You missed the signs. In retrospect they all seem so clear. Looking back on it, you should have known. But there were good signs as well as bad. And the bad did not seem that bad especially given the situation. And you were busy. And there could have been a lot of explanations for what you saw. And the friends who you confided in had different opinions based on their own experiences. It was just so hard to tell. If only you could have put the pieces together when you still had time to do something about it.
But you didn't put the pieces together because you couldn't. And you couldn't because you are a human. We might not like to think this about ourselves but studies from the 1950s onward have shown that humans are pretty awful at integrating multiple, noisy signals into an accurate view of a situation. This ability only improves as the signals become unambiguous. But, by then, it is often too late to change outcomes.
In business, our solution to the problem is to produce more data. But that actually makes the situation worse. Pick one of your KPI dashboards. Let's use Account Health as an example. You might have indicators for support tickets logged, NPS score, product usage, and so on. The implied expectation is that you will look at the indicators for an account and think to yourself: "Product usage is up, which is good. But support tickets are also up, which might be bad. Or it could be that an increase in product usage has led to more requests for support. Of course NPS scores are down, which is bad. But then again, the rep on this account says it is healthy. And they are a young account, which might explain some of this. But they are also a smaller account, so you would think that…"
Does this account need attention or are they healthy given their size and tenure? Would providing training that reduced the number of support tickets make the account healthier? Now imagine running some version of this same calculation with more indicators and for every account on your dashboard.
You are not equipped to do this well, and neither is anyone else. But the signals themselves aren't the problem. What's missing is a way to combine them accurately, consistently, and in enough time to influence outcomes.
← Back to Writing