A voice note is the highest-variance touch in outbound. Sent at the right moment it gets replies that no written message in the same sequence gets, because it is unmistakably a person and unmistakably not a template. Sent as a cold opener to someone who has never heard your name, it is a stranger’s voice on a personal device, and it gets blocked.
The difference is not the recording. It is where in the sequence it sits.
The one moment they work best
After an unanswered call. You dialled, they did not pick up, and a voice note arrives referencing that: “I just tried calling, wrong moment, clearly.” It is coherent, it explains itself, and it converts the missed call from an annoyance into context.
Ranked by how well they land:
| Moment | Works? | Why |
|---|---|---|
| After a no-answer call outcome | Best | The note explains the missed call. Nothing else in outbound has this coherence |
| After a reply that stalled | Strong | Tone does what three follow-up emails cannot: removes the pressure |
| Answering a specific question | Strong | Faster to say than to type, and it reads as effort rather than automation |
| After an accepted LinkedIn invite | Mixed | Works if the invite was personalised. Otherwise you are still a stranger |
| Cold first touch | No | An unknown voice on a personal number. This is the version that earns blocks |
Length, and the thirty-second rule
Under 30 seconds. A 90-second voice note is a meeting request disguised as a message, and the visible duration decides whether it is played at all: 0:22 gets tapped, 1:48 does not.
Thirty seconds is roughly 70 words, which forces the structure anyway: who you are (five words), why this person specifically (one sentence, and it has to be true), the ask (one question), stop. No summary, no thanks-for-your-time, no second ask.
What to actually say
After a no-answer: “Hi Priya, Ravi from Outboundry, I just tried calling. I noticed you are hiring three SDRs this quarter, which usually means the infrastructure question comes up. Worth a ten-minute conversation, or is this the wrong quarter for it?”
Reviving a stalled thread: “Hi Tom, Ravi again. You said timing was the issue in March, so I am not chasing, I just wanted to check whether the new budget year changed that. If not, I will leave it.”
Read both out loud. If you cannot say it without sounding like you are reading, rewrite it shorter. Delivery matters more than the words: normal pace, ambient room noise is fine, and a single take with a stumble in it outperforms a polished third take, because the polish is exactly what makes it sound like marketing.
Who tolerates them, and where
Region. Entirely normal in India, Brazil and most of LATAM, and in the Gulf. Distinctly odd in Germany, the Netherlands and the Nordics, where a voice note from an unknown sender reads as an imposition regardless of content. In the US and UK it depends more on the person than the market.
Persona. Founders, sales leaders and agency owners answer voice notes, they already run their working day on them. Engineering, finance and procurement generally do not, and prefer the thing in writing where it can be forwarded. Anyone in a role that measures itself on written records will resent audio.
Seniority matters less than people assume. What matters is whether the person’s job involves talking all day.
Record once, send to the step
The obvious objection is that voice notes do not scale, and one-recording-per-prospect genuinely does not. What does scale is one recording per step: a note that is specific to a segment and a moment rather than to a name. “You are hiring SDRs” is true of everyone in that segment, and it is the specificity that earns the reply, not the first name.
That is the practical version, record once for the segment, send as the step, and keep the genuinely bespoke recordings for named accounts. Two cautions. Never use synthetic voice for this: a cloned voice discovered is worse than no voice note, and it is discovered. And keep one note per prospect per sequence, a second unanswered voice note is where the channel starts to feel like surveillance.
Next: WhatsApp templates for B2B, the 21-day number ramp, or how the four channels fit in one sequence.
Frequently asked questions
Do voice notes work in sales outreach?
Yes, in one specific position: after an unanswered call, or to revive a thread that stalled. As a cold first touch they underperform text and generate blocks, because an unknown voice on a personal device reads as an intrusion.
How long should a sales voice note be?
Under 30 seconds, which is about 70 words. The visible duration decides whether it gets played at all, and anything over a minute usually does not.
What should a sales voice note say?
Who you are in about five words, one true sentence on why you are messaging this person specifically, one question, then stop. No summary and no second ask.
Are voice notes appropriate everywhere?
No. They are normal in India, LATAM and the Gulf, and read as odd in Germany, the Netherlands and the Nordics. Personas whose day is spent talking answer them; finance, procurement and engineering usually prefer writing.
Can voice notes be automated?
One recording per sequence step can be, which is different from one per prospect: the note is specific to a segment and a moment rather than a name. Do not use synthetic voice, and send at most one per prospect per sequence.