How to Write Cold Emails That Get Replies: Research, Personalisation and Follow-Up Sequences

Most cold emails fail before the recipient finishes the first line — not because cold email does not work, but because most cold emails are not really written for the person receiving them. They are written for the sender: our product, our features, our funding round, our fifteen-minute demo. The recipient, who has never heard of you and owes you nothing, deletes it in the two seconds it deserves.

Emails that get replies invert that. They are short, specific, honest about being outreach, and built on a genuine reason why this particular person might care right now. That is not a trick; it is a discipline — of research, of restraint, and of respecting the reader's time and legal rights. This guide walks through the whole craft: how to research prospects efficiently, how to personalise without being creepy, how to write subject lines and openers that earn the next sentence, how to structure follow-up sequences that persist without pestering, and what the law actually requires of every message you send.

A ground rule before anything else: everything below assumes you are sending relevant, truthful B2B outreach to people whose professional role makes your message plausibly useful to them, with a working opt-out and honest sender identity. Copywriting cannot rescue a bad list or a deceptive premise, and it should not try. Great cold email is good targeting made articulate.

Research First: The Reply Is Earned Before You Write a Word

The strongest predictor of a reply is not your subject line — it is whether the recipient, on reading your email, thinks "this is actually relevant to me, right now." That judgement is made in seconds, and you can only pass it if you did the work up front.

What to look for, in order of value

A trigger event. Something that changed recently and plausibly creates the problem you solve: a new role (people in their first 90 days are actively fixing things), a funding announcement, a product launch, hiring for roles adjacent to your solution, an expansion into a new market, a technology change visible on their website.
Evidence of the problem itself. Job postings are gold here — a company hiring three SDRs has outbound problems; a company hiring its first data engineer has data problems. So are public artefacts: a slow website, an outdated careers page, a review mentioning a specific pain.
Something they said. A podcast appearance, a LinkedIn post, a conference talk, a blog article. Referencing a person's actual ideas — accurately, and only if you genuinely engaged with them — is the deepest personalisation available.
Firmographic fit. Industry, size, business model. This is the minimum bar, not personalisation. "You're a B2B SaaS company" earns nothing; it merely qualifies them for the list.

Budget your research time honestly. A workable rhythm for most teams is three to five minutes per prospect: enough to find one specific, relevant observation, not enough to write a biography. If five minutes of looking produces nothing specific to say to this person, that is useful information — they probably should not be on this list.

The gotcha: personalisation that proves you looked is not the same as personalisation that proves you thought. "Congrats on the Series B!" or "I saw you went to Ohio State" shows you glanced at LinkedIn; every other sender did too, and recipients have learned to skim past it. The observation only earns a reply when it connects to the reason you are writing: "You announced a Series B and you're hiring four sales roles — most teams at that stage hit X problem within two quarters" is research doing actual work. If your personalised first line could be deleted without changing the email's logic, it is decoration, and readers can tell.

A second trap in the same neighbourhood: creepy personalisation. Referencing someone's personal social media, their family, their home city move, or anything that feels like surveillance rather than professional homework will kill the conversation and, under GDPR, drifts away from anything defensible as legitimate interest. Stay strictly inside their professional, public, work-relevant footprint.

Subject Lines: Small, Honest, Specific

The subject line has one job: get the email opened by the right person without misrepresenting what is inside. That second clause is not just ethics — it is law. CAN-SPAM explicitly prohibits deceptive subject lines, and it is also self-defeating: a bait-and-switch open converts to an annoyed delete and, often, a spam complaint that damages your sending reputation.

What works

Short and plain. Two to five words, lowercase or sentence case, reading like an email from a colleague: "question about your SDR hiring", "onboarding at Acme", "your post on churn".
Specific to them. Their company name, their initiative, their stated problem. Specificity signals "this was written for you" before the email is even opened.
Accurate. The subject should describe the actual content. If the email is about your analytics product, "quick question" is weak but tolerable; "re: our call" when there was no call is a lie and a legal problem.

What to avoid

Fake reply or forward prefixes ("Re:", "Fwd:") on first-touch emails — deceptive, prohibited, and instantly reputation-burning when noticed.
Clickbait, false urgency, and mystery ("You won't believe this", "Time sensitive"). These pattern-match to spam because spammers invented them.
Salesy formatting: ALL CAPS, multiple exclamation marks, emoji strings. You are not fighting a filter here; you are fighting the reader's instant categorisation of your email as mass mail.

Here is the trap most people miss: obsessing over open rates at all. Since Apple Mail Privacy Protection and similar features began prefetching messages, open tracking is unreliable — many "opens" are machines, not people. Worse, optimising subject lines for opens drags you toward curiosity-gap tactics that inflate opens while depressing replies. Judge subject lines by the reply rate of the emails behind them, and treat opens as a rough directional signal at best.

The Body: A Structure That Respects the Reader

A cold email is not a pitch document; it is a request for permission to have a conversation. The reliable structure is four moves, in under 120 words or so:

Relevance: one sentence showing why them, why now — your research finding. Not "My name is X and I'm the founder of Y"; your signature handles introductions.
Problem: one or two sentences naming the specific problem you believe they may have, in their language, tied to the relevance line.
Credibility: one sentence on why you are worth listening to — how you help companies like theirs. Truthful and concrete; if you cite results or name customers, they must be real and you must be permitted to name them. No invented statistics, no "we helped a company just like yours 10x revenue" unless you can stand behind every word.
Ask: one clear, low-friction question.

Write like a person

Plain text, or close to it. One-to-one business email is plain text with a simple signature. Heavy HTML templates, banner images, and buttons signal "marketing blast" to both the recipient and their mailbox provider.
Short sentences, short paragraphs. Most B2B email is read on a phone. If your email requires scrolling to find the ask, cut it.
Kill the jargon. "Leverage synergies to optimise workflows" says nothing. "Your reps spend Fridays copying data between HubSpot and spreadsheets" says everything. Concrete beats abstract in every line.
One idea per email. One problem, one offer, one ask. If you sell three things, pick the one your research says fits.

The call to action

Asking a stranger for 30 minutes in the first email is a big request from a position of zero trust. Lower the bar. Interest-based CTAs ("Worth a look?", "Is churn something you're focused on this quarter?") consistently outperform calendar demands because they let the recipient say a small yes. A calendar link is fine to include once someone has expressed interest; leading with "book time on my calendar here" asks them to do your job for you.

The gotcha: false personalisation at scale is worse than honest brevity. Mail-merge sentences that pretend to be handwritten — "I was reading your website and loved what you're doing in the space!" — are transparently templated, and the pretence itself costs trust. If you cannot genuinely personalise a segment, it is more effective and more honest to be a well-targeted, well-written relevant email without theatrical personalisation than a bad imitation of a personal note. Recipients forgive "this is clearly a targeted outreach email, but it's relevant and respectful"; they do not forgive being lied to about it.

Follow-Up Sequences: Persistence Without Pestering

Most replies to cold outreach come from follow-ups, not the first touch — not because nagging works, but because timing is everything and your first email usually arrives on the wrong day. A follow-up is a second roll of the timing dice. That justifies a short sequence; it does not justify a siege.

A sane sequence shape

Email 1: the researched, relevant email above.
Email 2 (3–4 business days later): a brief bump in the same thread, ideally adding one new element — a different angle on the problem, a relevant resource, a one-line clarification of who this is for. "Just bumping this to the top of your inbox" adds nothing; add something.
Email 3 (about a week later): a different value angle. Share something genuinely useful — a relevant article, a benchmark, a specific observation about their setup — with a soft ask.
Email 4 (one to two weeks later): a polite close: you will stop reaching out, the door is open if the problem becomes relevant. Keep it warm and brief; skip the passive-aggressive theatrics.

Three to five total touches over three to five weeks is plenty. Beyond that, reply rates flatline while complaint risk climbs, and under European legitimate-interest reasoning, repeated contact after sustained silence gets progressively harder to defend as proportionate.

Retire the "breakup email" clichés while you are at it. "Should I stay or should I go?", the guilt-trip about being ignored, the fake "closing your file" bureaucracy — recipients have seen every one of these hundreds of times, and manipulative framing from a stranger reads as exactly what it is.

Here is the trap most people miss with sequences: automation that keeps firing after circumstances change. A prospect replies "not interested" and step three goes out anyway. Someone books a meeting and still gets the breakup email. A person opts out and the sequence continues because the suppression list was not wired into the tool. Every one of these is a trust-destroying, sometimes law-breaking failure, and every one is preventable: configure your sequencer to halt on any reply (including out-of-office if you prefer to reschedule manually rather than auto-continue), sync opt-outs to a global suppression list instantly, and audit these settings whenever you change tools. Test the kill-switches with your own address before a campaign goes live, not after the first angry reply.

Handling replies — including the negative ones

Answer fast; interest decays by the hour. Answer "not interested" with grace and removal, not a rebuttal. Treat "not now" as what it says: log it, suppress from current sequences, and only revisit months later if there is a genuinely new reason. And treat an angry reply as free consulting: it usually means your targeting, tone, or frequency is off for that whole segment, not just that one person.

What the Law Requires of Every Message

Compliance in cold email is not a footer you paste in at the end; it shapes what you may send, to whom, and what every message must contain. The details vary by jurisdiction, and this is general guidance rather than legal advice — but the core obligations are clear and, helpfully, they overlap almost entirely with what makes outreach effective.

CAN-SPAM (United States)

No deceptive headers or subject lines. Your From name, domain, and subject must honestly identify who you are and what the message is.
Identify yourself and be reachable. Include your real business name and a valid physical postal address in every commercial email.
Provide a clear opt-out and honour it promptly. The law allows up to 10 business days; competent senders process it immediately, and you may not charge, quiz, or require a login to opt out.
You are responsible even when tools or agencies send for you. Outsourcing the sending does not outsource the liability.

Note what CAN-SPAM does not require: prior consent. Unsolicited B2B email is lawful in the US provided the rules above are met. That permissiveness is exactly why the honesty requirements have teeth — they are the whole framework.

GDPR and PECR (EU and UK)

European rules are stricter and apply based on where the recipient is, not where you are. For B2B outreach to people at companies, most senders rely on the legitimate interests lawful basis under GDPR, which requires you to actually do the balancing work: your message must relate to the recipient's professional role, the data you hold (name, work email, role) must be proportionate and professionally sourced, and their interests must not override yours. Document this reasoning — a short legitimate interests assessment — before campaigns run, not after a complaint arrives.

Alongside that, GDPR grants rights you must operationally support: recipients can ask where you got their data (your first email should make this easy to answer honestly), and an objection to direct marketing is absolute — once received, marketing to that person ends, permanently, no balancing test. PECR and its EU equivalents add country-specific rules on electronic marketing; several European countries effectively require prior consent even for B2B, especially when emailing individuals rather than corporate roles. If you target multiple European markets, check the rules per country or take advice — the "B2B is always fine in Europe" assumption is wrong often enough to be dangerous.

The gotcha: the sender-identity shortcut. Teams sending from a secondary outreach domain sometimes leave the footer vague — no legal name, no postal address, a bare unsubscribe — reasoning that a light footprint "looks more personal." In the US that omits legally required elements; in Europe it undermines the transparency that legitimate interest depends on; and everywhere it makes suspicious recipients more likely to report you rather than politely decline. Every cold email needs your real company name, a real postal address, and a frictionless way to say stop — even when it is one plain-text email from a founder. Honest identity is a requirement, not a design choice.

Testing and Iterating Without Fooling Yourself

Cold email rewards iteration, but only if you measure the right thing on enough data.

Optimise for positive reply rate, not opens, not clicks, not raw replies (which include "unsubscribe me"). A positive reply — interest, a question, a referral to the right person — is the unit of progress. Meetings booked is the metric behind that.
Change one variable at a time. Test the offer and the target segment before you test wording. A mediocre email to the right segment beats a brilliant email to the wrong one, so segment-level differences dwarf copy tweaks.
Get enough volume before judging. With 50 sends per variant, differences of a few replies are noise. Run comparisons across at least a few hundred sends per variant before declaring winners, and expect results to vary by segment.
Read the replies qualitatively. What people say when they respond — the objections, the misunderstandings, the "actually we need X" — is worth more than any percentage. It tells you what to fix in the email and, frequently, in the offer itself.

Keep a simple log: date, segment, variant, sends, positive replies, meetings, and notable qualitative feedback. Six weeks of honest notes will teach you more than any template library.

Conclusion: Relevance, Honesty, Restraint

Everything that makes cold email work is a form of respect. Research earns the right to interrupt. Honest subject lines earn the open without stealing it. Short, concrete, single-idea bodies respect the reader's attention. Modest, additive follow-up sequences respect their silence. Clear identity and effortless opt-outs respect their rights — and happen to be the law.

The compounding effect is real: senders who work this way get more replies per email, fewer complaints, better sending reputation, and therefore better inbox placement for the next campaign — a virtuous cycle that spammy shortcuts can never enter. Write every email as if the recipient will look you up afterwards, because the good ones will. If what they find is a real company, a truthful message, and a sender who took their time seriously, you have already done the hardest part of turning a cold contact into a warm conversation.