# Checklist for launching your first AI SDR campaign
> Set up an AI SDR for LinkedIn by connecting your account, defining a strict ICP, writing a sales playbook, and choosing the right reply mode.
- HTML: https://getkinlead.com/ai-sdr-onboarding-checklist
- Markdown: https://getkinlead.com/ai-sdr-onboarding-checklist.md

**Short answer:** To set up an AI SDR, connect your LinkedIn account through a provider that never holds your password, write a tight ideal customer profile with a fit line, and put your sales playbook in writing before launch. Then choose whether the AI books meetings itself or hands the conversation to you at the first reply.

## Connect and protect your account

Connect the account before you write a single message. Use a provider that holds the LinkedIn session on your behalf so nobody types your password into a form. If a tool asks for your LinkedIn password directly, close the tab. The [outreach safety checklist](/ai-outreach-safety-checklist) covers the other access questions worth asking before you sign up.

Then set the pace. LinkedIn does not publish exact invite or message limits, which means anyone quoting a number is quoting a guess about their own account, not yours. Start below what you think you can send, watch what comes back, and raise it slowly over a few weeks.

Look at your pending invitations once a week. Requests nobody accepts sit in the pool and make the account look like it is broadcasting rather than talking. Withdraw the ones older than a couple of weeks.

1. **Connect through a provider.** Authorise LinkedIn through a hosted session and confirm the tool never asks for your password. Check that you can disconnect the account from settings in one click.
2. **Set daily limits.** Pick a number of invites and messages you could plausibly send by hand on a normal working day. Cut it in half for a new account.
3. **Choose a send window.** Set the hours in the lead's time zone. Business hours suit most sellers. Extended hours help when your buyers are several time zones away.
4. **Leave gaps between actions.** One action every few minutes looks like a person working through a list. Fifty in a minute does not.

> LinkedIn does not publish exact invite limits. Work inside whatever your account allows: start low, watch the pending pool, raise volume slowly.

## Define your ideal customer profile

A vague profile is the main reason AI outreach reads like spam. Tell the tool to find marketing managers in the US and you get thousands of people with no reason to answer.

Describe your last five good customers the way you would to a new hire: job title, seniority, company size, industry, and one signal that the problem is live this week. That signal can be a job posting, a funding round, a comment on a competitor's page, or their own post about the thing you fix.

Then set a fit line. Score every lead against the profile and drop everyone below it, even when the job title matches. A good scoring pass produces a shortlist you would be happy to message by hand. If the list is four thousand people long, the profile is still too loose. [Setting the fit line](/ai-lead-scoring-fit-line-setup) is the step most teams skip, and it is the one that decides whether the campaign is worth running.

**Too broad**

- Target: marketing managers
- Location: United States
- Industry: any B2B

**Strict and specific**

- Target: head of demand generation
- Company: B2B SaaS, 11 to 50 employees
- Trigger: posted about lead quality in the last 30 days

## Build the sales playbook the AI writes from

The playbook tells the AI what you sell, who buys it, and how you talk. Without one it writes the average of every LinkedIn message in its training data, which is the message your buyer already has three of in their inbox.

Keep it under a page: the goal of a conversation, the tone, the three objections you hear most with your answer to each, and one piece of proof. Long playbooks get skimmed by the model the same way they get skimmed by a new rep. The [sales playbook guide](/ai-sales-playbook-for-linkedin-outreach) has a longer version if you want one.

Write the first message last. Once the playbook exists, ask for a 60 word opener that names why this person became a lead and ends with one question. If the opener would still work with the name swapped, it isn't finished.

| | Playbook part | What to write | Example |
| --- | --- | --- | --- |
| Goal | The outcome that ends the conversation well | A 15 minute call booked this week |
| Tone | How you sound when you are at your best | Short sentences, no adjectives, one question per message |
| Objections | The pushback you hear most and your answer | "Send me some info" leads to "Happy to. What would make it useful, pricing or a look at how it works?" |
| Proof | Something a stranger can check | A named customer, a public case study, or a number you can explain |

```
Write my LinkedIn sales playbook.

What we sell: [one sentence]
Who buys it: [job title, company size, industry]
The pain they feel this week: [one sentence]
What a good conversation ends with: [a booked call, a reply, an intro]
Tone: [peer to peer, short sentences, one question per message]

Objections and my answers:
1. [objection] -> [answer]
2. [objection] -> [answer]
3. [objection] -> [answer]

Proof I can use: [customer name, number I can explain, short story]

Then write a 60 word opening message and two follow-ups. Open with the specific reason this person became a lead.
```

## Choose who answers the first reply

Decide before launch who talks to a person who replies. This setting decides how much of your day the campaign takes, and it is the one most people change after the first week.

Hand-off suits high ticket deals and small lists. The AI writes the opener and stops the moment someone answers, then you take it. It is also the only way to read every reply before anything else goes out, so pick it if you want that control.

Interest mode lets the AI keep the conversation going until the person asks something that sounds like buying, then hands over. Booking mode goes further and sends your scheduling link. Both write in your voice without you reading first, so spend the first week reading transcripts and correcting the playbook where it drifts. The [hand-off workflow](/ai-sdr-hand-off-workflow) is worth reading even if you pick one of the other two.

- **draft:** Sarah, your comment on the Acme thread about lead scoring matched something we see in agency pipelines. Sending a request so I can follow what you post.
- **them:** Thanks. We're trying to fix that now actually. What do you do?
- **you:** We score every lead against your ICP before anyone messages them, so your reps only see people who fit. Ten minutes Thursday and I can show you the rules we'd use for an agency your size. Does 2pm work?

## Write connection notes that earn the accept

The note is the campaign in miniature. It has one job: give this person a reason to accept that would not apply to anyone else.

Generic compliments fail because they prove you didn't read the post. Name the thing they wrote or did, connect it to what you work on in the same sentence, and stop. The ask is the connection, nothing more. [Connection request templates](/ai-linkedin-connection-request-templates) help when you're staring at a blank box at 11pm.

**Reads like a template**

- Hi Sarah, I came across your profile and was impressed by your background. Would love to connect.
- Hey! Love your posts about marketing. Let's connect and explore synergies.
- Hi, I help companies grow. Open to a chat?

**Sounds like you read the post**

- Sarah, your comment on the Acme thread about lead scoring matched what we see in agency pipelines. Connecting to follow your posts.
- You hired two demand gen people last month and the role is still open. We fix the lead quality problem that usually causes that.
- Saw you took the Northwind team from agency to in-house. Curious how the handover went. Connecting in case comparing notes helps.

> If the note would work for five hundred people with the name swapped, it isn't personal.

## Read the queue before the first send

Treat the first week as reading week. Open the queue of scheduled messages and read the first twenty openers and the first twenty follow-ups before anything leaves your account.

You are checking two things: does each message name a reason this specific person is in the campaign, and would you send it yourself with your name on it.

1. **Read twenty openers.** Flag every one you wouldn't send. Two or three is a playbook problem. Ten is a targeting problem.
2. **Read the follow-ups.** Most conversations die on message two. The follow-up should add a different angle or a short example instead of repeating the ask.
3. **Fix and restart.** Pause the campaign, rewrite the playbook or tighten the profile, then relaunch to a small batch. Volume does not fix a bad template.

> A campaign you would be embarrassed to read is a campaign that gets you reported.

## Doing this with Kinlead

Kinlead covers the setup above in one place. Describe the customer you want or point it at your website, and it finds matching people on LinkedIn and scores each one from 0 to 100. Leads below the fit line are not kept.

Leads live in lead groups, and any campaign can point at any group. You can export a group to CSV or import a CSV of LinkedIn profile URLs. The [Steal Customers](/features/steal-customers) agent watches competitor company pages and any founder or employee profiles you add, reads the comments on their recent posts, and turns commenters who look like buyers into leads. Each lead keeps the post URL, what the post was about and what they commented, which gives the opener something real to say.

Outreach runs from your own LinkedIn account through Unipile, and Kinlead never stores your password. Each account sends at most one action every 5 minutes, inside daily invite and message limits you set and inside a send window in the lead's time zone. Connection notes are written by AI and stay under 200 characters. Queued messages can be read and edited before they send, and any campaign can be paused. If LinkedIn pushes back on an account, Kinlead pauses invites there and emails you.

Pro costs $49 a month after a 7 day free trial, and there is a minimum booking guarantee. The [terms](/minimum-booking-guarantee) explain when a week can be claimed.

1. **Set the goal, tone and playbook.** Paste your playbook into the campaign. The AI writes openers and follow-ups from it.
2. **Pick a reply mode.** Hand off at the first reply, let the AI reply until there is real interest, or let it book meetings with your scheduling link.
3. **Work the inbox.** Replies from every campaign land in one inbox, sorted by intent with interested people first. You get an email when someone replies.

> Extra LinkedIn accounts cost $40 a month each on Pro, less per account in larger numbers. Teammates join as seats and use their own login inside shared workspaces.

## Related

- [View pricing plans](https://getkinlead.com/pricing.md)
- [Book a demo](https://getkinlead.com/demo.md)
- [Read the booking guarantee](https://getkinlead.com/minimum-booking-guarantee.md)

## Common questions

**Do I need Sales Navigator for an AI SDR tool?**

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

**How do I stay inside LinkedIn's invite limits?**

LinkedIn doesn't publish exact limits, so set a conservative daily number you could match by hand and raise it slowly. Withdraw stale pending invitations so the pool stays healthy.

**Can I use an AI SDR to send cold email?**

No. This setup is for LinkedIn outreach. Leads come without email addresses or phone numbers, and nothing goes out by email.

**What happens if my LinkedIn account gets restricted?**

A tool worth using should pause outgoing actions on that account and email you the same day, so you can check the account before anything else sends.

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