Sales speech analytics
Alpina distributor · 2022
Find the conversation patterns linked to better sales.
- Record conversations
- Transcribe audio
- Compare sales
The problem
A field sales rep in mass consumer goods visits 60 to 80 stores a day. Each visit is a human interaction: the greeting, the tone, the order in which products are presented, the arguments, the handling of objections.
That conversation decides whether sales go up or down. But it had always been impossible to measure which conversations actually sell.
What we built
We equipped the sales reps of an Alpina distributor with high-fidelity microphones connected to their phones, recording every sales interaction throughout the day.
The system connected two worlds. Speech-to-text turned the recordings into structured text (greeting, order of products presented, arguments used, tone, handling of objections), and the reps already had sales tracking (average ticket, target achievement, daily results). When a rep had a better day, we analyzed their pitch to find correlations: which words they used, which order they followed, which offers they mentioned, which communication style they had.
A real limitation: the project ran in early 2022 under pandemic restrictions, and face masks affected audio capture. Today we apply the same principle internally: we record our own sales meetings, transcribe them automatically and correlate them with the deals we win.
The result
A system able to analyze thousands of sales interactions and detect the speech patterns that drive better sales results. We use it internally to optimize our own sales process.
Why it’s innovation
For the first time, sales conversations became analyzable data correlated with results. It is not just transcription: it connects what is said with what is sold.
In 2022 it was too early because of pandemic limitations. With today’s AI models it is fully viable and scalable.
