The future is unknown and although some people claim to be able to predict the future, most people can't. But to look a little ahead, into the future, to have a sense of when certain items on the Product Backlog are finished, predictability would be very useful. Although we can't be completely predictable, it is helpful if we are a little more predictable than we are now. And therefore we need some measurement results from the near past.
One way to get these results is just counting how many Product Backlog items (PBI) are Done per Sprint. After a few Sprints it is possible to calculate the average number of Done PBIs per Sprint, and also an optimistic and a pessimistic average. This might give the Scrum Team an idea of how many PBIs they can pick up per Sprint. This way of creating and using velocity works fine if all PBIs have more or less the same size.
But what if PBIs have different sizes? In that case it could be helpful if there is more information about the size of PBIs. A popular way of estimating is Planning Poker and using Story Points. After a few Sprints an average velocity in Story Points can be calculated.
These are just two ways to estimate, and there are more ways to do that, like for example T-Shirt Sizing, Bucket System, etc. But the main point here is that it is important to keep track of velocity, to have more transparency on how much work the Scrum Team can do within a Sprint, and to help the Product Owner managing stakeholder expectations.