What you get
Ask a model to review a call and it will find something nice to say.
That is what it is shaped to do. Which makes the output pleasant, encouraging, and completely useless on exactly the calls that needed reviewing — the ones that lost.
So the instruction has a floor written into it, in those words.
The floor
Three findable mistakes, and a rule against inventing any.
The two halves have to arrive together. A requirement to find fault, with nothing holding it down, produces a machine for manufacturing criticism — which a rep will discount within a week and never open again.
A call that lost, or where the rep talked more than two-thirds of the time, or where sentiment finished low, contains at minimum three distinguishable mistakes — find them, quote them, and list each separately.
- opening with a monologue instead of discovery
- ignoring what the prospect just said
- selling past the stated need
- guessing at a price or a fact
- dismissing a competitor instead of understanding them
- not confirming a concrete next step
- Missed moves are their own list —
Openings the prospect gave that the rep did not take — a stated pain not dug into, a buying signal not pursued, a stakeholder mentioned and never asked about. Usually more useful than the mistakes, because they are recoverable.
- What worked is only collected on a won call —
And lessons only on a lost one. So "what to repeat" is grounded in something that actually did.
- It will not open with a compliment it cannot evidence —
Every entry, positive or otherwise, carries the quote it came from.
Comparison
Against the middle of the team — and the middle excludes you.
Seven measures over a rolling four weeks. Each is ranged against the team's median rather than a target somebody invented, and the median is calculated without the person being looked at, so nobody is measured against a figure they helped set.
When somebody is the only person with calls in the window, the comparison is withheld and says so, rather than comparing them against themselves.
The intent is written into the code: a rep should read "you are third on discovery depth", not "you are a bad rep".
- No ranking, no severity colouring —
A position against a median is not a place in a queue, and the product does not turn it into one.
- Weeks start on Monday —
Because that is when the stand-up is.
- No calls returns a reason and a suggestion —
Not an empty dashboard.
- Live coaching exists, and is deliberately lower fidelity —
Prompts during the call from a cheaper pass over what has been said so far. The full analysis still runs at the end and remains the record. Two gates have to be crossed before it fires — enough new speech, and enough elapsed time — so a quiet stretch costs nothing and a chatty one cannot hammer it.
›Can a rep see their own coaching?
Yes, and it is built to be opened by the person it is about rather than delivered to them by somebody else.
›Does a manager get a leaderboard?
No. There is no ranking and no per-person severity colouring anywhere in this module.
›Is live coaching on by default?
No. It costs materially more than the analysis at the end, and the reason is recorded on the setting rather than left for you to discover on an invoice.
›Who can see somebody else's coaching notes?
Anyone who can read calls. That is worth being deliberate about — the critique of a call is a more sensitive artefact than the recording.