Messages written for one person.
Every lead gets their own emails and LinkedIn messages. This is how cold outreach should be written, and how our agent writes it for one person at a time.
Two failures
Cold outreach fails in two ways. The first is the template: one message sent to a thousand people with the first name swapped in. Readers recognize it within a line. The second is false personalization: a message that states something about the reader which is not true.
Good outreach avoids both. It is written for one person, and it says nothing about that person or their company that cannot be supported.
The email
A good cold email follows a few principles, and they are the ones our playbook sets.
Write like a peer, not a vendor. The email should read as if it came from someone who understands the reader's work.
Make every sentence earn its place. A cold email is short, and a sentence that does not move the reader toward a reply is cut.
Lead with the reader's world. The reader should see their own situation before they see your company.
Tie personalization to the problem. If the personal opening could be removed and the email still made sense, it was not doing any work.
Ask for one thing, and make it easy to say yes.
The subject line has one job, to get the email opened. The playbook asks for two to four words in lowercase, the kind of subject a colleague would write.
The sequence
One email is not enough. In the benchmark our playbook cites, the first email draws a little more than half of all replies, and the rest arrive on follow-ups. So outreach is a sequence: three to five emails, with the gap between them growing each time.
A follow-up is not a reminder. Each one has to stand on its own, since the reader may not have read the one before. Each one adds a single new thing: a different angle, a piece of proof, an insight.
The sequences our agent writes follow that rhythm. In the ones stored so far, the first email is on day 0, the first follow-up most often on day 3, the second on day 7 or 8, and the third on day 14 or 21.
LinkedIn is a different medium and gets messages of its own. A connection request is limited by the platform to 300 characters, so it has room for one thought.
Ours follows a fixed order: a greeting, what the person's company does and where you can help, what you do, a request for a short call, a demo or a chat, and your website. It carries no link other than your website.
The message for after they accept is not the request repeated. It is a short cold email in its own right, which takes the conversation one step further.
One lead, one run
The writing is done by an agent that works on one lead at a time. Each run starts fresh, with three inputs.
The first is the playbook: its writing principles, a catalog of copywriting frameworks, fifteen email templates for software companies, and its guidance on subject lines, follow-up sequences and LinkedIn messages. The agent must read all of it before it can save anything.
The second is your company: what you sell and to whom, and the whole of the research behind your profile, including the evidence, your competitors and the topics your buyers write about.
The third is the person: their role, their headline and profile, their work history, and their company and what it does.
The agent is not given the event through which the person was found. It writes from who the person is and what their company does.
The writing
The agent does not fill in a template. From the fifteen it picks the one that suits this person, judged by their company, who they are and what they do. Then it writes that email out in final words, from the details it was given.
Some rules hold for every message. An email opens with a greeting and the person's first name. It does not tell people their own title or what they built, because they know. It names your website and says how to book a call. It is signed by a named sender. It is written in final words, with no alternatives left in brackets.
It says what you do in words that fit the reader's work, and does not repeat your company description.
Checked facts
The rule against false personalization is enforced, not requested.
When the agent saves its work it must list each fact about the person or their company that the messages use: the exact words, copied from the lead's record, the field they came from, and why they were used.
Each quote is then compared with the stored record. If a quote is not there word for word, the save is refused and the agent has to correct it. The quotes are stored with the messages, so for any message you can see what it was based on.