Competency models are among the most useful tools in a soft-skills strategy and the most used yardstick for measuring skills. In our experience, though, they are neither the most intuitive way to describe performance to trainees nor the most complete way to represent it in a conversation. That is why Skillgym measures the outcome of a conversation along several dimensions. Reading them well requires knowing a little about how they are built.
A premise on where the numbers come from
Every measure described here is produced from what the trainee said during the conversation, by voice or by text, and from their own self-assessments. Spoken words are transcribed and treated exactly like typed ones. Nothing about the trainee's voice, face or state enters any score. The analysis recognizes which behavior each sentence expresses and how effective it was in that moment of that conversation; the scores are then computed with fixed rules, the same for everyone and the same over time.
1. Quality of the strategy
Quality of the strategy is the first and most important indicator. Skillgym measures it as the overall quality with which each skill is used along the conversation, in context. It is the metric that grows most with experience, and it is what builds a person's confidence in handling these conversations: the more conversations are held and reflected upon, the higher it gets.
Three things to keep in mind. The same skill may require different behaviors in different contexts; it can be developed through different behaviors as the situation changes. Behaviors live in what you say: "how would you like to proceed?" expresses the behavior of asking an open question, which belongs to the skill of asking questions. And whether a behavior scores high or low depends on the moment: Skillgym evaluates it taking into account the whole flow of the conversation so far.
The best way to raise it is to say things that are not only appropriate as dialogue but fitting for the real-time context: who you are talking to, how it has gone so far. Over time, reviewing conversations with the character's reactions explained at each step makes this judgment easier.
2. Self-awareness
The second indicator is self-awareness: the quality of your perception of your own performance compared with the evaluation of the conversation. Higher self-awareness is associated with better performance and faster improvement. At the end of a conversation you answer a few questions rating your performance on specific competencies and behaviors; the analysis produces the same evaluation from what you actually said; the comparison gives the score. The closer the two, the higher the self-awareness.
3. Objectives
Objectives are the specific outcomes you are asked to reach in a conversation. They are presented in the preparation room and connected to specific steps of the dialogue. In the review, each one is marked as reached or not, with an explanation of why.
A common mistake is to assume a perfect correlation between quality of the strategy and objectives. Usually good behaviors produce a good conversation and the objectives follow, but Skillgym is a simulation, not a deterministic exercise, and the flow matters. One unlucky sentence at the wrong moment can shift the character's attitude enough that, when the time comes to agree to what you propose, they decline. You may then see a reasonably high strategy score and one objective not reached. It can even happen that the character reacts well at that instant and still does not concede, because the weight of earlier behaviors prevails. These cases are not frequent, and they teach the most important lesson: every behavior counts. That is why we keep the two indicators separate.
4. Phases
A conversation is a continuum, where every step is influenced by what came before and influences what comes next. For learning purposes it helps to split it into phases and see the quality of your strategy in each, as a percentage and with a color code. This lets you pick a specific part of the conversation to work on next.
5. Competencies
Competencies are measured on observable behaviors, and observable behaviors are what you express in each sentence. Each behavior receives a score according to the context in which it appeared; the scores are aggregated according to the competency map, which is almost always the client's own model. Because competencies are an aggregate, they are most meaningful across several conversations over time, but they are informative even after one. Many people find their organization's competency model abstract and hard to connect to daily behavior; seeing a single conversation measured against it is a useful way to make it concrete.
Beyond the five: six families
Today the analysis of a session covers six families of indicators: the quality of the conversational strategy, the impact produced on the character turn by turn, self-awareness, knowledge of the subject matter, compliance with the rules that apply to the context, and transversal behavioral competencies. Each comes with a commentary on your specific choices, and an AI coach who knows your session is available to discuss any of them.
Who sees what
Your conversation analytics are yours. They sit in your private space, with your recordings, your debriefs and your coaching chats. The organization sees program progress in aggregate: participation, improvement trends across teams and over time, and where the competency model is holding up. That separation is deliberate, and it is what makes it safe to experiment, make mistakes and score wide rather than high.