "usually" — explicitly signals that we do not claim to know the prospect's situation, only to have modeled it against a pattern.
Messages without hedges read as accusatory. Messages with hedges read as collaborative. A single-word change shifts response tone measurably.
Rule 3 — Proof Shard (Peer-Framed)
One sentence of evidence that this pattern has a known resolution, drawn from peer data.
"Agencies in this spot usually cut SDR review time sixty percent once the first filter is automated" is a proof shard. "We have helped many companies with this" is not.
The primitive is social proof without brag. The sentence references a group the prospect belongs to, not the sender's own book of business. Cialdini's original work showed this distinction is neural, not cosmetic: peer references activate a different trust pathway than sender self-presentation.
Our proof shards are limited to patterns we have directly observed or quantified, never borrowed from generic marketing copy. If we cannot cite a specific range from our own data or a named study, the shard is cut.
Rule 4 — Async CTA
The final line must propose an action that costs the prospect less than ten seconds.
"Reply 'curious' and I'll send a write-up" is an async CTA. "Have twenty minutes next Tuesday?" is not.
The primitive is minimum viable commitment. Every increment of effort demanded from a stranger is taxed at roughly three to five times its actual cost, because the prospect has to model whether the sender is worth it. A one-word reply is priced at near zero. A twenty-minute meeting is priced at sixty to ninety minutes of mental overhead.
We also forbid any CTA that introduces scheduling, calendar links, or qualifying questions at first touch. Those are second-touch moves. One-word CTAs test two to three times higher reply rate than calendar CTAs at first touch in public benchmarks.
Rule 5 — Length Discipline
Sixty to ninety words, first touch. Not a style preference — a gate.
Under sixty words the message reads as dismissive. Over ninety it reads as a pitch. The band in between is where the first four rules can fire without any one of them crowding out another.
This is also below the "template" heuristic threshold used by most B2B inbox filters in 2026. Messages that mirror average marketing-automation length get de-prioritized by spam-score classifiers even when the content is human-written. Sixty to ninety keeps the message under that classifier ceiling.
The word-count check runs on every draft. If a message fails three rewrites, the lead goes back to Gate 1 of the Suppression protocol rather than being softened.
What Breaks It
Three anti-patterns the template gate kills before they leave the draft folder:
- Fake signals. "I saw you are interested in growth" is not a signal. Generic industry observations are the silent killer of reply rate. Any opener that could apply to more than fifty prospects is rejected.
- Two CTAs in one message. The moment a message contains both "reply" and "book a meeting," both fail. The prospect's decision cost doubles. One ask per message, always.
- Tool names and category jargon. "We use Apollo and n8n to automate..." is a sender-centered sentence. The prospect does not care what we use. Every tool name is stripped before send.
Receipts
The numbers below are the current state of first-touch outcomes across Waves 1 through 7 on the dedicated enpits.dev cold-outbound domain.
| Metric | Value |
|---|---|
| First-touch messages sent | 41 |
| Hard bounces | 0 |
| Spam complaints | 0 |
| Opt-outs | 1 |
| Median first-touch word count | 72 |
| Messages killed pre-send by the template gate | 8 |
Eight messages written during these waves never left the draft folder because the template gate failed them on Rule 1 (no Tier 1 signal) or Rule 4 (CTA too expensive). That is the formula at work. The gate's job is not to let good messages through. It is to kill the ones that would have hurt the domain.
What This Means For You
If we run your pipeline, every first-touch message your prospects receive has passed the five rules above. Signal specific, pain hedged, proof peer-framed, CTA one-word, length inside sixty to ninety words.
When a message does not pass, it does not soften — it dies. That discipline is the reason the Receipts table has the zeros it does.
This document is part of a four-part infrastructure series. See also: Suppression 3-Gate (Safety Logic), Architecture (Technical Logic), Vertical Analysis (Insight Logic) — forthcoming.