swan-gtm Applies to first contact with a stranger. Produces one message, or a batch where every row is grounded in that row's own data.
Sell the reply, not the product
You are not closing in the inbox. You are earning one small "tell me more." Everything below follows from that.
Two things do most of the work before a word is written: the offer and the list. Copy is the delivery vehicle. A great email carrying a weak offer to the wrong segment still loses, and rewriting it is the most common way teams waste a month. If the offer is soft, fix the offer. If the segment is generic, fix the segment.
The skeleton
Subject: 1–3 words, neutral, reads like internal mail
Line 1 — TRIGGER: one true, specific thing about THEM. No "I / we / our."
Line 2 — TENSION: the problem that trigger implies, in their language
Line 3 — PROOF: one outcome, with a real number and a real comparable
CTA — one soft question asking for interest, not for time
- Under 75 words; 25–50 is better. Corpus studies of tens of millions of sends put short, plain, low-reading-grade emails far ahead of long ones.
- Exactly one question. A second CTA lowers replies.
- "You" outweighs "I/we." Count them if unsure.
- Tentative beats certain — "looks like", "I could be off", "guessing". Certainty from a stranger reads as a pitch.
- Formatted for a phone. Most of these are read on one. One-line sentences, white space, no walls.
Relevance comes from the segment, not the merge field
A first name and "saw your post" is not personalization. Relevance means the message would only make sense sent to this kind of company, for this reason, right now. Personalize the first sentence or two; templatize the rest — the alternative doesn't survive contact with volume.
The test that settles it: could this exact email go to a different company and still make sense? If yes, the trigger is decoration. Rebuild around a real business event, not a piece of trivia.
The ask
Ask for interest, not for time. No calendar link on touch one. "Worth a look?" · "Is this a priority right now?" · "Want the two-minute version?" · "…or does this sit with someone else?" An easy-out — "if it's not a priority, no worries" — removes the neediness that kills replies. The meeting is earned in reply two or three.
Deliverability is a copy constraint
The best copy dies in spam, so the writing itself has to obey:
- No links, images, or tracking pixels on touch one. A reply-ask outperforms a link by a wide margin, and the pixel costs you placement.
- Vary the copy at volume so no two sends are identical.
- Plain text with a real signature. A bare, signature-less email reads less legitimate, not more.
- Cap sends per inbox at a conservative daily number and scale by adding inboxes — never by raising the per-inbox limit. This is the single most common self-inflicted deliverability wound.
- Send only to verified addresses. Bounce rate above a couple of percent damages the domain faster than any content mistake.
Follow-up
One email is a coin flip; the sequence is the system. Four to five touches over two to three weeks, each adding something genuinely new — a different angle, a proof point, a useful asset. Never an empty bump. Thread follow-ups as replies so the original personalization stays visible.
What good looks like
The tell of a good operator: they can name the segment-level reason the email exists before they write a line of it, and they fix offers and lists before they touch copy.
The mediocre version is fluent and forgettable — correct structure, a merge field in the opener, a polite ask, and nothing a stranger has any reason to answer. It reads fine to the team and produces nothing.
Good output survives the deletion test: strike the opening line, and the email still makes sense — because the relevance was in the substance, not the decoration.
Rules
- MUST pass the "could this go to anyone else?" test before shipping.
- MUST keep exactly one ask, for interest rather than for time, on the first touch.
- NEVER put a link, image, or tracker in a first touch.
- NEVER invent a signal, a statistic, or a customer to make a line land.
- NEVER raise the per-inbox send limit to scale volume.