Three Numbers, One Page

Can your CEO tell you cash on hand right now, without opening a laptop? At one company the answer was yes, and the reason was a discipline that looked almost primitive next to the finance team's work.

The CFO ran a 30-tab financial model. Inputs, assumptions, sensitivity tables, three scenarios with quarterly drill-downs, a cohort retention waterfall, a unit-economics tab built off a cap table legal had last touched nine months earlier. It was a fine piece of work and the right deliverable for the function.

The CEO ran the company on three numbers, written in pen on the back of every meeting agenda. Cash on hand. Net new ARR this month. The count of customers who'd logged in fewer than three times in the last two weeks. He could recite all three at any point in the day and tell you the trajectory of each over the past eight weeks. He could not tell you what lived on tab seventeen.

This is the right division of labor, and most founders run it backwards.

The CFO's job is the model. It exists for diligence, for board exposure, for the planning no human holds in their head. It is right and exhaustive and it exists so the CEO doesn't have to build it. The CFO is graded on the model being accurate.

The CEO's job is the three numbers. They are what the CEO steers against in every conversation, every hire, every reply sent at 11pm. They live in memory because they have to. A CEO who opens a dashboard to learn cash on hand has outsourced the steering to the dashboard that is quietly lying to them — one degree of separation too many, and the same failure a status meeting fixes when it is cut down to three numbers.

The three change by company. Pre-revenue: cash, monthly burn, the leading indicator of the bet. Series B SaaS: cash, net new ARR, gross retention. E-commerce: cash, contribution margin, repeat purchase rate. The discipline is naming the three and committing. You don't pick them by template. You pick them by asking which three numbers, if any one moved twenty percent in either direction, would change a decision this week.

Three is a cognitive limit, not a preference. The brain holds about that many live priorities at once. Four is hard. Five is a story you tell yourself. Most leadership dashboards carry fifteen, which means the dashboard is the priority and the priorities are missing. The CEO who can recite three has chosen. The CEO who needs the dashboard has been chosen for. Zoom out far enough and the model exists for one reason: so the three numbers can be trusted.