The transcript and the quote
Every observation comes with the words that triggered it. That quote is how you catch us being wrong, and it is the most important line in your snapshot.
Before anything is measured, your recording is turned into text. Every number and every observation in your snapshot is calculated from that text, not from your voice.
Which means the whole thing rests on one step that can be wrong.
Why every observation carries a quote
Machine transcription misreads accents, background noise, and quiet endings. When it does, the error travels silently: the text looks confident, the analysis treats it as fact, and the snapshot prints a verdict built on something you never said.
So every observation shows the words that triggered it:
“…people are buying vegetables…” — you described what people were doing, but never where they were.
That quote is not decoration. In your inbox you cannot play your audio back, so it is the only way you have to check our work. If the quote does not sound like what you said, the observation is wrong, and you should ignore it and tell us.
The failure this exists to prevent
The worst mistake this product can make is not a missing feature. It is telling you that you didn’t do something you actually did — because the transcription lost it.
Someone preparing for an exam changes what they practise based on what we say. A false negative sends you to drill something you already do well, and quietly teaches you to distrust everything else in the snapshot.
The quote is the cheapest possible defence: you read one line and you know.
What the transcript is not
It is not a spelling or grammar check of your speech. Written English and spoken English are different, and a transcript of natural speaking looks messy on the page even when the speaking was good. Nothing in your snapshot is scored on how the text reads.
If a quote looks wrong to you, write to us. Those are the most useful messages we get, because they are the only ones that show us where the machine breaks.