Sales: how many client voice notes did you save for later today?
It is 3.40pm. You come out of a forty-minute call and WhatsApp Web has eleven conversations in bold. Three of them are voice notes. The first runs 4 min 12 s and comes from the client who asked for a quote on Monday. Somewhere inside it sits the payment term he will actually accept. Playing it now is out of the question, because the next call starts in five minutes and your headset is already busy. So you do what everyone does: mark the chat unread and promise to come back.
You come back at 7pm. Or tomorrow, over coffee. And when you do, you play all 4 min 12 s from the start, because there is no way of knowing which minute holds the number.
This is the part of a sales day that never reaches a report. It is not disorganisation. The information simply arrived in a format nobody can read, nobody can search and nobody can paste into a CRM.
The deal now lives inside the messaging app
In Brazil this is measured, not assumed. In the TIC Empresas 2025 survey, run by Cetic.br, 68% of companies said they sell through some online channel. Among those that sold over the internet, 79% used WhatsApp or Telegram to carry the transaction. Fieldwork ran from February 2025 to January 2026, with 4,174 companies interviewed by phone.
And the line keeps rising: among online sellers, WhatsApp and Telegram use went from 74% in 2024 to 79% in 2025 in the same survey. Which means the channel where the order lands is the same channel where the question arrives, where the deadline is negotiated, and where the client sends four minutes of voice explaining why he needs another thirty days.
The same study found 17% of companies using artificial intelligence in 2025, up from 13% in 2024, and 15% among small companies. Put those two numbers side by side: almost everyone sells by message, and almost nobody has a machine reading those messages.
Clients would rather talk
On the other side of the negotiation the preference is not subtle. According to the PNAD Contínua household survey run by IBGE, the Brazilian official statistics agency, 90.5% of people aged 10 or over used the internet in 2025. Among those users, the most common purpose was voice or video calls, at 95.3%. Second came sending or receiving text, voice or image messages through apps, at 90.2%.
Voice ahead of everything. Nobody selling in Brazil chooses to receive voice notes. They arrive.
An illustrative sum (redo it with your own numbers)
Take one of your weeks. Say thirty client voice notes, averaging 1 min 40 s. That is 50 minutes of listening, and that is the good scenario, the one where you play each note once. In practice the voice note with the payment terms gets played twice: once to understand it, once to write the number down correctly. Replay a third of them and you are past an hour a week of pure audio playback, roughly four hours a month. The figures above are invented to explain the flow. Swap them for your own pipeline, which is where the sum starts to hurt.
What exists today, and where each route stops
Five ways out, none of them clean. The newest belongs to WhatsApp itself: on 21 November 2024 it introduced voice message transcripts, generated on the device, switched on under Settings and then Chats, with a long press on a message to transcribe it. It handles one voice note at a time, nicely. On the phone.
| Route | What it solves | Where it stops |
|---|---|---|
| Playing it on the phone at 1.5x speed | You find out what the client said | No text is left behind. Nothing to search tomorrow, nothing to paste into the CRM |
| WhatsApp native transcripts | Text of the voice note, generated on the device | One note at a time, by long press, and on the phone. The CRM is on the computer |
| Forwarding the audio to a bot | Comes back as text | Takes the audio out of the chat, adds a step, and the client waits for your reply |
| Downloading the file and uploading it to a transcription site | Faithful text, punctuation included | Four clicks per note, and the client file parked in your downloads folder |
| An extension inside WhatsApp Web | The text appears under the voice note, in the chat already open | Needs the computer on, and comes with a daily quota |
What separates the first four from the last one is where the work happens. No rep negotiates on a phone with the CRM open on their lap. The WhatsApp Web tab stays open all day, next to the spreadsheet and the pipeline, and that is where the text has to show up.
How it works in ZapVox, step by step, in a normal week
- Install the extension from the Chrome Web Store and open WhatsApp Web in that tab you already keep open from 9 to 6.
- Turn on Automatic transcription. A new voice note arrives and the text appears under the message, with no click. This is the point where the queue stops being a pile of sound and becomes a pile of text you can scan.
- For accounts in negotiation, open Per-chat settings and leave Bullet-point summary on. Four minutes of client become five lines, and the payment term is in one of them.
- Use Sentiment reading to order the queue. The note from someone in a hurry, or annoyed, moves up the list. On a Tuesday with eleven chats in bold, that single decision pays for itself.
- Copy with 1 click and the text goes into the CRM, the order or the billing email with the client's own wording intact, the way they said it.
- Find it later with Search inside voice messages. Ctrl+K, or Cmd+K on a Mac, type instalments and the passage comes up grouped by person. This is for the sentence you are sure the client said back in August.
- Export the chat transcriptions as a .txt when the deal closes and the whole negotiation needs to live in one file.
- Schedule messages and files for the quote you promised to send at 8am and finished building at 10pm.
- Register product, plan and contact names in Glossary and names your way. The transcription stops misspelling the words you repeat ten times a day.
None of this is magic. It moves the listening work to before your attention instead of after it.
Where this stops
The part no advert mentions.
- It needs the computer on. The extension lives in the browser. A whole Friday out of the office, laptop shut, produces no transcripts at all. The voice notes stay in WhatsApp and can be transcribed when you open the chat.
- There is a daily quota. The Free plan gives 8 transcriptions a day and audio of up to 15 minutes, with a watermark. Starter goes to 60 a day and 30 minutes. Pro reaches 150 a day and 150 minutes. During the 7-day trial the cap is 15 operations a day and up to 30 minutes of audio.
- On Free, Search inside voice messages covers the last 7 days. Hunting for a sentence from three months ago needs a paid plan.
- Product names come out wrong. Brands, models, plan names, a buyer's surname: transcription trips over them. The Glossary fixes what repeats, not what appeared once.
- The text is a working note, not a document. It is there to remind you, to feed the CRM and to align the team. Quotes, discounts and deadlines are formalised in writing, in the order or the contract the sale requires. The 4 min 12 s client in this piece is illustrative, invented to explain the flow.
- It does not replace the call. Someone who sends three minutes complaining about delivery wants a call back, not a text reply. Reading first only means you call knowing what it is about.
Seven days to measure it on your own pipeline
This is not settled by argument. It is measured over one week of pipeline. Every ZapVox signup starts with 7 days of trial with the features unlocked, no credit card. When the seven days end, the account returns to the Free plan and nothing is charged.
Try it free for 7 days, no card: switch automatic transcription on Monday, work a normal week, and on Friday count how many times you reopened a voice note just to check a number.
In two lines
- The order arrives by message and the terms arrive by voice. Between those two things sits a slice of your day that nobody measures.
- On the phone, native transcripts handle one note at a time. On the computer, where the CRM is open, an extension is what makes the voice note arrive as text, searchable and ready to paste.