A program dashboard fails in one of two ways. Either it shows so much that no one can find the signal, or it shows so little that no one trusts it. Most fail the first way, because adding a chart feels safe and cutting one feels risky.
The discipline is to start from the decisions the dashboard should drive, then work backward to the few numbers that drive them. If a chart doesn't change a decision, it doesn't earn the space. Pretty and irrelevant is still irrelevant.
I want three things visible at a glance. Is the program on track for the date, with a real basis and not a gut feel. What's blocked and who owns clearing it. What's the top risk right now. A senior person should get the program's state in ten seconds and know what, if anything, they need to do.
The hardest part is keeping it honest under pressure. There's a constant pull to color something green that's really yellow, because green is comfortable and yellow invites questions. A dashboard that's always green is worthless. The whole point is to surface the yellow while it's still cheap to fix.