Churn is a lagging metric when contracts are involved
Customer churn is a lagging indicator.
I once worked at a company in which customer churn was considered an important metric. A great many things revolved around it, including the monthly bonus (if any) for the support teams. One thing I noticed is that they were using churn as an indication of whether their service level was any good. That's a bad idea. Here's why.
Most of the customers at that company were tied up in some kind of contract. The usual term was 12 or 24 months. That means if you did something really bad to that customer and they didn't disappear next month, that doesn't mean anything. They may be sticking it out until the contract elapses just because it's easier (and possibly cheaper) than trying to cancel early.
When things started getting really sticky in support, with horrible products being rolled out, lazy people on the floor, and managers who didn't care, some of us tried to point out that customers were not happy. The managers and bean counters just pointed at the churn numbers and said "but churn is better than ever", and ignored us.
Meanwhile, on the other side of the phones and tickets, those customers who had been burned were probably looking at our competition to see what their options were. They had time to figure this out and make a leisurely transition into a new hosting company. That way, when their contract ran out, they could just walk away and stop returning our calls.
The account managers would go nuts and try to figure out what had gone wrong by reading their recent ticket history. But, odds are, there wouldn't be anything useful to read since they had long since given up on the company and had moved on. Any useful data about what broke was buried in the past, and hardly anyone cared enough to go back and review all of it to get the full picture.
Somehow, this company is still in business. It looks like they managed to find a customer base which doesn't care about mediocrity. Amazing.