---
title: "Which DM Openers Actually Get Replies (Data)"
slug: dm-opener-performance-data
author: "Leonardo Maldonado"
category: "DM Closing & Scripts"
articleType: data-research
tags: ["DM openers that get replies","best DM opener","cold DM reply rate","permission-based opener"]
publishedAt: 2026-09-27T09:00:00.000Z
updatedAt: 2026-09-27T09:00:00.000Z
canonical: https://setluca.com/blog/dm-opener-performance-data
---# Which DM Openers Actually Get Replies (Data)

> No public dataset measures Instagram DM openers directly. But the largest cold-outreach studies agree on what earns replies: permission-based, personalized openers that ask instead of pitch. Gong found a permission-based cold-call opener converted at 11.18% versus 2.15% for a throwaway line. Luca applies the same shape to your first reply after a lead messages you, replying automatically in your voice, and you can switch that channel to review if you'd rather read it first.

## Key takeaways

- No public dataset ranks Instagram DM openers, so the best evidence comes from cold-call and cold-email studies covering 350M+ contacts.
- Openers that ask permission beat openers that apologise, by a lot: 11.18% against 2.15% across 300M+ cold calls (Gong Labs, 2024).
- Personalized message bodies got a 32.7% higher response rate than generic ones across 12M emails (Backlinko, 2019).
- Length matters: emails of 50-125 words drew responses above 50%, and those asking 1-3 questions were 50% likelier to get a reply than those asking none (Boomerang, 2016).
- Every figure here comes from another channel, so run your own split test. Detecting a 5% to 10% lift needs roughly 440 sends per version.

---

Most advice about DM openers is opinion. This post collects the DM opener performance data that can actually be verified, with the sample size and the method behind every number. No public study ranks Instagram DM openers by reply rate, so nobody can hand you a clean leaderboard. What does exist is the cold-call and cold-email research: hundreds of millions of contacts, measuring the same thing an opener does. Whether a stranger engages or ignores you.

## What does the DM opener performance data actually measure?

No public dataset ranks Instagram DM openers, so the DM opener performance data worth using comes from a handful of large studies of cold calls and cold emails. Together they cover more than 350 million contacts. Gong Labs scored cold-call openers across 300M+ calls in its 2024 update, the closest proxy anyone has published.

Below is every study this post leans on, with its sample, its method, and the reason it can't be read as an Instagram number.

<table>

<thead>

<tr><th>Study (year)</th><th>Sample</th><th>What it measured</th><th>Method</th><th>Main caveat</th></tr>

</thead>

<tbody>

<tr><td>Gong Labs (2024 update)</td><td>300M+ cold calls</td><td>Success rate of four opener types</td><td>Automated transcript scoring</td><td>Gong never publishes its definition of "success"</td></tr>

<tr><td>Gong talk-ratio research (2025)</td><td>326,000 sales calls</td><td>Rep talk time against deal outcome</td><td>Won-and-lost transcript analysis</td><td>Measured across the whole call, not the opener</td></tr>

<tr><td>Backlinko with Pitchbox (2019)</td><td>12M outreach emails</td><td>Response by personalization, length, follow-up</td><td>Observational, live campaigns</td><td>Link-building outreach, not sales; correlation, not cause</td></tr>

<tr><td>Boomerang (2016)</td><td>40M+ tracked emails</td><td>Response by word count, reading level, questions</td><td>Emails users chose to track</td><td>Largely warm email, so absolute rates sit far above cold</td></tr>

<tr><td>Woodpecker (2026)</td><td>20M+ emails on its platform</td><td>Reply rate by personalization depth and sequence length</td><td>Aggregated platform data</td><td>Self-selected users of one tool, no control group</td></tr>

<tr><td>EZ Texting (2026)</td><td>Consumer survey</td><td>Expected business response time to a text</td><td>Self-reported survey</td><td>A stated expectation, not observed behavior</td></tr>

</tbody>

</table>

Read all of it as direction from a neighbouring channel. The mechanics differ; the decision the reader makes does not. Do they feel pitched, or do they feel like answering? For opener examples built for Instagram specifically, see [Instagram DM openers that get replies](/blog/instagram-dm-openers-that-get-replies).

## Which opener types actually win?

Permission-based and recognition openers win, and the spread is wide. Gong Labs scored more than 300 million cold calls in 2024 and ranked each opener by how often it led to a successful call. The gap between best and worst ran past 5x, and both winners lead with the reader instead of the pitch.

**Chart:** Across 300 million cold calls, a permission-based opener converted at 11.18% versus 2.15% for 'did I catch you at a bad time?' (Source: Gong Labs, 2024 (300M+ cold calls).)

Coaches pick from a wider menu than Gong tested. The table maps each common opener type to the closest measured evidence, so you can see which rest on data and which rest on folklore.

<table>

<thead>

<tr><th>Opener type</th><th>Closest measured evidence</th><th>Source (year)</th><th>DM version</th></tr>

</thead>

<tbody>

<tr><td>Recognition ("heard of us?")</td><td>11.24% cold-call success, the top opener tested</td><td>Gong Labs (2024)</td><td>"Not sure if my name's crossed your feed yet."</td></tr>

<tr><td>Permission ask</td><td>11.18% cold-call success</td><td>Gong Labs (2024)</td><td>"Random question, okay if I ask it?"</td></tr>

<tr><td>Value-first / stated reason</td><td>Naming the reason for the call raised success 2.1x</td><td>Gong (2025)</td><td>"Messaging because you asked about macros in my comments."</td></tr>

<tr><td>Observation about them</td><td>Personalized bodies drew 32.7% higher response; deep personalization ~17% reply vs ~7%</td><td>Backlinko (2019); Woodpecker (2026)</td><td>"Your reel on postpartum programming was the clearest I've seen."</td></tr>

<tr><td>Open question</td><td>Emails asking 1-3 questions were 50% likelier to get a reply than those asking none</td><td>Boomerang (2016)</td><td>"Are you taking 1:1 clients this month, or group only?"</td></tr>

<tr><td>Warm check-in</td><td>7.60% cold-call success</td><td>Gong Labs (2024)</td><td>"How's your week going so far?"</td></tr>

<tr><td>Compliment only</td><td>No published reply-rate data for a compliment with no question attached</td><td>--</td><td>"Love your content!"</td></tr>

<tr><td>Direct ask / pitch</td><td>No cold-open study; on calls, naming price on call one lifted win rates 10%</td><td>Gong (2025)</td><td>"I help coaches scale to six figures. Interested?"</td></tr>

<tr><td>Easy-exit apology</td><td>2.15% cold-call success, last of four</td><td>Gong Labs (2024)</td><td>"Sorry to bother you, bad time?"</td></tr>

</tbody>

</table>

Read down the evidence column and a rule shows up. Every opener with real data behind it gives the reader something before it asks for anything: recognition, an honest heads-up, a reason, or proof you read their work. The compliment-only opener, a line coaching circles often recommend, is the only one in the table with no public evidence at all. The easy-exit line fails because it hands the reader the exit first, and a pitch asks for a decision before there's a reason to make one.

## How do reply rates differ by channel?

Reply rates vary more between channels than between openers, which is why importing a benchmark is risky. A 2.15% cold-call opener and a 3.43% cold-email sequence are not measuring the same event, and neither is measuring a DM.

<table>

<thead>

<tr><th>Channel</th><th>Best published figure</th><th>Source (year, sample)</th></tr>

</thead>

<tbody>

<tr><td>Cold call</td><td>11.24% success for the strongest opener; 2.15% for the weakest</td><td>Gong Labs (2024, 300M+ calls)</td></tr>

<tr><td>Cold outreach email</td><td>8.5% average response across all campaigns</td><td>Backlinko (2019, 12M emails)</td></tr>

<tr><td>Cold sales sequences</td><td>3.43% platform average, down from 5.1% in 2024</td><td>Woodpecker (2026, 20M+ emails)</td></tr>

<tr><td>Tracked business email</td><td>Above 50% at the 50-125 word length</td><td>Boomerang (2016, 40M+ emails)</td></tr>

<tr><td>SMS / text</td><td>No reply rate published; ~70% expect a business reply within an hour</td><td>EZ Texting (2026, consumer survey)</td></tr>

<tr><td>Instagram DM</td><td>Nothing published</td><td>--</td></tr>

</tbody>

</table>

The Woodpecker trend is the one to sit with. Cold reply rates on their platform fell by roughly a third in two years, so any benchmark you read has a shelf life. Reply-rate context for Instagram sits in the [Instagram DM response rate benchmark](/blog/instagram-dm-response-rate-benchmark).

## How much does personalization actually lift replies?

Personalization is the biggest lever anyone has measured, and two separate datasets put it in the same place. Personalized message bodies drew a 32.7% higher response rate than generic ones across 12 million outreach emails (Backlinko), and deeply personalized emails ran near a 17% reply rate against roughly 7% for basic sends (Woodpecker, 2026).

Two different methods landing in the same place is about as much agreement as this evidence base offers. Neither is a controlled experiment. Backlinko says so itself: the study shows a link, not a cause, and people who personalize probably pick better targets too.

In a DM, personalization is cheap. Reference their last post, their offer, or the question they asked in your comments. One true sentence about the person beats a thousand copy-pasted sends. Doing this without sounding stiff is its own skill, covered in [how to sell in the DMs without being salesy](/blog/how-to-sell-in-the-dms-without-being-salesy).

## Does opener length change the reply rate?

Short wins, but there's a floor. Messages between 50 and 125 words drew responses above 50%, while 25-word and 500-word emails both sat near 44% (Boomerang, from more than 40 million tracked emails). Very short is not automatically better.

Two more findings from Boomerang's dataset carry into DMs. Emails written at a third-grade reading level got a 53% response rate against 39% for college-level writing. And emails asking one to three questions were 50% likelier to draw a reply than emails asking none. Ask eight and you do worse than asking three.

Scale Boomerang's findings down for a chat window. A DM lives at maybe 25 to 60 words, so the useful translation is the shape, not the count: plain language, one or two questions, no interrogation. Boomerang's emails were largely warm threads, so the absolute rates mean nothing for cold DMs. The direction of each effect is what transfers.

## A failing opener, rewritten

Here is the same outreach done two ways. Both are illustrative examples, not tested benchmarks.

> Weak: "Hey! I help coaches scale to 6 figures with my proven system. Interested?"

The weak line is mass-sent, pitch-first, and gives no reason to reply. It could land in any inbox on the platform.

> Better: "Hey Jenna, loved your post on programming for postpartum clients. Quick one, are you taking new 1:1 clients this month, or is it group only right now?"

The rewrite names her, points at a real post, sits at plain reading level, and asks exactly one thing. The fix is never a cleverer hook. It's proof you read before you typed.

## How to run your own opener test

Since no DM opener dataset exists, generate one. Split your outreach into two versions that differ in a single element, send them in the same week to the same kind of account, and count replies at 72 hours rather than the same day. Change one variable at a time, or you learn nothing about which change moved the number.

The hard part is volume. The table shows how many sends each version needs before a difference stops being noise, calculated with the standard two-proportion test at 80% power and 95% confidence.

<table>

<thead>

<tr><th>Baseline reply rate</th><th>Lift you want to detect</th><th>Sends needed per version</th></tr>

</thead>

<tbody>

<tr><td>2%</td><td>to 5%</td><td>~600</td></tr>

<tr><td>5%</td><td>to 8%</td><td>~1,080</td></tr>

<tr><td>5%</td><td>to 10%</td><td>~440</td></tr>

<tr><td>10%</td><td>to 15%</td><td>~700</td></tr>

<tr><td>10%</td><td>to 20%</td><td>~200</td></tr>

</tbody>

</table>

Many coaches cannot reach those send counts in a month, and that's the honest answer. Two things help. Test big differences rather than small ones, because a 2x gap needs a fraction of the sample a 20% gap does. And pool across months instead of calling a winner every Friday.

## A worked example: Dani, a strength coach

Dani runs a postpartum strength program and sends about 40 outbound DMs a week. She wants to know whether a permission ask beats an observation opener.

> A: "Hey Priya, quick question about your training, okay if I ask it?"

> B: "Hey Priya, your deadlift form breakdown was the clearest I've seen. Are you programming for yourself right now, or following something?"

She splits 20 and 20 each week, alternating so neither version always goes out on Monday. After four weeks she has 80 sends per version and a baseline near 10%. The table above says a 10-to-15-point move needs roughly 700 per version, so 80 is nowhere close. B is ahead, and the gap sits well inside noise.

Dani does two things with the result. She keeps both versions running instead of declaring a winner, which costs her nothing. And she reruns it as a 10% against 20% question by pushing the versions further apart: B gains a specific reference to the prospect's last post, A stays generic. That test needs about 200 per version, which she reaches in ten weeks. At 40 sends a week you can only detect big differences, so test big differences.

## Which mistakes do coaches make when reading opener data?

**Quoting a cold-call percentage as a DM reply rate.** Gong's 11.18% describes a phone conversation. Nothing survives the trip to Instagram intact except the shape of the opener.

**Treating correlation as cause.** Backlinko flags this itself. Senders who personalize also research and target better, so the 32.7% lift bundles several habits together.

**Calling a winner after 30 sends.** At a 10% baseline, a difference under roughly 200 sends per version is indistinguishable from luck. Many opener "tests" are coin flips reported as findings.

**Changing three things at once.** New hook, new question, new send time, then the numbers move. You now know something worked and nothing about what.

**Copying a winning line verbatim.** "Heard the name tossed around?" works on a call because recognition does the lifting. Sent by a coach with 2,000 followers, it reads as a bluff.

**Assuming a benchmark holds.** Woodpecker's platform average dropped from 5.1% in 2024 to 3.43% in 2026. A two-year-old number may describe a market that no longer exists.

## Which edge cases does the opener data miss?

**They messaged you first.** Every study here measures cold contact. A permission ask sounds strange when they already asked for you. Answer their question, then qualify.

**They already follow you or bought before.** Recognition was the strongest thing Gong measured, and a warm follower already has it. Skip the introduction and name the thing that connects you.

**Your audience is tiny.** Below roughly 200 sends a month, no split test will resolve. Pick the opener the published evidence supports and spend your effort on targeting instead.

**They're in another timezone.** A great opener delivered at 3am local is a great opener read fourteen hours late, so [the best time to send a cold DM](/blog/best-time-to-send-cold-dm) matters more in a two-timezone audience than word choice does.

**Your niche has a script problem.** Where every coach uses the same three lines, the winning opener is the one that doesn't sound like the other three. No dataset can tell you which line your prospects are sick of.

## Why is your reply rate low?

<table>

<thead>

<tr><th>Symptom</th><th>Likely cause</th><th>Fix</th></tr>

</thead>

<tbody>

<tr><td>Openers get read, no replies</td><td>Nothing to answer; the message ends in a statement or a pitch</td><td>End with one low-friction question. Boomerang found 1-3 questions lifted replies 50% over none.</td></tr>

<tr><td>Replies arrive, then the thread dies</td><td>The reply came too late to matter</td><td>Cover the first hour. Nearly 70% of consumers expect a business reply to a text inside an hour (EZ Texting, 2026).</td></tr>

<tr><td>Reply rate drops as volume climbs</td><td>Personalization thinned out as sends went up</td><td>Cut volume until every message carries one specific detail. Deep personalization ran ~17% reply against ~7% (Woodpecker, 2026).</td></tr>

<tr><td>Good replies, no calls booked</td><td>The opener never signals what you do</td><td>Name the reason early. Stating the reason for the call raised success 2.1x in Gong's call data.</td></tr>

<tr><td>Test results flip week to week</td><td>Sample too small to resolve the difference</td><td>Pool weeks until you clear the sample in the table above, or widen the gap between versions.</td></tr>

</tbody>

</table>

## Which opener should you use?

Use a **permission ask** if the prospect has never heard of you and you have nothing specific to reference. It's the highest-scoring opener that works without recognition or research.

Use an **observation opener** if you can name one real, recent thing about them. Two independent datasets put personalization ahead, and it's the only lever you fully control.

Use a **recognition opener** if you have genuine visibility in their niche and the claim would survive a raised eyebrow. If it wouldn't, use the permission ask.

Use a **stated-reason opener** when there's a concrete trigger: they commented, they joined a webinar, a mutual client referred them. Name it in the first sentence.

Do both when the situation allows. An observation plus a light question is the shape the evidence supports across every study here, and it's what the examples in [Instagram DM openers that get replies](/blog/instagram-dm-openers-that-get-replies) are built on.

## How much does timing change the reply math?

Speed decides whether a great opener ever gets read. Nearly 70% of people expect a business to answer a text within the hour (EZ Texting, 2026), and a DM is held to the same clock. Send the perfect line, answer the reply a day later, and the moment has passed. For the timing detail, read [DM response time and speed to lead](/blog/dm-response-time-speed-to-lead).

## How many follow-up openers should you send?

Send at least one, and stop around three. A single follow-up raised replies by 65.8% over one message alone (Backlinko), and campaigns with no follow-up averaged a 4.1% reply rate against 8.3% for sequences of three to five steps (Woodpecker, 2026).

One opener, then silence, leaves replies on the table. The trick is stopping before you nag. Two or three DM touches, each with a fresh angle, tends to be the ceiling before a prospect starts to feel chased. For a cadence you can copy, see [the DM follow-up sequence](/blog/dm-follow-up-sequence).

## One honest limit

Every number here comes from cold calls, cold emails, and texting surveys, not Instagram DMs. There is no published dataset ranking DM openers by reply rate, so anyone quoting an exact "DM opener reply rate" is guessing.

The gaps run deeper than channel. Gong has never published its definition of a successful call, so the 11.18% figure can't be recalculated. Backlinko's data is link-building outreach, not sales, and its authors say it shows correlation rather than cause. Boomerang's emails were largely warm threads between people who already knew each other, which is why its response rates sit five times above any cold benchmark. Woodpecker's numbers come from one tool's self-selected users with no control group. None of that is a flaw in the research. It's a limit on how far you can carry it.

Use the figures as direction, not as a promise. The safe read is the pattern every study agrees on: ask, personalize, and move fast. Then run the test above and trust your own numbers over anyone else's.

## A DM opener built on the data

Here is an illustrative opener that applies the pattern. It is an example, not a tested benchmark.

> Hey Mara, saw your reel on macro tracking for busy moms. Quick one, is your 12-week program still open for September, or is there a waitlist? No pressure either way.

The Mara opener names her, references a specific post, asks a low-friction question, and lands at 35 words in plain language. That is the permission-based, personalized shape the data rewards. Luca doesn't do cold outreach, so it won't send an opener like this for you. What it does is apply the same shape to your first reply once a lead messages you, replying automatically in your own voice. Switch any channel to review if you'd rather read a reply before it goes out. Either way, that personalization is the point the data makes plain: openers and replies win when they sound personally written, not generic. Plans and limits are on [Luca's pricing](/pricing). For the closing sequence after the reply lands, see [how to close clients in the DMs](/blog/how-to-close-clients-in-dms).


## FAQ

### Is there real data on which DM openers get replies?

Not directly. No public study ranks Instagram DM openers by reply rate. The best evidence comes from cold-call and cold-email research measuring the same first step: whether a stranger engages. Gong Labs (2024, 300M+ calls) and Backlinko (2019, 12M emails) both found permission-based, personalized openers beat generic lines by wide margins.

### What is the best-performing opener structure?

Lead with context, acknowledge you are reaching out cold, then ask permission or a low-friction question. In Gong's 2024 analysis of 300M+ cold calls, this permission-based structure converted at 11.18% versus 2.15% for "did I catch you at a bad time?" The shape matters more than the exact words.

### Does personalizing a DM actually help?

Yes, and two datasets agree. Backlinko's 2019 study of 12 million emails found personalized bodies earned a 32.7% higher response rate. Woodpecker's 2026 platform data put deeply personalized emails near a 17% reply rate against roughly 7% for basic sends. Neither study is a controlled experiment.

### How long should a DM opener be?

Short, but not one line. Boomerang's 2016 analysis of 40M+ tracked emails found messages of 50-125 words drew responses above 50%, while 25-word and 500-word emails both sat near 44%. Scaled to a chat window, aim for 25-60 words with one or two questions.

### How many DMs do I need to send to trust a test result?

More than most coaches expect. Using a standard two-proportion test at 80% power and 95% confidence, moving a 5% reply rate to 10% needs roughly 440 sends per version, and 10% to 20% needs about 200. Below 200 sends per version, differences are usually noise.

### Is cold DMing still worth it in 2026?

It can be, if openers are personalized and follow-up is fast. Woodpecker's 2026 data shows cold reply rates fell from 5.1% in 2024 to 3.43%, so the bar has risen. Whether the channel is dead is covered in is cold DMing dead.

### What makes a DM opener fail?

Three things: it reads as mass-sent, it pitches before earning attention, and the reply comes too late. HubSpot Research found in 2018 that 82% of buyers rate an immediate response as important for a sales or marketing question, so a slow reply undercuts even a strong opener.

### Should you follow up if your first DM gets no reply?

Yes. Backlinko found a single follow-up raised replies by 65.8%, and Woodpecker's 2026 data shows sequences of three to five steps averaged 8.3% versus 4.1% with no follow-up. Space two or three DM follow-ups with a fresh angle each, then stop before it feels like chasing.


## Sources

1. [Gong Labs -- The Best Cold Call Openers, Backed by Data From 300M Calls](https://www.gong.io/blog/the-best-and-worst-cold-call-openers-backed-by-data-from-300m-calls)
2. [Gong -- Sales Statistics (win rates, reason for the call, discovery questions)](https://www.gong.io/blog/sales-stats)
3. [Gong -- Talk-to-Listen Ratio (326,000 sales calls analyzed)](https://www.gong.io/blog/talk-to-listen-conversion-ratio)
4. [Backlinko -- Email Outreach Study (12M emails analyzed)](https://backlinko.com/email-outreach-study)
5. [Boomerang -- 7 Tips for Getting More Responses to Your Emails, With Data (40M+ emails)](https://blog.boomerangapp.com/2016/02/7-tips-for-getting-more-responses-to-your-emails-with-data/)
6. [Woodpecker -- Cold Email Statistics (20M+ emails on platform)](https://woodpecker.co/blog/cold-email-statistics/)
7. [EZ Texting -- 2026 Consumer Texting Behavior Report](https://www.eztexting.com/report/2026-consumer-texting-behavior-report)
8. [HubSpot Research -- Consumers Expect an Immediate Response (via HubSpot Sales Blog)](https://blog.hubspot.com/sales/live-chat-go-to-market-flaw)

---

Published by SetLuca, the company behind Luca.