LinkedIn drives 70 percent of B2B social replies we track, yet most SDR inboxes sit under 5 percent. If your reply rate is under 8 percent, the script is broken.
It’s not volume. It’s context. You don’t need 200 messages a day, you need 20 messages worth replying to.
Below are the exact cold DM scripts, frameworks, and sequencing rules we use at Growleads across 200+ outbound campaigns. Benchmarks, word-for-word templates, and a testing plan you can run this week.
Why your cold DMs get ignored
Most SDRs make this mistake, they write for themselves, not for the recipient’s moment. You’re competing with internal pings, board decks, and 40 other sellers.
We’ve seen replies triple when messages reference a fresh signal and ask one precise question. Generic intros and fluffy CTAs bury you.
If you fix nothing else, fix context, timing, and the ask. X matters more than Y, context more than copy polish.
- No visible trigger, you lead with a pitch instead of why you’re in their inbox today.
- Bloated intros, 4-6 lines before the point. Aim for 280-420 characters total.
- Vague CTAs, “open to connecting?” invites ghosting. Ask one specific yes-no or multiple-choice.
- Zero credibility. One short proof point, not a paragraph of logos.
- Wrong timing. Week 1 of a new role, day of a launch, or right after a funding round changes what lands.
- Wrong channel stage. Cold DM without warming via comment or connection first usually halves reply rate.
The 5-part cold DM that gets 12-22 percent replies
Here’s the skeleton we use at Growleads when ICP and offer are proven. Keep it tight.
Checklist, 5 parts:
Templates you can paste:
- Signal hook, 1 line, cite something specific they did in the last 14 days.
- Micro-credibility, 1 line with a relevant metric or peer.
- Problem hypothesis, 1 line framed as a guess.
- Micro-offer, 1 line scoped to 10-15 minutes or one artifact.
- Soft CTA, 1 line with a binary choice.
- Template A, recent post signal "Saw your note on hiring 3 more AEs after Q1 ramp. We helped 3 seed-stage sales teams cut ramp by 2 weeks using a peer-call library. Wild guess, onboarding content lives in Notion right now? Worth a 12-minute screen share to show the exact outline we use, or pass and I’ll send the doc?"
- Template B, tech stack change signal "Noticed you moved from Zendesk to Intercom in Feb. We’ve migrated 11 support teams and kept CSAT flat during the switch. Guessing macros are still half-baked while you tune workflows. Want the 7-macro template we deploy on day 1, or prefer I send the Loom?"
- Template C, role-change signal "Congrats on stepping in as VP Sales. We’ve seen first-90-day pipelines jump 18-27 percent when the first enablement sprint hits 3 plays, not 12. If I send the 3-play sprint draft we use with new VPs, would you skim it, or should I leave you be for a month?"
- Template D, competitor peer proof "Your team reminds me of Acme’s motion pre-Series B. We cut their no-show rate from 28 to 11 percent by changing the confirmation DM and first 5 deck slides. If I share the 2-line DM and slide order, would that be useful, or not a focus?"
| Component | Word Count Target | Goal | Bad Version | Good Version |
|---|---|---|---|---|
| Signal hook | 8-14 words | Prove relevance | Saw your profile | Listened to your RevOps AMA on forecasting last week |
| Micro-cred | 6-12 words | Earn attention | We’re trusted by leaders | Helped 4 PLG teams lift PQL to SQL 19 percent |
| Problem guess | 8-14 words | Show insight | You have pipeline issues | Bet SDRs chase MQLs that never self-serve trial |
| Micro-offer | 8-14 words | Lower risk | Let’s meet 30 minutes | 12-minute teardown or one page you can paste |
| Soft CTA | 6-12 words | Make reply easy | Thoughts? | Want it, or should I drop a Loom? |
Connection requests and first-touch scripts that stack the odds
Blank connection requests sit at 15-20 percent. Personalised reach hits 50-65 percent when the note references a fresh signal and a reason to connect.
We use this at Growleads when we need fast pipeline from LinkedIn without InMail spend. Don’t sell in the note. Earn the right to DM after acceptance.
- Connection note, comment-first"Caught your take on usage-based pricing in the FinOps thread. Smart point on ACV cliffs. Would love to compare notes with you and 2 RevOps leads we help who measure margin by cohort."Day 0 after accept"Appreciate the connect, [Name]. You wrote that tracking gross margin by cohort is messy. We built a 1-sheet model that plugs into HubSpot for a seed-stage CFO last month. If I send the sheet with dummy data, useful or noise?"Day 2 nudge if no reply "No stress if timing is off. Want the model, or should I send a 3-minute Loom of how it works so you can judge fast?"
| Touch | Expected Acceptance | Expected Reply | When To Use | Notes |
|---|---|---|---|---|
| Blank request | 15-20 percent | N/A | Volume mapping | No note, accept lower acceptance for speed |
| Signal note | 50-65 percent | N/A | Priority accounts | Reference post, podcast, launch, hire |
| Post-accept DM 1 | N/A | 10-18 percent | Within 6 hours | Short hook, micro-offer, binary CTA |
| DM 2 with asset | N/A | 6-12 percent | Day 2-3 | Offer Loom or doc, keep decision easy |
| DM 3 bump | N/A | 3-7 percent | Day 6-7 | Alternate CTA, multiple-choice question |
Signal-based threads: comment first, DM second
If your reply rate is stuck, start the conversation in public. Commenting with substance warms the DM and earns context you can cite.
We’ve seen threads first, DMs second, double reply rates on enterprise personas who ignore cold InMails.
- Like with intent, then leave a 1-2 sentence comment that adds data, not praise.
- Wait 30-90 minutes, reply to someone else in the thread with a second data point.
- Send the DM referencing your comment and the author’s response.
- Offer one asset that extends the thread, spreadsheet, checklist, snippet.
- Ask a binary question to close the loop.
| Signal | Public Comment Example | Private DM Follow-up | Why It Works |
|---|---|---|---|
| Hiring post | "Curious if you index onboarding on calls listened per week or deals touched. We saw a 1.8x ramp when AEs hit 12 calls by week 2." | "Happy to send the 2-week call library outline if you want a head start. Want it, or are you set?" | Shows value where they are already focused |
| Product launch | "Congrats on Launch. What surprised you in beta retention by persona? We saw ops users out-stay ICs 1.3x in a similar tool." | "If useful, I can share a 5-row retention cut by role we track in Sheets. Want the template?" | Adds data, not a pitch |
| Podcast clip | "Loved the take on MEDDICC. We tested dropping MED, kept DIC, closed rate rose from 23 to 27 percent." | "If you want, I’ll send the 1-page MEDDICC-lite questions we use. Want it?" | Peers plus a micro-offer |
Follow-up that earns replies without nagging
Most SDRs either never follow up or spam daily. Neither works. Space your nudges, vary the angle, and keep the ask tiny.
Sequence we use when connection is accepted and DM1 goes out:
- Day 0, DM1, core script with signal hook, micro-cred, problem guess, micro-offer, soft CTA.
- Day 2, DM2, offer asset swap, “send the 1-pager or Loom?”
- Day 6, DM3, multiple-choice, “Is X or Y closer to your current focus?”
- Day 12, DM4, fresh signal, “Saw your team hired 2 CSMs, still relevant or pause?”
- Day 20, DM5, opt-out plus long-term value, “I can drop you on a quarterly list with one practical teardown, yes or skip?”
- Day 35, DM6, close the loop, "Parking this, happy to circle back in Q3 unless you point me at someone else."
| Follow-up | Purpose | Reply Target | CTA Style | Example CTA |
|---|---|---|---|---|
| DM2 | Lower friction | 6-12 percent | Asset choice | Doc or Loom? |
| DM3 | Clarify need | 3-7 percent | Multiple-choice | Is trial-to-paid or expansion the blocker? |
| DM4 | Show you track signals | 3-6 percent | Reframe around event | Still useful after the CSM hires? |
| DM5 | Respect opt-out | 2-4 percent | Opt-down | Quarterly teardown, yes or skip? |
| DM6 | Exit cleanly | 1-3 percent | Close loop | Circle back in Q3 or someone else better? |
Testing, metrics, and the 20 percent reply playbook
Without this, you’re guessing. Instrument your outreach and run simple tests. If a variant doesn’t move replies by 3-5 points over 150 sends, drop it.
We’ve seen teams jump from 4 percent to 14 percent replies in 21 days by changing only the hook and the CTA format. It’s not a new tool. It’s tighter experiments.
Want a faster path? Our LinkedIn Profile Optimizer audit takes about 3 minutes and returns a prioritised fix list built for founders and SDR teams. The full CTA is at the end of this post.
- Track by persona, signal type, and CTA type, not just total replies.
- Test one variable per 150-250 sends, keep baselines stable.
- Keep messages under 420 characters, replies fall off above 500.
- Remove friction, no calendar link in message 1, offer a doc or Loom first.
- Measure positive replies, not just any reply. “Not now” is a data point, not a win.
| Variable | Test Range | Expected Impact | Sample Size | Cut Rule |
|---|---|---|---|---|
| Hook source | No signal vs post vs hire vs tech switch | ±3-10 points | 200 per variant | Kill anything under 6 percent |
| CTA type | Yes-no vs multi-choice vs open | ±2-6 points | 150 per variant | Kill open-ended if under 5 percent |
| Asset offer | Doc vs Loom vs snippet | ±1-4 points | 150 per variant | Keep winner only |
| Length | 220-520 chars | ±2-5 points | 200 per variant | Cap at 420 chars |
| Cred line | Metric vs logo vs peer name | ±2-5 points | 150 per variant | Use the one that boosts 30-day meetings set |
Mini cases and scripts to steal
Real numbers, stripped of PII. Use them as-is or adapt by swapping the signal and the asset.
We use this at Growleads across seed to Series C motions. That’s not a theory. It’s what we see consistently.
- Dev tools to VP Eng, post-launch signal, 19 percent replies "Congrats on rolling out Feature Flags v2. We helped 2 teams cut deploy rollbacks 31 percent by templating canary checks in CI. If useful, I can send the 5-check step that caught 7 regressions last month. Want it, or should I send a 2-minute Loom?"
- Fintech CFO, new hire signal, 16 percent replies "Welcome aboard. We’ve seen first-45-day closes jump when finance owns pricing experiments. Guessing you’re wrangling Stripe exports and NetSuite mappings right now. Want a 1-sheet mapping we used at AlphaCo to clear reconciles in 3 days?"
- HR tech to Head of People, hiring surge, 14 percent replies "Saw 12 new reqs go live this week. We cut time-to-slate by 5 days for 3 teams using a backchannel list of niche boards. Want the list sorted by role and geo, or not a focus?"
- RevOps to CRO, tool sprawl, 22 percent replies "Noticed 11 tools in your stack across GTM. We trimmed touchpoints by 3 tools at BetaCo and lifted stage 2 to close 4 points. If I send the de-dupe workflow and the before-after funnel, would that help, or should I hold off?"
Common mistakes
The patterns we see most when auditing profiles and outbound:
- Pitching in the connection note instead of earning the right to DM post-accept.
- Asking for 30 minutes in message 1. Offer a doc or Loom first.
- Writing 4-6 lines before the point. Keep it under 420 characters.
- No specific signal. If you can’t name the trigger, don’t send it.
- Using open-ended CTAs like “thoughts?” instead of a binary choice.
What to do next
If you want to run this against your own LinkedIn profile, the LinkedIn Profile Optimizer audit takes about 3 minutes and gives you a prioritised fix list.
Frequently asked questions
What is a good cold DM reply rate for SDRs?
Under 5 percent means fix the script or targeting. 8-12 percent is workable. 12-22 percent is strong for LinkedIn DMs with signal-based hooks. Track positive replies separately from total replies.
How long should a LinkedIn cold DM be?
280-420 characters tends to win. Above 500 characters, reply rates drop. Aim for 5 short lines max, one idea per line, no walls of text.
Should I include a calendar link in my first DM?
No. It adds friction before value is proven. Offer a low-friction asset first, a 1-pager or Loom. Drop a link only after interest or on follow-up 2-3.
Is it better to connect first or send InMail?
Connect first when you can. Personalised connection plus DM outperforms InMail on cost and reply rates for most ICPs. Use InMail as a supplement for hard-to-reach titles.
How many follow-ups should I send on LinkedIn?
Three to six over 35 days is a solid range. Vary the angle and offer a new asset or a multiple-choice question each time. Stop when you get a clear no or opt-out.
What CTA works best in a cold DM?
Binary choices win, “Want the 1-pager, or a 2-minute Loom?” Multiple-choice can also work, “Is onboarding or expansion your priority right now?” Avoid vague asks like “open to chat?”
Do images or GIFs improve reply rates in LinkedIn DMs?
Usually no. They can look salesy and get clipped on mobile. A clear text hook plus an optional Loom link after interest performs better in most tests.
When is the best time to send a LinkedIn DM?
Weekdays 8-10am or 3-5pm local time perform well. But the signal matters more than the clock. A relevant DM tied to a fresh event beats perfect timing by a wide margin.
How do I personalize at scale without wasting 10 minutes per lead?
Batch by signal. Pick one trigger per day, like new role or tech change, and templatize the hook line. You can personalize in 60-90 seconds while keeping the rest of the script fixed.
How should I handle a 'not interested' reply?
Thank them, ask a one-click preference, and opt-down. "All good, want me to park this for Q3 or send one teardown per quarter?" Respect earns future opens and referrals.
Can I reuse the same script across industries?
Keep the framework, swap the signal, proof point, and asset. If replies stall under 6 percent after 150 sends, your hook or CTA needs a rewrite for that vertical.