Connection request playbook for technical buyers

Write LinkedIn connection requests that engineers and developers actually accept by avoiding sales language and focusing on technical utility.

Get started

Short answer

Engineers ignore messages that sound like sales pitches. To get an acceptance, lead with a specific technical observation, a shared problem, or a request for their expertise on a niche topic. Reference a specific library, a known bug, or documentation instead of using adjectives like innovative.

Why technical buyers ignore you

Developers read a lot of bad outreach, and they read code, changelogs and issue threads for a living. That makes them quick at telling a specific message from a generated one. If your note could have gone to any of two thousand people, they file it as noise and move on.

The other thing working against you is status. A senior engineer gets vendor and recruiter requests every week. A general value proposition gives them no reason to answer yours in particular. It asks them to do the work of figuring out why you're relevant.

A technical observation flips that. It shows you looked at something they made, and it gives them a concrete thing to answer. Now the exchange is between two people who use the same tools, which is all a connection note has to achieve.

Sales approach

  • I see you are a Lead Engineer at Acme and I think our tool can help you scale your infrastructure.
  • I would love to connect and learn more about your current challenges with DevOps.
  • I've been following your company's growth and want to share some insights on your tech stack.

Technical approach

  • I saw your comment on the Kubernetes forum about the latency issues with gRPC in multi-cloud setups.
  • I am working on a similar implementation of Rust for data pipelines and wanted to follow your work on GitHub.
  • I read your post about the memory leak in version 2.4 of the X library and wanted to connect.

The three angles that work

There are three reasons a technical person accepts a request from a stranger. Pick the one that matches what you actually have, not the one that sounds smartest.

Peer works when you can honestly say you use the same thing. Problem works when the person has publicly described a frustration. Expert works when they're senior enough that an opinion question reads as a compliment rather than a chore.

The peer angle fails the moment you name a tool you don't use. Engineers ask follow-up questions, and the conversation dies in the first reply. If you can't hold a five-minute conversation about the stack you mentioned, use a different angle.

The peer angle

Angle
Mention a tool or language you actually use.
Logic
I saw you're using Go for your backend; I'm curious if you've tried the new concurrency patterns in 1.21.

The problem angle

Angle
Reference a pain point they've described in public.
Logic
I've been reading about the memory leak in the latest version of X and saw you're using it.

The expert angle

Angle
Ask for an opinion on a choice they make often.
Logic
I'm deciding between X and Y for a high-throughput API and saw your project on GitHub.

Writing the request

The note box is small and the reader decides in a second or two. Spend the space on the reason you're writing, not on who you are. Your profile does the introducing.

Go in a straight line: trigger, connection, request. Don't stack two questions, don't describe your company, and don't ask for a call.

  1. 1

    Find one specific trigger

    A post, a comment, a repo, a talk. Something that exists outside their profile summary.

  2. 2

    Open with it

    The first words in the note are the observation. No long greeting, no line about hoping they're well.

  3. 3

    Say how you relate

    One sentence that ties the trigger to your own work. Same tool, same problem, same open question.

  4. 4

    Ask to connect and stop

    No call, no demo, no attachment. The accept is the whole outcome you're after.

Example

  1. You

    Hi Sarah, I saw your comment on the Postgres thread about indexing large JSONB columns. I'm running into a similar issue with a 1TB dataset and wanted to connect.

  2. Prospect

    Hey! Yeah, that's a nightmare. Happy to connect.

Words to delete immediately

Some words are red flags for developers. They signal a script or a generic sales playbook, and removing them makes you sound like a person typing.

Avoid any language that implies you have a solution to a problem they haven't mentioned. Assume they already have a tool that works and you're discussing the tech.

Avoid these

  • I noticed you are the head of engineering...
  • Our innovative solution can optimize your...
  • I'd love to jump on a quick call to see if we have synergy.

Use these

  • I saw your post about...
  • I'm curious how you handled the...
  • I'm trying to figure out if X is better than Y for...

Remove words like synergy, solution, value-add, optimize and innovative. Never ask for a quick call in the first request.

Where to find a real trigger

Most people look at the profile headline and stop there. The headline says what someone is, not what they're working on. Go where technical people leave evidence of their thinking.

  • Their comments, not just their posts. A reply under someone else's thread is usually more specific than the post itself.
  • GitHub. Open issues, pull requests, or a repo they maintain. Read the last few merged changes before you say anything.
  • Conference talks and meetup recordings. A talk title tells you what someone has thought about for months.
  • Documentation and changelog contributions. A name in a docs commit is an unusual trigger and hard to fake.
  • Job postings on their team. The stack listed in an ad tells you what they're building right now.

What to send after they accept

The accept is not the win. The first message decides whether this turns into a conversation. Keep it on the thing you opened with, because switching to your product here makes the accept look like bait.

Ask one narrow technical question. One. A message with three questions gets an answer to the easiest one, or no answer at all.

Example

  1. You

    Thanks for accepting. We're running Postgres 15 with about 900GB in one JSONB column and queries on nested keys are getting slow. You mentioned a partial index on the extracted path. Did that hold up as the table grew?

  2. Prospect

    It did, but we had to pin the query planner to use it. Otherwise it kept falling back to a seq scan after about a week of writes.

  3. You

    That's useful. Did you hit the same thing with autovacuum on the TOAST table, or was that a separate problem?

Paste this job

Hi {{first_name}}, thanks for connecting. The part of your {{post_or_talk}} about {{specific_detail}} is where I'm stuck.
We run {{your_stack}} at {{your_scale}}, and {{problem}} started showing up {{when}}.
How did you end up handling {{narrow_question}}?

Doing this with Kinlead

Finding the trigger is the part that eats time. The Steal Customers agent watches competitor company pages and optional founder or employee profiles, 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 the person commented, which is usually the technical detail you'd otherwise spend ten minutes hunting for.

Notes are written by AI from the campaign's goal, tone and an optional sales playbook, which is your own script. Connection notes stay under 200 characters and open with why that person became a lead. Queued messages can be read and edited on the Actions page before they send, which is where you fix a wrong library name.

Sends go out at one action every 5 minutes per LinkedIn account, inside daily limits and a send window you set. When LinkedIn pushes back on an account, invites pause on that account and you get an email.

Frequently asked questions

Related

Get started with Kinlead

Prefer markdown? Connection request playbook for technical buyers.md