Connection requests that software engineers accept
How to write LinkedIn connection requests that software engineers accept. Find a technical hook, keep the note short, and skip the sales language.
Get startedShort answer
Software engineers accept connection requests that name something specific about their work: a comment they left, a tool in their stack, or a problem they publicly discussed. Skip the sales language, keep the note short, and don't ask for a meeting in the request itself.
Why engineers ignore most connection requests
Engineers get the same notes everyone else gets, just with different words in them. The note says the sender is impressed by the work and would love to explore synergies. Nobody reads past the first clause, because the first clause already told them this was sent to two hundred people.
What they're actually doing when they look at a request is deciding whether you cost them anything. A note that asks for a call costs them time. A note that asks for nothing and says something true about their work costs them nothing. The second kind gets accepted.
The other thing worth knowing: engineers are unusually good at spotting a template. They spend their working lives reading code written by other people and noticing when something was copied. A note with a placeholder swapped out reads exactly like that.
Sounds like a template
- I'd love to learn more about your business and see how we can help you scale.
- I have a game-changing solution for your current development workflow.
- I'm looking to connect with innovative leaders in the software space.
- Hi there, I came across your profile and was impressed by your background.
Sounds like a person
- Saw your comment on the Postgres locking thread. The bit about advisory locks matched something we hit last month.
- You mentioned migrating off a monolith in a comment on Northwind's post. Curious how that went with the shared schema.
- Your repo for the retry wrapper is the cleanest version of that I've seen. Adding you in case you post more on it.
- Noticed you're hiring for a platform role. I work with teams at that stage and follow what you write about it.
Find a hook before you write anything
A hook is one true, specific thing about that person that you can name in a sentence. It doesn't have to be flattering. It has to be real, and it has to be something you could only know by looking at them.
Most engineers don't post much. That's fine, because you're not looking for posts. You're looking for comments, repositories, job posts, and the tech stack listed in their experience. Any one of those is enough.
Spend two minutes per person. If you can't find a hook in two minutes, skip the person. Sending a generic note to them is worse than not sending one, because a rejected request sits in your pending pool and does nothing for you.
- 1
Open their activity tab and read the comments
Comments are where engineers say what they actually think. A post might be a repost of a company announcement, but the comment underneath it is usually an opinion. Look for the last three or four comments and pick the most specific one.
- 2
Check the stack in their experience section
You're looking for a named language, database, cloud provider or framework. 'Backend engineer' is not a hook. 'Go and Kafka at a payments company' is a hook, because you can say something about it.
- 3
Look for a public repo or a writing link
If there's a GitHub or personal site in the profile, open it. A pinned repository, a README, or a short post gives you something concrete to reference that almost nobody else will have read.
- 4
Check whether they're hiring
A job post on the company page or a 'we're hiring' comment tells you they have a problem right now. That's a legitimate reason to connect and it doesn't require you to pretend you read their code.
- 5
Write the hook as one sentence before you write the note
If the sentence is boring, the note will be boring. 'Works at a fintech' is boring. 'Complained about flaky integration tests on a competitor's post' is not.
Write the note in under 200 characters
Connection request notes on LinkedIn are capped, so brevity isn't a style choice, it's the format. Treat the cap as a gift. It stops you from writing the paragraph where you explain your company.
The shape that works: name the thing you saw, say why it's relevant to you, stop. No question mark needed. No call to action. The accept is the whole goal of this message.
Don't ask for a meeting. A stranger asking for thirty minutes before you've exchanged a single sentence is asking a lot, and engineers in particular get a lot of these. Save the ask for after they reply to something.
If the note would still make sense with a different person's name in it, it isn't a hook. Rewrite it or skip the person.
Paste this job
Three notes you can adapt. Keep each under 200 characters. Comment hook Saw your comment on [Name]'s post about [topic]. We hit the same thing with [tool] last quarter. Adding you in case you write more on it. Stack hook Noticed you're running [specific tool] at [company]. I'm researching how teams handle [problem] in that setup. Would like to follow what you post. Work hook Your [repo / post / talk] on [topic] is the clearest explanation I've found. Connecting so I can keep up with what you're building.
What to do after they accept
An accepted request is not a conversation. Most people accept and then never hear from you again, which is fine, but if you want a reply you have to send something worth replying to.
Send one message, a few days later, that continues the thing you mentioned in the note. Don't restart with an introduction. They already know who you are and why you connected.
Ask a question they can answer in one line from memory. 'How did the migration go?' is answerable. 'Would you be open to a call to discuss your priorities?' is not, because answering it requires a decision about their calendar.
Example
- You
Saw your comment on the Postgres locking thread. We hit the same advisory lock problem last month and ended up wrapping it in a retry. Did you go the same way or did you change the schema?
- Prospect
Changed the schema eventually. The retry just hid it until traffic went up.
- You
That tracks. Was the trigger the write volume or the batch job?
- Prospect
Batch job. We were doing a nightly rebuild that locked half the tables.
- Suggested reply
Makes sense. If it's useful, I can send over the write-up our team did on splitting that rebuild into chunks. No call needed, just the doc.
Check the note before you send it
Run every note through the same short list. It takes ten seconds and it catches the notes you wrote while distracted.
The last item matters most. If you can't say out loud why this specific person got this specific note, the note is a template wearing a costume.
Length
- Check
- Under 200 characters, ideally under 150
- What good looks like
- Cut the second sentence
Specificity
- Check
- Names a tool, a comment, a repo or a role
- What good looks like
- Replace with something you actually saw
Adjectives
- Check
- No innovative, seamless, cutting-edge, robust
- What good looks like
- Delete the adjective, keep the noun
The ask
- Check
- No meeting, no demo, no call
- What good looks like
- Remove it and send the note
Swap test
- Check
- Wouldn't work with another name in it
- What good looks like
- Rewrite or skip the person
Where people get this wrong
The most common failure isn't a bad note, it's a good note sent to the wrong person. Engineers are often not the buyer. If you're selling a tool that a VP of engineering signs off on, the engineer is a useful person to know and a bad person to pitch.
The second failure is mistaking a technical hook for a technical pitch. Naming their stack doesn't earn you the right to explain your product in the same message. It earns you the right to be read.
The third is volume. Sending a hundred notes a week with a swapped first name gets accounts flagged and gets you nothing. A smaller number of notes you actually wrote will outperform it, and it won't put your account at risk.
- Write the hook first, then the note. If the hook is weak, the note will be too.
- One sentence of context is enough. Two is the maximum.
- Never mention your product in the request. It's not what the request is for.
- If they don't accept, leave it. Don't send a follow-up request.
- Keep a note of which hooks got replies so you stop guessing.
Doing this with Kinlead
Kinlead finds the people and keeps the hook attached. The 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 the person commented, which is usually the sentence you need for the note.
You can also describe the customer you want and have Kinlead score each lead from 0 to 100 against that profile, keeping only the ones above the fit line. Connection notes can be AI-written under 200 characters and open with why that person became a lead.
Sending runs from your own LinkedIn account at a human pace, at most one action every 5 minutes, inside the daily limits and send window you set. Queued messages can be read and edited on the Actions page before they go out, and any campaign can be paused. If you'd rather review every reply yourself, set the campaign to hand off at the first reply. Pricing and the trial are on the pricing page.
Frequently asked questions
Related
Get started with Kinlead
Prefer markdown? Connection requests that software engineers accept.md