Velocity is the rate of change of the position of an object.
Change, by its everyday as well as physical and even mathematical meaning, implies that you need to compare two things to each other - in this case it is, for an object in motion, first the position of the object at some given time, and then at a given later time. Hence, the difference you can measure between those two positions at different times are your "change" (expressed as a number in meters, kilometers, miles etc.).
Dividing the change (the distance driven by your car) by the time it took to do so gives you a number which we define as velocity (expressed as kilometers per hour, or miles per hour, or meter per seconds, etc.).
So far so good. This is the non-instantaneous kind of velocity which should be no problem and is not what you are asking about.
The instantaneous velocity is an idealized, mathematical concept; not something that occurs in reality. It would be like making a single photograph of a car and asking how fast the car goes. It is not possible to say - there is not enough information.
But what you can do is go back to your original non-instantaneous velocity and make the time between your two measurements ever smaller. You might start out with an hour, go to a minute, then to a second, then 1/10 of a second, and so on. For each of these ever smaller time spans, given that your measurement apparatus is accurate enough, you can find the two positions of the car. All of these different measurements give you possibly different velocities - i.e., the car may be driving slow during the first half hour, and fast during the second. If you measure over the whole hour, you will end up with some average speed. But if you measure only during a minute near the end of the drive, you might end up with the faster speed. All of this is totally normal and what we do all day, every day, when judging how long it will take us from A to B when commuting.
Now we come to the trick: let's make the time ever smaller, and keep either the first or second of the two timestamps fixed. Since cars do not accelerate that quickly in normal circumstances, you will end up with always nearly the same velocity, down to measurement errors. E.g., if you measure your non-instantaneous velocity over a time of 1s or of 0.5s (with the same start time), the car will have driven twice as far in the first measurement, but dividing the displacement by the time still gives the same result since the first timespan is also twice as large.
If you take this to the extreme, you end up with a (still and always non-instanaeous) real, physical velocity which is not changing at all between further finer measurements with one of the timestamps fixed, within your measurement accuracy. At this point, you define this to be the instantaneous velocity at that point in time (the point which you held fixed).
If you are a normal person, you're done and can go on with your happy life. If you're a mathematician, you spend the next half of your life figuring out how to create a formal mathematical mechanism to work rigorously with all of that, and eventually call it "calculus", much to the chagrin of generations of young students wrestling with those ideas!