If there's multiple subscription changes occurring for one customer during the same calendar month, we bundle these together, and show you the end result of what occurred during that month.
Thus, if somebody signs up and churns within the first month of purchase, we won't include them in your metrics whatsoever. The reasons here are twofold: seeing churn increase from a customer who likely isn't part of product-market fit seems misleading, and many companies choose to refund if an unhappy customer wants to cancel immediately after initial signup.
Another example of how bundling these intra-month changes manifests is a new customer that signs up and then upgrades a few days later, within the same calendar month. We'll show this as a new customer for the full amount, bundling in the upgrade. It would be misleading to place this in expansionary revenue, essentially, when the customer most likely was a fit for the more expensive plan in the first place.
Finally, imagine a customer who signs up, upgrades to a very expensive plan, and then churns all of that revenue within the same month. You would see the revenue from this customer have an outsized impact on your metrics, both positively and negatively, when in actuality, the net in this situation should be 0, for all intents and purposes.
We believe that MRR is a momentum-metric, meant to be actionable, and the bundling of changes at the intra-month level are a result of this belief. We want to avoid inflating or deflating your other metrics in an unwarranted manner, when those actions don't represent the intended outcome.