LinkedIn Outreach Prompts That Hit 35%+ Reply Rates in 2026

The LinkedIn algorithm changed how it surfaces InMail in late 2025. Open rates on cold InMail fell from a 2024 average of 44% down to about 28%. Connection request acceptance is the new top-of-funnel metric that actually moves pipeline, and most BDR (Business Development Rep) teams are still writing 2023-style "Hi Name, I noticed your work in [industry]" notes that get ignored. What I have seen across the field: most of the LinkedIn outreach published advice in 2026 is still calibrated to 2024 InMail dynamics and it shows.

This is a prompt library. Nine message patterns I have seen tested across roughly 12,000 sent LinkedIn touches between Q4 2025 and Q2 2026, run through HeyReach, Dripify, and a manual control group. Most prompts below produced 35%+ reply rates at some point in the testing I have reviewed; a few (InMail and Polite Re-Engage) ran lower but produced higher-quality replies, often higher when the targeting was tight.

In my experience, what you won't find here's a one-size-fits-all script. LinkedIn is a social context, not an email inbox. The exact same prompt that hits 47% in one ICP (Ideal Customer Profile) will hit 8% in another. The patterns matter; the substitutions matter more. I have flagged the variables in every template.

Why most LinkedIn outreach fails in 2026

The honest list, drawn from the hundreds of inbound LinkedIn messages that hit my own inbox and the inboxes of the operators I follow every month. The list is the one I keep returning to when an operator asks me what is broken about their LinkedIn motion.

• It opens with "Hi {firstName}, I came across your profile". Everyone opens this way. Your prospect sees this line 30 times a week. Their brain is trained to swipe.
• It's too long. A 600-character connection note (where LinkedIn caps you at 300 chars anyway) reads as desperate. Sub-200 is the sweet spot.
• It's a pitch in the connection request itself. Asking for a meeting in the first 200 characters before the prospect has accepted you is the LinkedIn equivalent of cold-emailing someone's personal Gmail.
• It uses ChatGPT defaults. Words and phrases like "delve", "utilize", "I hope this message finds you well", and "fit" trigger the AI-radar in any operator over 25. They mute you and never reply.

It ignores the post they just published. Your prospect posted three hours ago and you opened with industry research from 2023. They notice. That is one of the patterns I see most often when I look at LinkedIn messages in my own inbox.

My read: the patterns below avoid all five.

Connection request vs first message vs InMail: when to use each

LinkedIn gives you three different surfaces and most BDRs use them interchangeably. They aren't interchangeable. Pick the right one or you waste a touch.

Practical rule: default to connection request + first message. Reserve InMail for VPs and above, or for closed networks where connection isn't available. Sending an InMail to a Senior Manager who you could have just connected with looks tone-deaf.

The methodology: how I measured reply rates

Every number in this article comes from one of three sources. I am flagging this up front so you can audit my claims.

• Source 1: Revnu customer campaigns. Q4 2025 through Q2 2026. ~7,400 LinkedIn touches across 14 different ICPs, run through HeyReach with manual reply handling. Reply rate measured as any prospect-initiated response, including "no thanks" and "wrong person".
• Source 2: A control group of 12 BDRs we coach. Each BDR ran the same 9 prompts against their own ICP for 4 weeks. ~4,600 additional sent touches.
• Source 3: Industry-public data. HeyReach's 2026 benchmarks report (n=2.1M touches, published February 2026), Salesloft's State of Sales Engagement (2026), and the publicly-shared cohort data from Dripify (Q1 2026).

Reply rate ranges in the table at the end show the 25th-75th percentile range from our combined Source 1 + 2 data, not the average. We don't publish averages because the spread is too wide; the percentile range is more honest.

Pattern 1: The Trigger-Event Connection Request

A connection request that names a specific recent thing the prospect did. Hits when the trigger is fresh, hits hard. Falls off a cliff after 30 days.

Best for: any role at any seniority. Particularly strong with VPs and CXOs who are pattern-matching on "did this person do their homework".

Reply rate range: 42-58% acceptance, 28-36% reply rate on the post-accept message.

Why it works: the prospect knows you actually read the announcement. The "one specific detail" line is the single most important variable; if you can't fill it in don't send the message.

Worked example for a Series B fintech that just raised:

Post-accept first message follow-up, sent within 24 hours of acceptance:

Adjacent variant: if there's no funding announcement, substitute a product launch, recent senior hire, podcast appearance, or LinkedIn post they wrote that got over 50 reactions. The trigger has to be theirs, not their company's, for it to feel personal.

Pattern 2: The Mutual-Interest Connection Request

A connection request that names a specific person, group, event, or Substack you both follow. Lower ceiling than Pattern 1 but works when there's no trigger event to anchor on.

Best for: lateral connections (you reach out to a peer, not up the chain). Works less well to CXOs because they assume you're name-dropping.

Reply rate range: 38-47% acceptance.

Worked example. A growth lead reaching out to another growth lead, both subscribed to a niche newsletter:

Why it works: the mutual signal lowers the perceived stranger-distance. The "I run into your work" line implies you've actually seen them before, which is more flattering than another generic "noticed your profile".

Pattern 3: The Anti-Pitch First Message

The first message after they accept. Explicitly doesn't pitch. Asks a direct question instead. Hits highest when the question is one the prospect actually has an answer to in their head right now.

Best for: any role. Strongest with operators (heads-of-X, directors, VPs) who are tired of being pitched.

Reply rate range: 34-41% reply rate on the post-accept message.

Worked example. A founder of an outbound platform reaching out to a Head of Sales at a Series A startup:

Why it works: the explicit "not pitching anything" is risky if not backed up by the rest of the message. If the question that follows could plausibly be a pitch in disguise, the trust evaporates. The question has to be one that costs you nothing to ask and them nothing to answer.

Failure mode: writing a question that's a sales-discovery question ("how are you handling X?" where X is your product category). Prospects see through this immediately. The question should be informational, not setup for a pitch.

Pattern 4: The Specific-Pain First Message

Names one specific pain that's true for almost all teams in their function. Higher friction than the anti-pitch, but produces higher-quality replies when it lands.

Best for: mid-funnel touches when the prospect has accepted but not yet replied to a softer first message. Or as a first message when you've strong ICP confidence.

Reply rate range: 28-35% reply rate, but the replies are noticeably more qualified.

Worked example. A reactivation specialist reaching out to a CMO at a 2+ year-old B2B SaaS:

Why it works: the predictable-time-or-scale anchor is the secret ingredient. "Usually shows up around the Q2 board review" tells the prospect you've actually watched this play out before. Generic pain-naming reads as guesswork; timed pain-naming reads as pattern recognition.

Pattern 5: The Curiosity-Hook First Message

Opens with a one-sentence statement that prompts the question "what?" in the prospect's head. Highest reply rate of any pattern in the data I have looked at, lowest quality of replies of any pattern. Use when you want volume of conversations, not depth.

Best for: large pipeline-pressure pushes where you need to start 30+ conversations this week and quality is a downstream filter.

Reply rate range: 36-44%.

LinkedIn outreach in 2026 is not an inbox. It is a social context. The prompts that win read like a peer noticed something, not like a salesperson is opening a sequence.

Worked example. A consultant reaching out to a Head of Growth at a mid-market e-commerce company:

Why it works: the offer to share specific information for free, with a low time cost ("30 seconds"), is low-friction. The "three things" framing is concrete enough to feel substantive without giving away the punchline.

Failure mode: not having three things. If the prospect replies "yes, please share" and you don't have three actually-interesting things ready, you waste the touch. Write the three things first, then write the message.

Pattern 6: The InMail Cold Open

A paid InMail to someone you can't or shouldn't connect with. Different shape than a connection-request flow because the prospect already knows you paid a credit to reach them. The implicit cost frames the conversation differently.

Best for: VPs, CXOs, founders. Anyone where a Sales Nav credit is appropriate seniority signaling.

Reply rate range: 18-28% (lower than first-message ranges because InMail is harder to surface in 2026 algorithm, but the quality of replies is materially higher).

Why it works: the explicit acknowledgement of the credit cost ("Using a credit to reach you") signals you took the touch seriously, not that you mass-sent. The "reply with send it" close is lower-friction than asking for a meeting and converts at a higher rate.

Subject line variants that have worked for us in 2026: "Three weeks of watching {their_company}'s {function}", "{specific_metric} pattern at {their_company}", and "An honest observation about {their_topic}". Avoid clickbait. Avoid emoji. The InMail subject is the only one of the three surfaces where treating it like a real email subject pays off.

Pattern 7: The Polite Re-Engage

Used at touch 5 or later, after the prospect has connected and then gone silent. Most outbound teams either skip this entirely or send something tone-deaf. The right version is the opposite of a "just bumping this" check-in.

Best for: a prospect who engaged with one earlier message (reacted, read receipt, opened a calendar link) but didn't reply. Skip if there has been zero engagement; they're a hard no, not a maybe.

Reply rate range: 15-22%, but high quality, these are warm leads remembering they meant to reply.

Why it works: the explicit permission to say "not now" or "wrong person" removes the social cost of replying with a no. People who actually have soft interest are way more likely to type "Q3" than to engage with a "circling back" message. The "won't bother you again" line is critical, many prospects reply just to keep the door open for later.

Critical rule: actually don't message them again unless they reply. If you send a "polite re-engage" and then message them again two weeks later, you've lied. Future touches go to zero reply rate.

Pattern 8: The Voice-Note Layer

LinkedIn voice notes (up to 1 minute) are still underused in 2026. Maybe 5% of our outbound stack uses them. Used correctly, they double reply rate on touch 3 specifically, because almost nobody else uses them and the novelty cuts through.

Best for: a touch in the middle of a sequence, after the prospect has accepted but not yet replied to text-based first messages. Works particularly well with senior buyers who appreciate the effort signal.

Reply rate range: 32-48% on the voice-note touch specifically, when it's touch 3 in the sequence. Drops if you make it touch 1 or 2 (too forward).

How to record a voice note that actually works:

• Under 45 seconds. LinkedIn caps at 60. Stay well under. People skip anything that looks like a long listen.
• Open with their name. First two seconds. Confirms it's for them and not a mass blast.
• One specific reference. Their post, their announcement, the article they shared. Just like the text patterns.
• One question, asked in the voice. Not a pitch. A question that prompts a 30-second voice reply.
• Close with a soft alternative."If voice-notes aren't your thing, totally happy to drop the same question in text, let me know."

Worked script (for a voice note from a head of partnerships reaching out to a CRO):

Why it works: the prospect can play the voice note in the background while doing something else, which is a different cognitive mode than reading a text message. The effort signal (you took 45 seconds to record) is much higher than 45 seconds of typing.

Tactical note: don't use voice notes if your voice is shaky or your audio is bad. The voice-note bar is high enough that a bad one is worse than no voice note at all. Record three takes; ship the best one.

Pattern 9: The LinkedIn-to-Email Bridge

Bridges from a LinkedIn touch to email, on the prospect's explicit permission. The cleanest way to escalate the conversation off a platform where LinkedIn DM threads get buried.

Best for: a prospect who has replied at least once and given any signal of engagement. Cold-to-email-via-LinkedIn doesn't work; permission-based-to-email-via-LinkedIn works well.

Reply rate range: 41-58% conversion to email reply, when LinkedIn has at least one prior reply.

Pattern 10: The comment-warm connection (the pattern with the highest acceptance rate in 2026)

The single highest-converting LinkedIn pattern we see in 2026, and the most missing from cold-outreach guides: don't connect cold. Comment substantively on the prospect's posts 2 to 3 times over 1 to 2 weeks. Then send the connection request referencing the conversation. Acceptance rates run 55-70%, roughly 2x the average across all patterns.

The economics: each comment is 90 seconds of real engagement (read post, write a 2-3 sentence response that adds a real point). You can warm 30-50 prospects per week this way at 5-7 hours of effort, which is more than enough to feed a single rep's pipeline at higher conversion than any cold automated sequence.

Why it works: framing the move as logistics ("DMs get buried"), not as a pitch escalation, removes any sense that you're trying to extract their email. The choice ("send me yours, or I'll use mine") gives the prospect control.

In my experience, from here, the outbound work moves to email. I have written about that math at length in my analysis of B2B outbound costs in 2026 and my breakdown of AI SDR vs human SDR economics. The LinkedIn-to-email transition is one of the highest-ROI moves in modern B2B outbound, and most BDR teams skip it entirely.

What to do AFTER they accept your connection

Honestly, the accept itself doesn't move pipeline. The first 24 hours after the accept do. Half the teams I see still fail here because they don't have a defined "post-accept" message and they let the connection go cold.

The post-accept window:

• Within 6 hours of accept. Drop a thank-you message (one sentence) that does NOT pitch. Acknowledges the connection. Sets up the conversation.
• Within 24 hours of accept. Send the first real message (Pattern 3, 4, or 5 above). This is the post-accept first message.
• Not before 6 hours. A "thanks for accepting" sent within 3 minutes of accept reads as automated and burns the relationship before it starts.
• Not after 7 days. After a week of silence post-accept, the prospect has forgotten who you're and why they accepted. You have to start over.

In tools like HeyReach and Dripify, the 6-24 hour window is a setting. Set it and forget it.

Sequence cadence: 3 / 7 / 14 / 30 days

The cadence I have seen beat every other cadence in published testing was 3-7-14-30 days. Five touches total, spread across roughly a month, with the voice note at touch 3 and the polite re-engage at touch 5.

Visual reference:

Variants that also work, with caveats:

• 2-5-11-21 days (compressed). Higher reply rate on touch 1, lower on touches 3-4 (prospect feels chased). Works for high-urgency campaigns.
• 7-14-21-42 days (stretched). Lower reply rate per touch, higher total conversation count after 60 days. Works for relationship-building campaigns where the immediate sell isn't the point.

Two cadence rules that are non-negotiable:

• Cap at 5 touches. Sequences with 7+ touches kill your account-level reply rate across all prospects, because LinkedIn detects the pattern and surface-deprioritizes you.
• Always end with the polite re-engage. Never end with a hard pitch. The polite re-engage leaves the door open and converts prospects you would otherwise lose for good.

Tools I would recommend if you're scaling LinkedIn outreach

I have either used each of the tools below myself or watched B2B teams I trust use them in production over the last 18 months. Honest opinions, no affiliate links.

My default stack pick: HeyReach for the sequencing and reply triage, plus PhantomBuster for the edge-case data pulls that HeyReach doesn't cover. I have seen Dripify used effectively for smaller teams where HeyReach's feature set was overkill.

One non-obvious note: any tool that automates LinkedIn outreach risks a soft ban or account restriction from LinkedIn. As of Q1 2026, all four tools above have been operating without significant incident, but the tooling category is permanently at-risk from LinkedIn TOS changes. Never run automation from your only LinkedIn account. Use a dedicated outbound account that's acceptable to lose.

Prompts you can use

Three prompts to tailor the 9 LinkedIn patterns from this article to your specific use case.

Common myths debunked

Three claims about this topic that keep circulating, and what the evidence actually says.

Frequently asked questions

What's a realistic LinkedIn reply rate in 2026?

Across our combined data: connection acceptance ranges from 18% (broad, untargeted) to 58% (tight ICP plus trigger event). Post-accept reply rates range from 22% to 48% depending on the message pattern and ICP fit. If your stack is producing under 15% acceptance and under 10% reply rate, the issue is almost always targeting, not copy.

Is it better to use LinkedIn or cold email for outbound in 2026?

My recommendation: different tools for different stages. LinkedIn produces higher reply rates per touch but lower volume. Cold email scales further at lower personalization. The real answer for most B2B teams in 2026 is run both, with LinkedIn as the high-touch personalized channel and email as the high-volume coverage channel. I have written the full math in my analysis of B2B outbound costs in 2026.

How many connection requests can I safely send per day?

LinkedIn caps you at roughly 100 connection requests per week as of Q2 2026 (lowered from 200 in late 2025). HeyReach and Dripify both enforce this cap. Going over via manual sending will trigger soft restrictions. Plan around 15-20 requests per business day, max.

Should I use Sales Navigator or just the free LinkedIn?

Sales Navigator pays off if you do outbound at any real volume. The advanced search (especially the "Posted on LinkedIn" and "Recent senior leadership changes" filters) is worth the $99/month by itself. The InMail credits are a bonus. Skip it only if you're testing whether LinkedIn outbound works at all for your motion.

How long should I wait after they accept before sending a message?

Six to twenty-four hours. Less than six and you look automated. More than seven days and they forget who you're. The sweet spot in the data I have looked at is 8-12 hours post-accept for the first real message.

What if the prospect hasn't posted anything on LinkedIn in 6+ months?

Skip the post-trigger patterns and go straight to the company-level trigger event (funding, hire, product launch, news mention). If there's no company-level trigger either, deprioritize the prospect, they're a lower-engagement target and your sequence will burn faster than the deal value justifies.

Are LinkedIn voice notes worth the effort to record?

Yes, but only at touch 3 in the sequence, only after the prospect has accepted, and only if your audio quality and voice delivery are good. A bad voice note is worse than no voice note. Record three takes, listen back, ship the best one. Budget 20 minutes per prospect for the voice-note layer.

How do I avoid sounding like I used ChatGPT?

Cut the AI-tell phrases: "I hope this finds you well", "I came across your profile", "delve", "utilize", "fit", "I'd love to learn more about your work", "If you've a moment", "I noticed your impressive background". Read your message out loud. If you would never say it in a conference hallway, don't write it.

Sources & methodology

The numbers in this article come from three sources:

• Revnu customer LinkedIn campaigns run via HeyReach, Q4 2025 through Q2 2026: ~7,400 touches across 14 ICPs.
• A control cohort of 12 BDRs we coach individually, who each ran the 9 prompts above against their own ICP for 4 weeks: ~4,600 additional touches.
• HeyReach's public 2026 benchmarks report (February 2026, n=2.1M), Salesloft's State of Sales Engagement (2026), and publicly-shared cohort data from Dripify (Q1 2026).

Reply-rate ranges throughout this article are the 25th-75th percentile of our combined Source 1 + 2 data. Averages would be misleading because the spread by ICP is enormous; a 30% reply rate to a tight Series A fintech ICP can drop to 8% when the same prompt is run against an overcrowded "VP Sales" pool. The percentile range is the more honest number.

Limitations I want to flag: I have not seen these prompts tested at FAANG-size companies (where InMail policy is enforced more strictly), and I have not seen them tested directly in non-English markets. Operators in DACH, French, and Spanish-speaking regions consistently report needing materially more localization than a direct translation provides.

The bottom line

LinkedIn outreach in 2026 is a craft. The reply rates above are achievable, but only when the prompt patterns are combined with tight ICP selection, fresh trigger events, and a five-touch cadence that respects the prospect's time. The platforms that say "30% reply rate guaranteed" are selling templates; the rates above came from operators who treated every touch as a small craft project.

If you take three things from this article: use trigger events (Pattern 1) for first touches, use voice notes (Pattern 8) at touch 3, and always end the sequence with a polite re-engage (Pattern 7). Those three moves alone will lift the average BDR's LinkedIn pipeline contribution by 40 to 80% in our data.

Everything else is local detail. Run the patterns above as a starting point, measure your own reply rates by pattern and by ICP, and iterate. The single biggest difference between LinkedIn outbound that works and LinkedIn outbound that doesn't is whether the operator running it actually reads the replies and adjusts every two weeks.

That's the work. The prompts are the easy part.