Forstå poengsummene dine
Hva tallet betyr, hvorfor det endrer seg mellom forsøk, og hvordan du sammenligner folk rettferdig.
Oppdatert
Dette innholdet er ennå ikke oversatt til norsk. Den opprinnelige engelske versjonen vises.
The score itself
Each criterion is scored from 0 to 100 based on the transcript, then combined using its weight. Mandatory criteria that are missed cap the final score at the value configured on the scorecard.
Why a score can change between attempts
Two conversations are never identical. A slightly different persona reaction, a question asked in a different order, a shorter answer — all of these legitimately move a score. This is a feature: it is why five attempts tell you more than one.
Comparing people fairly
Only compare scores on the same scenario. Scores across different scenarios use different criteria and different difficulty, so a 78 on a renewal and a 78 on an escalation mean different things.
Reading the dashboards
The team dashboard shows the median and the spread per criterion. The spread is usually more actionable than the median: a criterion where everyone is average needs a process fix, one where the spread is wide needs coaching.
Exporting
Everything is available through the API and webhooks, including per-criterion scores and the transcript. See the developer documentation for the endpoints.