A scale shows a mass of 0.03060 kg. In grams the same mass is 30.60 g, and the zeros at the front have vanished, which shows they only placed the decimal point. The other digits survive any change of unit: the 3, the 0 between the 3 and the 6, the 6, and the final 0. That final 0 matters because the scale chose to display it: it measured the hundredths of a gram and found 0. So the reading has 4 significant figures.
The rules follow from that example. Nonzero digits always count. Zeros between nonzero digits count, so 4005 has 4. Leading zeros never count, so 0.0050 has 2. Trailing zeros after a decimal point count, because someone chose to write them, so 4.50 has 3. Trailing zeros in a whole number like 1500 are unclear, and scientific notation settles it: 1.5×103 has 2 significant figures, 1.500×103 has 4.
Rounding to significant figures is not rounding to a decimal place. Round 0.004873 to the nearest thousandth and you get 0.005, with only one significant figure. Round it to 2 significant figures and you start counting at the 4, the first nonzero digit, giving 0.0049. Decimal places count from the decimal point; significant figures count from the first digit that was really measured. Calculations follow the same idea: when you multiply or divide measurements, the answer keeps as many significant figures as the least precise one.
Think of it like this. Picture each significant figure as a digit you could defend to a judge by pointing at the instrument that produced it. Leading zeros were never read off any display, so they don't count. A trailing zero the display showed does count, and a calculator's extra digits never do.
Core idea. Significant figures count which digits in a measurement actually carry real information about its precision.