You send 20 messages. You personalize them. You wait. Nothing happens.

The advice to just reach out isn't wrong. What's missing is giving someone a reason to reply.

The good news is that this is something you can design.

Before you send the message

Warm it up first

Find the person, meaningfully engage with something they posted or said publicly, then send the request after. This removes the surprise of "why is this stranger contacting me?"

Received from a CTO:

This is also how I ended up speaking with Aline Lerner, founder of interviewing.io. A few messages later, I had a signed copy of her book sitting on my desk.

Time it to something real

Reach out right when something changes for the person, like a new role or company news. That gives you a specific reason to write instead of a random one.

This became a resume review that turned a ghosted application into a phone screen.

Send a connection request, not an InMail

A connection request doesn't register as spam the way an InMail does, it registers as someone wanting to know you.

A request I sent to an EM at a target company:

This cold message turned a ghosted application into an offer.

Start with why you're reaching out

Does the message answer, even indirectly, why you're reaching out to this person, not just anyone with their title?

Compare:

With:

If you're early career, don't assume you need an impressive achievement. You can create value by asking a thoughtful question and making it easy for someone to help.

This became a mentoring relationship that supported a career transition and the first months in a new company.

Make it easy to reply

A genuine, specific question invites a reply. A request for something makes the other person do the work first.

Clear ask:

This led to a podcast conversation with John Crickett and several other interesting connections.

What kills replies

Asking for too much

I get asked for time, a lot. A 30 minute call to "pick my brain" assumes they're the only one asking, which is far from true. I appreciate the outreach, but I can't do it at that rate.

The same applies to resume review requests that attach a resume and want me to read, assess, and reply. The person has asked me to do the hardest part before we've even had a conversation.

What does work is a message that shows you understand the load on their side and makes it easy to help anyway, usually a specific, answerable question.

Generic outreach

You can feel the formula. This is the same message sent to a few hundred people with a few words swapped.

The same problem appears when people contact the wrong person. I frequently get messages from people who aren't a fit for my team, asking me to evaluate whether they are. They haven't read my profile, because if they had, they'd already know.

Find the right person for what you're looking for, and tell them why they're right. Don't ask if you're a fit. Prove it.

Referral requests from strangers

A referral isn't the start of a professional relationship. It's usually the result of one.

It works because someone is willing to attach their reputation to yours. That's easy when you've worked together or built a relationship over time. It's much harder when you've exchanged 2 LinkedIn messages.

Yes, some people refer strangers, and sometimes it works. But that's the exception, not the system I'd build my job search around.

If you're reaching out cold, your first goal isn't a referral. It's a conversation. The referral becomes much easier if that conversation goes well.

Closing insight

Cold outreach works when the other person can quickly see why this conversation makes sense.

Most people optimise for sending more messages. The better strategy is making each message feel like it was written for exactly one person.

What's next

So far in this series we've covered the inputs: positioning, resumes, networking, and outreach.

Next comes the underlying system: how to run your job search like an engineer.

We'll build a pipeline that tracks every application, interview, recruiter conversation, and follow-up, helping you identify your bottlenecks, capture the information you'll need later, and make decisions from data instead of memory.