User experience design is a tricky thing, full of tiny, seemingly insignificant pitfalls that can end up causing major frustration for users. One common pitfall is being more precise than you are accurate. The typical example uses pi: 3 is accurate, but not precise, while 3.6789 is more precise, but less accurate. Accuracy in a system is controlled by a wide array of factors, but generally you’re aware of the limitations in place and you have a general idea of your accuracy. Precision you can control directly through interface design, so you should always have it match your accuracy, never exceed it. Any estimation, extrapolation, aggregation, or rounding can introduce a loss of precision. Digital floating point math is inherently imprecise.
Users naturally presume that any number they’re looking at is as accurate as it is precise. If you show four decimal places, they assume that number is accurate to four decimal places, and rightly so. If you show numbers in millions (37M), users will assume this is accurate to the nearest million. They naturally trust you to present them with accurate information, so they assume that whatever information they’re given is accurate. This is exactly why you should ensure that you don’t present information that you don’t know to be accurate.