Copywriting & personalisation at scale
Write short messages a busy stranger would actually reply to, then make the system write them for thousands.
Why it matters
Outbound lives or dies on the message. GTM engineers rarely own the brand voice, but they build the systems that assemble copy for thousands of people: variables, segment-level angles, AI-written lines and the checks that keep them sane. Bad automated personalisation is worse than none: "I loved your recent post about {{topic}}" is instantly recognisable. Knowing what good cold copy looks like is what lets you build systems that produce it.
What it's made of
- Writing short, plain, specific cold emails
- Leading with the prospect's situation, not your product
- Segment-level angles vs one-to-one personalisation
- Using signals and research as the reason for reaching out
- Low-friction calls to action
- Prompting AI to write one line well, with guardrails
- Quality-checking generated copy at scale
- Follow-ups that add something new
What a good cold email looks like
Short: often 50–90 words. Plain text, no images, at most one link (often none). Written the way a thoughtful person writes to a colleague. The structure most good ones share: a reason you're reaching out to this person now, a specific problem people in their role usually have, a brief line on how others solved it, and an easy question.
Read it on a phone. If it needs scrolling, it's too long. If you'd be embarrassed to send it to one person by hand, don't send it to 2,000.
Where personalisation actually comes from
Personalisation that works is relevance, not compliments. In order of usual impact:
- The segment: a message written for heads of RevOps at 200-person SaaS companies beats a generic one with a first name merged in.
- The signal: a true, recent reason ("you're hiring three SDRs in Austin") tied to a likely problem.
- Specific research: something about their company that only makes sense for them, used to support the point rather than to flatter.
- Mentioning their post or podcast only works if the email then says something useful about it.
AI-written lines with guardrails
Ask the model to write one specific line, not the whole email. Give it the research inputs, the segment angle, three examples of good lines and two of bad ones, a word limit, and an explicit instruction to return nothing if the inputs don't support a good line. Then fall back to a segment-level line.
Before any send, read a random sample of 30–50 generated lines. Look for invented facts, weird phrasing, repeated openers, and anything that sounds like marketing. If more than a few fail, fix the inputs or prompt, not the individual rows.
Follow-ups and calls to action
Most replies come after the first email, so plan the sequence. Each follow-up should add something: a different angle, a short proof point, a useful resource. "Just bumping this" adds nothing. Keep the ask small: "Worth a look?" or "Is this on your radar this quarter?" gets more replies than a calendar link in email one.
Mid-level looks like
Writes clear three-to-five sentence emails, uses variables correctly, and generates a personalised first line with AI.
Senior looks like
Builds angles per segment and signal, constrains AI output with examples and rules, samples output for quality before every send, and kills copy based on reply data rather than taste.
Common pitfalls
- Opening with the company and product ("We are a leading platform that…").
- Fake personalisation: flattery about a post the writer clearly didn't read.
- Asking for 30 minutes in the first message when a yes/no question would do.
- Letting AI write the whole email, which tends to produce long, polished, generic text.
Practice drills
Drill 1
Write a three-email sequence for heads of finance at 300–1,000 person companies who just hired a new controller. Keep each under 90 words.
A strong answer: The signal used as a reason, a role-specific problem, no product-first opening, a small ask, and each follow-up adding a new angle.
Drill 2
Build an AI step that writes a first line from company research. Run it on 50 rows, grade every output pass/fail with written criteria, and revise until 90% pass.
A strong answer: Clear pass/fail criteria, a fallback for weak inputs, examples in the prompt, and measured improvement between versions.
Drill 3
Rewrite this opener for a cold email: "I hope this email finds you well! I'm reaching out because our innovative AI-powered platform helps companies like yours unlock growth."
A strong answer: A version that's about the reader, specific, shorter, and free of filler and hype words.
Best resources for this skill
Repos, courses, articles and podcasts, ranked by how much they teach.
Cold-IQ/ColdIQ-s-GTM-Skills
Claude Code skills for Go-To-Market: 137 sales triggers, 34 email templates, 11 GTM plays, and Clay enrichment workflows.
Apollo Academy
Free courses and plays on outbound programs, cold email, cold calling and data enrichment, useful for understanding what the sequences you automate should actually do.
PodcastScaling Smarter: Cold Email, GTM Engineering & Operational Rigour with Noah Adelstein of Rippling
In this episode, Jeff sits down with Noah Adelstein, who leads international growth at Rippling and hosts The GTM Engineer podcast. They unpack how Noah helped scale Rippling’s Australia team from 10 to 80+, the evolving state of cold email, and how GTM engineers are reshaping the way we think about growth. Noah shares why marketers must obsess over follow-up as much as lead gen, how AI fits into…
Read next
Fanatical Prospecting
Jeb Blount
A blunt guide to the daily prospecting work your tooling is meant to amplify. Read it to understand what reps actually need from lists, sequences and call blocks.
SPIN Selling
Neil Rackham
Research-backed questioning for large, complex sales. It helps you design discovery fields, call notes and AI summaries around the questions that matter.
The Challenger Sale
Matthew Dixon, Brent Adamson
Argues that teaching buyers something new beats relationship building in complex deals. Useful context for writing outbound copy and insight-led sequences.
Gap Selling
Keenan
Frames every deal as the gap between a buyer's current and future state. A practical lens for deciding what research and enrichment fields a rep needs before the first call.