LinkedIn lead generation: strategy, how to and examples

LinkedIn lead generation is a loop: pick a narrow audience, source people from a signal rather than a title alone, send a connection request, open with one true observation, follow up a few times, answer replies quickly and book a short call. The evidence from four real accounts is consistent on the order of importance. The source of the list moved results most: in our own data intent signal lists replied at 54.6% against 21.7% for Sales Navigator searches, and at ShortVibe 16 of 19 calls came from intent lists while title searches booked 1 in 225.

This guide walks through the loop in the order you build it, with the numbers from four accounts that ran it, including the parts that did not work. It is long, so use the headings.

Step 1: choose an audience you can describe in one line

"Heads of growth at software companies with 20 to 200 staff in the US" is an audience. "Anyone in sales" is not. A narrow audience lets the message be specific and lets you see what works. It also lets the agent or you decide, for each person, whether they belong.

Write three things down: who you want, what problem you solve for them, and what you are asking for. If the ask is a 15 minute call, say so. If it is a reply to a question, say that. The ICP template helps.

Step 2: pick the source of your list

There are four common sources, and the evidence ranks them.

The ShortVibe founder saw the same order from a different business: 16 of 19 discovery calls came from intent lists, and title searches booked 1 in 225 (case study). And the source of the signal matters. In our data, people who engaged with posts by AI SDR vendors and automation creators booked at 2.6% to 3.7% per person messaged, while those who engaged with enterprise sales thought leaders booked under 0.4%.

So: choose a signal tied to the problem you solve. The posts on intent signals worth acting on and companies that just raised funding show how.

Step 3: send the connection request

A connection request with or without a note is the normal first touch. In our own account 68,818 requests were sent and 41% accepted. Notes are optional. The playbook that worked for us used no pitch and no link at this stage. Keep volume inside the limits: Sellinger defaults to 20 invitations a day and 150 a week per account. See invitation limits.

Step 4: open with one true observation

When they accept, the first message states one observation about their work and asks nothing heavy. Within the same month, in our data, observation openers out-replied question openers by 10 to 15 points: 43.5% against 33.5% in February 2026, 45.2% against 33.6% in March, 51.9% against 37.3% in June. Three or four short sentences. No calendar link, no vendor name.

Honest limit: only 5 of 30,000 of our openers mentioned the post that put the person on the list. Mentioning it is an untested idea, so test it. The post on observation openers has examples.

Step 5: follow up, a few times

Follow ups add replies. In our own account the first message alone got 26.5% to reply, and with follow ups the total reached 38.4%. Lamothermic saw 44% of all replies come from follow ups. The cost: negative first replies rose from 0.9% on the intro to 14.7% by the third follow up. So follow up with something new each time, keep the gap near five days, and end with a clean exit line.

Step 6: answer replies, then book

A reply is a door. In the referral network case, 1,751 of the first replies were pleasantries such as "thanks for connecting", and the agent turned 1,035 of them into real conversations by asking how the prospect handles their own clients. When someone asks "what is this?", answer in one line, tie it to their words and offer 15 minutes.

Speed helps but is not the lever. Our reply time fell from about 19 to 2 minutes with no rise in meetings per conversation. Lamothermic's median reply time was 2.3 minutes, and Uberwood's was 18 minutes within the winning conversations. Reply fast enough that nobody waits overnight, and then work on the list and the message.

Four examples, in short

• Sellinger. Our own agents, a year of data. 27,737 messaged, 40% replied, 351 showcase meetings booked through the agents, 33 confirmed held. Meetings per person peaked at 2.4% in November 2025 and drifted below 1% from June 2026. Case study.
• Lamothermic. A foundry president selling to engineers and buyers. 293 of 879 replied (33%), 42 buyer side outcomes such as quote requests, drawings and NDAs. Case study.
• ShortVibe. A solo founder. 935 conversations opened, 395 replied (42%), about half courtesies, 19 discovery calls and 10 documented as held. Case study.
• A referral network. 26 profiles, 9,808 messaged, 44.8% replied, 33.7% with substance, 95 calls booked. Case study.

Different audiences, different offers. Do not average them into a benchmark. Use them to see which parts of the loop show up in all four: a narrow audience, a specific opener, follow ups and a quick answer.

What mistakes cost the most?

• Starting with the message instead of the list. Polishing wording for a list of the wrong people is wasted effort. The list was the largest lever in every account we looked at.
• Pitching in the first message. It turns a conversation into a sales call and the reply into a refusal.
• Counting pleasantries as interest. A reply rate that includes "thanks for connecting" overstates demand. Split real replies from courtesies.
• Counting booked as held. In our own account 351 meetings were booked through the agents and 33 were confirmed held. Track attendance.
• Ignoring the account. Pace, quiet hours and relevant audiences are not decoration. No tool can promise LinkedIn will never restrict an account, so keep to modest volumes and stop early if something looks wrong.

A weekly routine

• Monday: refresh the list from your signal. Remove anyone who does not fit.
• Tuesday: read the first messages queued for the week, or the agent's drafts.
• Wednesday: read replies, hand over the warm ones.
• Thursday: review last week's numbers: accepted, replied with substance, calls booked.
• Friday: change one thing for next week.

A checklist

• [ ] My audience fits in one sentence.
• [ ] My list comes from a signal tied to my offer.
• [ ] I stay inside daily and weekly limits.
• [ ] The first message has one true observation and no pitch.
• [ ] I follow up about every five days with something new.
• [ ] Pleasantries get a question about their work.
• [ ] I count calls held, not only booked.