Do LinkedIn Voice Notes and Video Messages Lift Reply Rates?

The short answer

I'm Hlib Storchak. I build outbound systems for B2B founders and sales teams, and I've booked 2000+ meetings for B2B clients running cold email and LinkedIn side by side. This question comes up constantly because it has an obvious appeal: a 15 to 30 second voice note or a 40 second selfie video is a genuinely different experience for the person receiving it than another wall of text, and it feels like it should convert better.

The honest answer is that it probably does help, in the right context, but the specific numbers circulating (30 to 40% lifts, "double your response rate", "up to 40% response") come from outreach tool marketing blogs, not from an independent or disclosed-methodology study. I went looking for the primary source behind those figures. There isn't one you can actually check.

Two different features people lump together

"Voice notes" and "video messages" get used interchangeably in most outreach content, but they are two separate LinkedIn features with different friction and different signal. A voice note is a short audio clip recorded and sent inside a LinkedIn DM thread, playable in place without leaving the conversation. A video message is either a native LinkedIn video attachment or, more often in outbound tooling, a link to an externally hosted personalized video (Vidyard, BombBomb-style tools, or a screen recording). Voice notes take the sender seconds to make. Video messages take longer to record, and asking someone to click out to a third-party video host adds a step that a native voice note does not have. Any claim about "video" performance should be read with that extra friction in mind, since a link-out video is a bigger ask of the recipient than an in-thread voice clip.

What outreach vendors are actually claiming

The figures that show up most often across outreach tool blogs: replies "jump by 30 to 40%" when a voice note replaces or follows a text message, voice messages can reach "up to 40%" response versus a quoted 10 to 15% baseline for text, and one growth-tooling vendor puts it more simply, describing its own data as showing that voice messages "can double response rates when used as part of an outreach sequence." A solo-agency case example puts a specific client's lift at 40%, attributed to a boutique recruiting firm.

Read individually, those numbers sound like a research finding. Read together, they are the same handful of claims getting repeated and rounded across different blogs, several citing each other rather than an underlying dataset.

Where those numbers actually come from

I traced the citation chain on the most specific claim I could find, the "30% higher reply rate" figure from a solo-agency growth blog. Its own citations point to an outreach platform's marketing page for the "30% higher engagement" number and a separate agency's blog for the "roughly 40%" figure and an "around 80% of text messages get ignored" baseline. Neither source discloses a sample size, a date range, or whether the comparison controlled for anything besides message format. The one named example in that chain, a recruiting firm's video and voice campaign, is described anecdotally with no campaign metrics, no audience size, and no control group.

That is not evidence the effect is fake. It is evidence that nobody publishing these numbers has shown their work, which means you cannot check whether their audience, industry, or message quality looked anything like yours.

What the one real LinkedIn benchmark says

The most rigorous LinkedIn outreach dataset publicly available in 2026 is Expandi's benchmark report, built from 13,218,869 connection requests, 6,730,447 outbound messages, and 3,766,161 accepted connections across 13,302 active accounts, tracked between May 2025 and April 2026. Source: Expandi's 2026 LinkedIn outreach benchmarks. It puts the platform-wide average message reply rate at 10.4% and connection acceptance at 28.5%.

Here is the part that matters for this question: that dataset's "outbound messages" category lumps together welcome messages, follow-ups, group messages, event messages, InMails, and emails. It does not break out voice notes or video messages as a separate format at all. The single largest, most methodical LinkedIn benchmark in circulation right now simply has nothing to say about whether voice or video outperforms text. Anyone quoting a specific voice-versus-text percentage is not quoting this dataset, because this dataset does not measure that comparison.

The claims, side by side, ranked on rigor

Laying out what I actually found, from most to least verifiable.

Why a real lift is plausible anyway

I would not throw the idea out just because the numbers are unverified. There is a reasonable mechanism behind it. A voice note carries tone, pace, and hesitation that text strips out, and on a channel where the median message is a templated pitch, a genuinely human-sounding clip is unusual enough to earn a second of attention it would not otherwise get. It also signals effort. Recording and sending a 20 second voice note takes longer than firing off a templated line, and recipients pick up on that asymmetry even without consciously noticing it. Neither of those is a controlled finding, they are the same behavioral logic that makes any high-effort, low-volume channel outperform a high-volume, low-effort one, at least until everyone starts doing it and the novelty wears off.

Why almost every good example is warm outreach

Nearly every strong example I found in vendor content involves a recipient who already has some context on the sender, a mutual connection, a recent post they engaged with, a follow-up after a call, not a cold stranger opening a connection request for the first time. That distinction matters more than the format itself. A voice note from a name you do not recognize, on a platform where audio scams and impersonation are a known nuisance, can read as invasive rather than personal. The same clip sent to someone who accepted your connection request last week and saw your name on a shared post lands completely differently. Any lift you have read about almost certainly shrinks, and may reverse, once you apply it to genuinely cold, first-touch outreach to strangers.

How to test it on your own list before you believe a vendor

Since no public dataset answers this cleanly, the only real answer is the one you generate yourself. A simple version that does not require new tooling: split your next outreach batch into two matched groups by industry and seniority, send one group your normal text follow-up and the other a short voice note with the same core message, and track replies over the same seven day window for both. Run it on at least 100 to 150 sends per group before drawing a conclusion, since anything smaller will bounce around on noise alone. If you want to check whether a result is real or just normal variance before you act on it, a two-proportion test on the two reply counts will tell you faster than eyeballing the percentages. The number that comes out of your own list, on your own ICP, is worth more than any vendor's blog post, including this one.

When I actually reach for it with clients

The setup I run for clients treats voice notes as a second or third touch, not an opener. A cold stranger gets a short, clear text message first, since that is the lowest-friction way to establish who you are and why you are reaching out. If they accept the connection or reply with mild interest, that is when a short voice note earns its place, following up on something specific they said or a detail from their profile, because now there is enough context for a human voice to land as warm rather than odd. I have not seen this move a dead list back to life. I have seen it shorten the gap between a lukewarm reply and a booked call, which is the actual place in the funnel where the format's strengths, tone and effort signal, do the most work.

Mistakes that waste the format's advantage

The most common one is using a voice note as a cold opener to a total stranger, where it reads as intrusive rather than personal and gives up the format's one real advantage, context. Second is recording a generic script and sending the identical clip to hundreds of people, which defeats the entire point since the format's value is the appearance of individual attention, not the audio itself. Third is defaulting to a link-out video for a first touch, adding a click and a wait for a stranger who has not yet decided you are worth thirty seconds of their day. Fourth is skipping a written fallback entirely, since some recipients will not or cannot play audio in a given context (an open office, a phone on silent, an accessibility need) and a voice-only message with no text summary loses those replies for free.

FAQ