Sometimes there's no warm signal. You just know which titles, company sizes and industries buy your product. That's cold outreach, and on LinkedIn it lives or dies on two things: how tight the list is, and whether the first message could only have been written to that one person.
Short answer: build the list from structured filters, score it against an ICP, send the connection request without a note, then open with one real detail about their company and one question. Follow up once with proof, then stop. Updately's Cold outreach to your ICP template runs exactly that, with an AI agent qualifying replies and booking a discovery call.
Key Takeaways
| Question | Answer |
|---|---|
| When should you use it? | When you have a clear ICP but no warm signal |
| What carries the campaign? | A tight list and a personalised first message |
| Does the invite have a note? | No. The message after acceptance does the work |
| How long is the opener? | Under 250 characters: one detail, one question |
| What's the goal? | Qualify interest and book a discovery call |
When cold is the right choice
Warm campaigns start from something the prospect did. Cold campaigns start from who the prospect is. Our comparison of warm outbound vs cold outbound goes deeper, but use this template when:
- Your market is defined by firmographics. "Heads of finance at UK SaaS companies with 50 to 200 staff" is a list you can build from filters.
- You've used up your warm lists. Community members, post engagers and competitor users are smaller pools.
If you do have a signal, use it: the post engagement and shared community templates start warmer.
How to build a tight cold list
LinkedIn caps connection requests at roughly 100 a week per account. Every request spent on a poor fit is one you can't send to a good one, so the list matters more than the copy. Useful sources:
- Profile search: LinkedIn people search by title, industry, location and headcount, or a search URL you've refined and pasted in.
- Sales Navigator list import: lead lists you've already built, imported as they are.
- LinkedIn jobs: companies hiring the role you sell to, turned into the decision makers there. The open role gives the opener a real detail to use.
- Tech stack: companies using a technology that pairs with yours.
Then attach an ICP filter so every lead is scored before it reaches the campaign. The template's own description is blunt: it relies on personalisation and a tight ICP filter to stay out of spam territory. If you're starting from a company list, see how to find decision makers at target accounts.
The cold outreach sequence, day by day
Delays count from the previous step. Messages after the invitation wait until it's accepted, and everything stops as soon as they reply.
| Day | Step | What it does |
|---|---|---|
| 0 | View profile | A quiet first touch |
| 1 | Connection request, no note | A plain request, withdrawn automatically if not accepted within 14 days |
| 1 day after they accept | Personalised opener | AI-written per person: one specific, real detail of their company or role, then one question about how they handle the problem you solve. Under 250 characters, no pitch |
| 4 days after they accept | Proof follow-up | AI-written from the conversation so far: one short, concrete result from your Knowledge if there is one, then asks if it's worth a quick look |
| 11 days after they accept | Breakup | AI-written: says it'll stop here and leaves one line on when you'd be useful. No question |
Why the invitation has no note
A note on a cold request has to explain why a stranger wants to connect, and the honest answer is usually "I want to sell you something". A blank request skips that, and the opener after acceptance can be specific without the note's length limit. If you want to test a note anyway, read our LinkedIn connection request best practices.
What a good cold opener looks like
- Weak: "I see you're doing great things at Acme, would love to share how we help companies scale."
- Better: "Saw Acme opened a second support office in Leeds. When volume jumps like that, how are you routing tickets between the two teams?"
The second could only have gone to that person, asks about their process, and leaves the product out. That's the bar the AI step is told to hit.
The proof step depends on you
The day 4 follow-up brings in a result, but only one that exists in your Knowledge. If Knowledge has none, the AI won't invent one. Add a short case study or two before launch.
Why there's a breakup
The last message says you'll stop and gives one line on when you'd be useful. A prospect who wasn't ready this month is left with a polite exit and a reason to come back, not a fourth ask.
How the AI agent handles replies
The objective is "Qualify interest and book a discovery call." The rules:
- "Why are you reaching out?": it answers honestly, saying what it noticed about their company and why it seemed relevant.
- Interest: one qualifying question about their current process, then a call suggestion with your booking link.
- A price question: answered only from Knowledge. If the price isn't there, the chat comes to you.
- "We already have a tool": it asks what they'd improve and, if there's an opening, offers a quick comparison.
- "Not interested": it thanks them and ends the chat.
Like every template, it also hands over (HUMAN INTERVENTION) whenever it lacks the facts to answer, or when someone shares an email, phone number or calendar link. Once the call is booked, it ends the chat. Handed-over chats land in Needs you.
Follow-ups when a cold prospect goes quiet
The Follow-ups card sends two messages, with waits of 3 days and 7 days, a little more patient than the warm templates. The first picks up where it stopped with one low-pressure question. The second adds one new, concrete reason to talk and ends with a yes or no question.
How to adapt the cold template
- Knowledge: two or three results you can stand behind, the questions prospects really ask, and pricing (or leave it out on purpose so price questions come to you).
- Who you're reaching out to: say how the list was built ("finance leads at 50 to 200 person SaaS companies, from a profile search"), so the opener and the honest "why you" answer match.
- Voice: two or three cold messages you wrote that got replies.
- Files & Links: a comparison page, for when the agent offers a comparison.
Common mistakes: a loose list (no opener saves "everyone with manager in their title"), editing a pitch back into the opener, launching with empty Knowledge, and running at full weekly volume from day one instead of starting with a slice. See is LinkedIn automation safe in 2026 for pacing.
How to launch it in Updately
- Build the list with a signal and an ICP filter, then create a campaign.
- On the AI Agent tab, click Start from a template and choose Cold outreach to your ICP under B2B SaaS & founders. The sequence, reply rules and follow-ups are filled in; your voice, Knowledge, files and booking link stay.
- Add results to Knowledge and a discovery call booking link.
- In the simulator, play "why did you message me?", strong interest, a price question, "we already have a tool" and "not interested". Then use Test Sequence on the Workflow tab to read real openers.
- Add leads and activate.
The other founder templates are on the campaign templates page. If your list is mostly users of one competitor, the "My guess is…" opener is a sharper cold opener for that case.
FAQ
Does cold LinkedIn outreach still work for SaaS?
It works when the list is narrow and the first message is about the person, not your product. It doesn't work as a volume game with a generic pitch.
How long should a cold LinkedIn opener be?
Short enough to read in a preview. The template keeps it under 250 characters.
Should a cold connection request have a note?
This template sends none and lets the first message do the work. A note isn't wrong, but a cold one tends to read as a pitch.