Why we score thinking and delivery separately

Most feedback on speaking blends two different things into one impression. Did it sound good? The trouble is that sounding good and thinking well are separate skills, and each one hides the other. A confident voice makes a thin idea sound substantial. A shaky voice makes a sharp idea sound unsure. If you only get one number, you cannot tell which one you need to work on.

Five for what you said

Depth of Ideas is how far past the obvious answer you get. Thinking Lens is whether a mental model shaped the answer. Framework Use is whether the answer had a shape a listener could follow. Use of Examples is concrete instances rather than general claims. Conciseness is saying it once and leaving it alone, rather than circling. It is not about speed, and it is not about length.

Five for how it sounded

Conviction is statements that land rather than lift into a question. Pace and Cadence is your rate, and how deliberately you change it. Energy is the pitch and volume dynamics that make a listener lean in. Articulation is word endings that land and vowels that hold under pressure. Intentional Pauses is placed silence rather than stalls.

Two of these deserve a note, because they are where scoring can go wrong. Conviction is about whether a point lands, not how loud you are. A quiet speaker can have plenty of it. And Articulation is not an accent score. An unfamiliar accent is never marked down.

Two numbers, not one

So a Cold Stage report gives you two scores, one for thinking and one for delivery, and the ten dimensions underneath them. Each part opens to what it measures, the evidence from what you said, and one fix. The point is that you can always trace a number back to something you actually did.

Why only thinking steers the program

Your daily program is steered by your thinking, not your delivery. That was deliberate. My view is that delivery improves most when there is something worth delivering, and a lot of what reads as a delivery problem, the fillers, the trailing off, is really a thinking problem: you are searching for the next idea out loud. Fix the thinking and the delivery often follows. Delivery is still measured in every Cold Stage, and Learn has a whole track for it, called Your Voice.

Being honest about the AI

Every rep is scored by AI, and you should be able to check its work. So every report carries an AI-generated label, and every section of it has its own thumbs up and down with its own question. Did it hear this right? Do these scores feel fair? Was this the right thing to work on? There is one for the whole report too. We read those ratings. They are how we find the places where the scoring is off.