# AI SDR playbook for cold LinkedIn leads
> A step by step AI SDR playbook for cold LinkedIn leads: define the target, write the sales script, send connection requests and book meetings.
- HTML: https://getkinlead.com/ai-sdr-playbook-for-cold-leads
- Markdown: https://getkinlead.com/ai-sdr-playbook-for-cold-leads.md

**Short answer:** An AI SDR playbook moves a stranger from a connection request to a meeting by giving the AI a specific sales script to follow. The AI uses the lead's profile and the campaign goal to write a personal opening, answers objections in your tone, and asks for a scheduling link once interest is confirmed.

## Define the customer before you write a word

An agent given a vague target writes vague messages. Tell it to find marketing managers and you get a list mixing interns, directors and one CMO at a 12 person company. Every message then has to hedge, and hedged messages get ignored.

Pick the trigger that makes someone worth messaging this week. Company size, a funding round, a job post that hints at the problem, a line in their headline. Write it as a sentence you could say to a colleague: heads of growth at 30 to 80 person B2B software companies who just hired their first sales rep.

Then write down what disqualifies someone. A profile that matches the title but works at an agency selling to consumers is a bad lead for most B2B tools. The exclusions save you from arguing with the agent later.

**Too broad**

- Target: marketing managers
- Goal: sell a lead gen tool
- Message: I help marketers get more leads.

**Narrow enough to write for**

- Target: head of growth at 30 to 80 person B2B software companies
- Goal: replace manual list building with scored leads
- Message: you just hired your first sales rep, so the list building probably landed on you.

> Write the disqualifiers first. The leads you leave out decide whether the list works.

## Write the sales playbook the agent follows

A playbook is your own sales script written down. It says what you sell, who it's for, the outcome a customer gets, the objections you hear most and the answers that work. Without one, the agent writes plausible sentences that sound like every other outreach message in the inbox.

Here's the test. Hand the playbook to a new hire who has never sold your product. If they could run a call from it, the agent can run a message thread from it.

1. **State the outcome in the customer's words.** Write what changes for them after they buy. 'You stop rebuilding the same list every Monday' beats 'we have an advanced filtering engine.'
2. **List the objections you actually hear.** Three or four is enough: no budget, already using something, no time, not my call. Write the reply you'd give on a call, in two sentences.
3. **Pick a tone and keep it.** Casual peer, direct operator or formal consultant. Put one example message in that tone so the agent has something to copy.
4. **Name the win.** A 15 minute call, a reply with a question, or a forward to whoever owns the budget. The agent needs to know when to stop asking.

```
Product: [what it does in one sentence]
Who it's for: [role] at [company type and size]
Outcome: [what changes for them]
Proof: [one specific thing you can point to]
Objections:
- Too expensive: [two sentence answer]
- Already have a tool: [two sentence answer]
- Not my decision: [two sentence answer]
Tone: casual peer, short sentences, no jargon
Win condition: 15 minute call booked
```

## Write connection requests people accept

A connection request has one job: get accepted. Pitching inside the note gives people a reason to hit ignore, and the note is capped at 200 characters anyway.

The notes that get accepted mention something true and specific about the person. A comment they left, a post they wrote, a change on their profile. If a competitor's post is how you found them, say so, because that's a real reason you're writing.

Send the same note to 200 people and it stops being specific. If you can't name what the person did, they shouldn't be in the campaign yet.

- **you:** Your comment on the pricing thread about list quality matched what three of my clients said last month. Worth connecting.
- **them:** Sure, happy to connect.
- **you:** Thanks for accepting. You mentioned the list goes stale fast. How are you handling it now, rebuilding by hand or something else?
- **them:** Mostly by hand, which is why it's a mess.

## What to send after someone accepts

The first message after acceptance is where most sequences die. People send a paragraph about the company and a calendar link. The lead has no reason to care yet.

Ask about the thing that made them a lead. If they commented on a post about messy lead lists, ask how they're handling theirs now. One question, two sentences, no pitch.

When they answer, give them something they can use. A short observation about how teams their size solve it, a question about the part of the workflow they didn't mention. Then ask for time.

1. **Open with the reason you found them.** One line. The post, the comment, the job change. It shows the message was written for them.
2. **Ask one question about their process.** Something they can answer in a sentence from their phone. 'Are you building lists by hand or with a tool?'
3. **Reply with something useful.** A pattern you've seen at similar companies, in two or three sentences. No deck, no link.
4. **Ask for 15 minutes once they describe the problem.** Then send the scheduling link. Asking before they've named the problem turns a conversation into a pitch.

## Follow up without becoming noise

Most replies land after the second or third touch, which is why follow-ups matter and why each one needs to add something. 'Just bumping this' gives the reader nothing to respond to.

Write each follow-up as a new angle on the same problem: a short observation, a question about a different part of their workflow, a link to something you wrote. Three or four touches over a few weeks is plenty. If someone hasn't answered by then, they aren't a lead right now, and the next message costs you more than it earns.

| | Touch | What it does | Example opener |
| --- | --- | --- | --- |
| Day 1, invite | Gives a reason to accept, no pitch | Your comment on the pricing thread made me think of a client with the same problem. |
| Day 3, first message | Asks about the problem, no link | How are you handling the list building side at the moment? |
| Day 8, follow-up | New angle, short | One thing I see with teams your size: the list goes stale in about a month. |
| Day 15, last touch | Clear close, easy to answer | Should I close the loop here, or is this worth 15 minutes next week? |

## Check the campaign before you turn it on

Ten minutes of reading saves a week of cleanup. Go through these before the first message leaves.

- Read the first five queued messages. If any of them could have gone to anyone on the list, the targeting or the playbook is too loose.
- Check the send window against the lead's time zone. A message landing at 3am reads like automation.
- Decide what happens at the first reply: hand off to you, AI until there's interest, or AI until a meeting is booked. Pick one before launch.
- Look at your pending invitations. If a large share of old invites were never accepted, withdraw them before adding more.

> Read the queued messages before they send. Editing five is faster than repairing a campaign that went out wrong.

## Doing this with Kinlead

Kinlead finds people who match the customer you describe, or your website, and scores each one from 0 to 100 against that profile. Leads under the fit line aren't kept, so the list starts narrow. The [Steal Customers agent](/features/steal-customers) watches competitor pages and optional founder profiles, reads the comments on their recent posts, and turns commenters who look like buyers into leads with the post URL and their comment attached, which gives the AI a real reason to open.

Messages and follow-ups come from the campaign goal, tone and your playbook. Each LinkedIn account sends at most one action every 5 minutes inside the daily limits and send window you set in the lead's time zone. You can read and edit queued messages on the Actions page, and each campaign chooses whether replies hand off to you, get answered by the AI until there's interest, or get answered until a meeting is booked with your scheduling link.

| | What you need | How Kinlead handles it |
| --- | --- | --- |
| Finding leads | Standard LinkedIn search, scored 0 to 100, no Sales Navigator seat required. |
| Replies | Every campaign lands in one inbox sorted by intent, with an email when someone replies. |
| Account pressure | Invites pause on an account LinkedIn pushes back on, and you get an email. |
| Connection | Runs from your own LinkedIn through Unipile; Kinlead never stores your password. |

## Related

- [Book a demo](https://getkinlead.com/demo.md)
- [AI outreach features](https://getkinlead.com/features/ai-linkedin-outreach.md)
- [Steal Customers agent](https://getkinlead.com/features/steal-customers.md)
- [Pricing plans](https://getkinlead.com/pricing.md)

## Common questions

**Do I need Sales Navigator for AI outreach?**

No. Kinlead uses LinkedIn's standard search to find and score leads, so a Sales Navigator seat is not required.

**How do I stop the AI from sounding like a bot?**

Give it a detailed playbook with a named tone and one example message to copy. Then read the first few queued messages and fix anything that could have gone to anyone.

**What happens if LinkedIn warns my account?**

Kinlead pauses invites on that account and emails you, so you can slow down or stop before it becomes a restriction.

**How do I handle objections automatically?**

Put the objections you actually hear in the playbook with the two sentence answers you'd give on a call. The AI uses those instead of inventing a rebuttal.

[Create a Kinlead account](https://getkinlead.com/signup) or [book a call](https://getkinlead.com/demo) to talk through pricing and setup.
