Open your own LinkedIn inbox and count the messages from strangers that begin with fake familiarity and end with a calendar link. That inventory is why your prospects don't answer your messages: every DM from someone they don't know gets pattern-matched against the pile.
And yet — this is where referrals land, where warm leads surface, where "been reading your posts, do you have capacity in Q4?" arrives. A DM is less formal than email and less confrontational than a phone call, which makes it the easiest place on the internet for a good conversation to start, provided you've earned the right to be there.
The difference between a message that gets answered and one that gets deleted is rarely the wording. It's everything that happened — or didn't happen — before you hit send.
Key takeaways
- The #1 killer is the pitch-slap: connecting and immediately selling. It doesn't underperform; it actively converts goodwill into distrust.
- Replies are a function of familiarity × relevance × size of ask. Most people optimize the wording and ignore all three.
- Sequence matters: comment before you DM. A message from "the person who's been engaging thoughtfully for weeks" opens; a message from a stranger doesn't.
- First messages should carry zero ask or a micro-ask (an opinion, a reaction) — never a meeting.
- With a prospect who genuinely matters, assume weeks of visible, useful engagement before a DM is welcome. The number of days is less important than the order of events.
Why your messages get ignored (it's not the copy)
A reply is a small act of trust. Before answering, the recipient runs a subconscious check:
Do I know this person? Not "are we connected" — does the name ring warm. If you've been commenting on their posts for three weeks, yes. If you connected yesterday and messaged today, no.
Is this about me or about them? People detect within one line whether a message serves their interests or the sender's quota. "I noticed you're scaling the team — how are you handling onboarding?" is about them. "I help companies like yours…" is about you, wearing their name as a costume.
What does answering cost me? A meeting request from a near-stranger costs 30 minutes plus the awkwardness of a sales conversation. An opinion question costs one typed sentence. Price your ask to the trust you've actually accumulated.
Fail any of the three and the wording can't save you. Pass all three and even a clumsy message gets a warm reply.
The sequence: earn the DM before you send it
For a prospect who matters — someone who could become a serious client — "connect and pitch" is not a shortcut, it's a forfeit. What works is a ladder, and the rungs have to be climbed in order. Don't count days; count whether the previous rung actually held.
Rung one: be visibly present. After they accept, a two-line thank-you with no ask. Then read what they publish and react to it honestly. Nothing here is a move; you're simply becoming a name they've seen more than once, attached to reactions that weren't stupid.
Rung two: be useful in public. Comments that add something they'd have wanted said — an experience that supports their point, an angle they didn't cover. This is where you stop being a name and become a competence. You've climbed this rung when they reply to you.
Rung three: be useful in private. Now a DM is defensible, provided it costs them nothing: a resource that fits a problem they've described out loud. Ungated, not yours, no form. "Saw your post on hiring ops people — this checklist helped a client of mine with the same thing. No agenda, just thought of you."
Rung four: start a conversation. Ask their opinion on something they visibly care about. Opinion questions work because they're cheap to answer, they flatter without flattery, and they open a thread rather than requesting one. If they lean in, a call is a natural next line, not an ambush.
The whole ladder is slower than a pitch and enormously faster than the alternative, which is a burned prospect who now screens your name. Trust with a serious buyer can be built or it can be faked, and faking it is legible from the first sentence.
(Doing this for one prospect is easy. Doing it for forty in parallel is a memory problem — who's on which rung, who mentioned the hiring struggle, who went quiet three weeks ago. That bookkeeping, honestly, is a big part of why we started building Quarz. A spreadsheet works at small scale; what doesn't work is winging it.)
Five LinkedIn messages that get replies
Skeletons to adapt — each assumes prior engagement, each asks for almost nothing:
The post reaction:
Your post on [topic] this morning — the point about [specific detail] is one I haven't seen anyone else make. Curious whether you're seeing [related thing] as well, or is that just my corner of the market?
The no-strings resource:
You mentioned wrestling with [problem] in a recent post. This [resource] helped a client of mine with the same thing — sharing in case it's useful. No agenda.
The congratulations with a question:
Congrats on [new role / milestone]! Curious — does this change how your team approaches [relevant area], or is it steady as she goes?
The opinion ask:
I'm seeing [trend] pop up across my clients and I keep going back and forth on what it means. You're closer to [their vantage point] than almost anyone I follow — what's your read?
The referral thread (when a mutual is real):
[Mutual] suggested I say hello — we've been working on [topic] together and your name came up as the person who'd have the sharpest view on it.
Common DNA: specific, short (under 100 words), zero pressure, and each one is answerable in one sentence. The goal of message one is message two.
The asks that kill threads
Never in a first message: a meeting ("15 minutes next week?"), a demo, a "who's the right person at your company for…", a gated whitepaper, or the fake-casual "quick question" that unfolds into a pitch. Each of these tells the recipient the relationship was a setup — and recovering from that is harder than starting cold.
One more: don't send AI-generated messages unedited. Recipients pattern-match the cadence instantly now, and nothing says "you weren't worth three minutes" like machine-flavored warmth.
Managing the inbox side
Replies create an inbox, and the inbox is a deal source — treat it like one. Check it once or twice daily at fixed times (lunch and end of day work well) rather than living in it. Triage fast: prospects and referrals first, thoughtful networking messages second, cold pitches get a polite one-line decline or silence, spam gets deleted. When a thread turns warm, drop your scheduling link only once they've signaled interest — and move real opportunities into whatever system you track clients in, because inbox threads scroll away and die. A conversation that took four weeks to earn deserves better than being forgotten in week five.
For the upstream half of this system — deciding who deserves the 30-day investment — the answer is a named, deliberately chosen list, not whoever happens to appear in your feed.
FAQ
What should I write in a LinkedIn message to a prospect? Reference something specific they published or did, ask a question answerable in one sentence, and make no bigger ask. A good LinkedIn message reads like the middle of a relationship, not the start of a pitch — which is why the engagement that precedes it matters more than the wording.
How long should a LinkedIn message be? Under 100 words for a first message. Long messages transfer effort from you to the reader — the opposite of what an unearned relationship can afford.
Should I use LinkedIn InMail or regular messages? A message to a first-degree connection you've engaged with beats InMail to a stranger every time. InMail carries a paid-outreach smell; connection-based messaging carries context.
How soon after connecting can I message someone? A no-ask thank-you immediately is fine. A conversational message with any ask: after you've genuinely engaged for two to four weeks. The calendar matters less than the engagement being real.
What reply rate is realistic? Cold pitch-style DMs: low single digits. Messages sent after weeks of visible engagement, with a micro-ask: routinely 40–60%. The work happens before the message, and so does the reply rate.