Healthy on the Dashboard, Dead Eighteen Months Later

The number that ends up costing a company everything is rarely the one that was wrong. It's usually the one that was right.

We imagine the dangerous figure in a board deck is the misstatement: the aggressive recognition, the metric quietly redefined, the fudge an auditor eventually catches. Those exist, but they're not what takes companies down. What takes companies down is a number that's accurate, computed correctly, defensible line by line, and chosen, by everyone in the room, precisely because it's the one nobody has to act on.

A board has to run on summaries. You can't govern a company by reading its ledger; you need the few figures that compress the whole into something a non-operator can hold in their head for ninety minutes a quarter. Net retention above 100%, 40% growth, a healthy magic number: these aren't lazy shorthands. They're the distilled judgment of an entire industry about what predicts a company's future, and most of the time that judgment is right. The frameworks exist because they work. A board looking at green has, by the ordinary standards of governance, done its job.

The trouble is what green does to the people looking at it. A number that disappoints gets taken apart. A dip in MRR triggers a cohort analysis, customer calls, an emergency session. A number above benchmark gets a slide and a nod. We interrogate bad news and wave through good news, and call the difference discipline. The same curiosity we'd consider rigorous if the number were falling, we consider paranoia when it's rising. Nobody is thanked for pulling apart a metric that's reassuring the room.

There was a company. Eight million in ARR, growing forty percent, net retention at 110%. Every figure a Series B investor would underwrite without a second meeting. Eighteen months later it was shopping itself to acquirers for a fraction of what it had been worth. None of the data that explained the collapse was hidden. It was sitting in the same system that produced the dashboard. One client was carrying nearly a quarter of the revenue and all of the net expansion; underneath that client, the base was quietly shrinking. Anyone could have seen it. The reason no one did isn't that the metric lied. It's that the metric was pleasant, and pleasant numbers don't get questioned.

I think about a founder who, for three months running, sent his board only net MRR growth. Up and to the right. He hadn't lied. The number was true, and he chose it because it was true. A true number is the safest thing in the world to hide behind. We don't talk ourselves into the story we want by ignoring the data; we do it by finding the one figure that's both accurate and flattering and deciding that's the one that counts. And the board took it, because a reassuring number asks nothing of the people who receive it. Nobody in that room wanted the breakdown. They wanted the version they walked in with. When someone finally asked, the growth was real, and so was a churn rate large enough to make it meaningless. The sales team had been running flat out to stay in place, and everyone had agreed, without ever saying so, not to notice.

I don't think the answer is a better metric, or a more honest one. The numbers were honest. The dashboard worked exactly as designed. The failure was upstream of all of it, in the simple, nearly universal preference for a number that lets you stop looking. The companies where the metrics look best are sometimes the ones where the least looking is being done. And the people inside them are not fools or frauds. They're just relieved.