What you'll get
- 4 copy-ready blocks
- 6 min read
- Step 1: Extract
- Step 2: Build the line
- Step 2: Build the line
A first line built from a LinkedIn profile goes wrong in one of two directions: creepy (quoting their anniversary post back at them) or hollow ("impressive journey in the logistics space"). The fix is a two-step prompt: extract professional facts first, then build a line from the one fact that connects to why you are writing, with flattery banned outright.
The rules before the prompt
1. Professional facts only: role, work history, posts about work,
skills, company announcements they shared. Nothing personal, ever.
2. No compliments. Relevance is the flattery.
3. Every line must be checkable against the profile, and survive the
reader thinking "they looked me up".
4. If the profile offers nothing real, skip the personalised line.
A strong company-level opener beats a stretched personal one.And the standing caution: check your tools and your company's rules before pasting profile data anywhere, and work from profiles you have genuinely viewed.
The two-step prompt
Step 1: Extract
Below is the text of a LinkedIn profile (and optionally their three
most recent work-related posts). Extract only professional facts:
1. Role and what they likely own, in one line
2. Trajectory: promotions, company moves, what that suggests they
care about now
3. Anything they have said about problems, tools or priorities,
quoted exactly
4. Overlaps with us: shared sector, known peer companies, technology
List facts only. No adjectives, no praise, no inferences about
personality. Mark anything uncertain with (?).
Profile text:
[paste]Step 2: Build the line
We sell [what] and I am writing to this person because [the real
reason: the trigger, the role, the account].
From the facts above, draft three first lines, one sentence each:
- Each must connect ONE fact to the reason I am writing
- No compliments, no "impressive", no "loved your post"
- Nothing that merely proves I read the profile; it must earn its
place in a short email
- Plain words, as a colleague would write
If no fact genuinely connects, say "skip the personal line" and
suggest a company-level opener instead.Good and bad, side by side
BAD "Impressive 15-year journey in supply chain."
(flattery, uncheckable, screams template)
GOOD "You wrote in March that depot scheduling is still spreadsheets
and prayer; that is the exact problem we work on."
BAD "I see we are both passionate about logistics."
(nobody is passionate about logistics in their inbox)
GOOD "You moved from running one depot to running eleven; I am
guessing the scheduling that worked for one is the thing
breaking at eleven."
BAD "Congratulations on your recent post's engagement."
GOOD skip the personal line; open with the company's new distribution
centre instead.The good lines share one shape: their fact, your reason, one sentence, no adjectives.
What to do with the output
- Check the fact against the profile yourself. Thirty seconds. A wrong quote or a misread job change does more damage than no line.
- Rewrite one word minimum. Your voice has to touch it; the line signs with your name. The full-email versions of this discipline are the cold email prompt pack and the role-personalisation prompts.
- Respect the skip. When step 2 says skip, that is the prompt working, not failing. Empty profiles are common; stretched lines from them are the creepy ones.
- One line per thread. The first line opens; the LinkedIn message sequence or the email body carries the rest. Profile facts sprinkled through a whole email reads as surveillance.
Worked example (illustrative, not a real company)
Sam Carter preps outreach to an operations director at Pellow and Rook. Step 1 extracts: eleven years in logistics, promoted from single-depot to regional ops last year, one post quoting "we plan at 6 and replan by 9", a shared peer company in the history. Step 2's best line connects the quote to the reason: "You wrote that the day gets planned at 6 and replanned by 9; that replan is the exact thing we build for." For the finance lead, whose profile is a name and a title, step 2 returns "skip the personal line" and the email opens with the company's depot expansion instead. Both emails get replies; neither reads as a bot's work, because one fact was real and the other line was never forced. All names are invented.
Five mistakes to avoid
- Asking AI to "personalise this email". The vague instruction produces flattery, because flattery is the statistical average of personalisation. The two-step structure exists to prevent exactly that.
- Using personal-life content. The marathon, the wedding, the bereavement post: off limits without exception. Rule 1 is not a style preference; it is the line between research and creepiness.
- Sending the unverified line. The model misreads tenures and attributes posts wrongly. The thirty-second check is the whole quality system.
- Forcing a line from a dead profile. "I see you have been at Acme for six years" proves only that you looked. The skip rule produces better emails than the stretch ever does.
- Scaling it into a mail merge. The prompt works at the speed of accounts that matter. Run across two thousand profiles it regresses to the template everyone deletes, because nobody is checking the facts.
Why this works on named accounts
Common questions
How do you personalise cold outreach with AI without sounding fake?
Separate extraction from writing: have the model list checkable professional facts first, then build one line that connects a single fact to your reason for writing, with compliments banned. Verify the fact yourself and rewrite a word or two. Fake is what happens when flattery substitutes for relevance.
What should you never use from a LinkedIn profile?
Anything personal: family, health, hobbies, life events, and anything from posts about them. Professional facts only, and only ones you would be comfortable being asked about on the call. The test is whether the line reads as preparation or surveillance.
Is a personalised first line always worth it?
No, and the prompt says so: when a profile offers nothing genuinely connected to your reason for writing, a sharp company-level opener (their announcement, their hiring, their market) beats a stretched personal one. Personalisation is a means; relevance is the end.
More from the Academy
ChatGPT and Claude Prompts for Writing Cold Emails
For sellers who use ChatGPT or Claude to draft outreach and get emails that sound like everyone else's.
You leave with a set-up prompt, a research block, 10 prompts that draft, cut, criticise and fact-check, and one worked draft.
AI Prompts to Personalise by Buying-Committee Role
For sellers who send the same message to every person in an account.
You leave with a fact sheet, 7 prompts, a consistency check and one fact turned into 3 role-specific openings.
LinkedIn Message Templates
For anyone generating B2B leads on LinkedIn by hand.
You leave with templates for the whole sequence: the connection note, the first message after accepting, two follow-ups and the line that moves it to email or a call.
Want this in your inbox?
Optional. We use your email only to send you this, nothing else.