Line up 100 students by height, shortest on the left. The student in place 91 has 90 people shorter than them, so their height is at about the 90th percentile. A percentile is just 'what share of the line stands behind me'. Put that same student in a line of professional basketball players and their percentile drops sharply, though their height has not changed. Percentiles are always relative to a group.
That is why a percentile is not a percentage score. On a very hard contest, 30 points out of 100 might beat 85% of the entrants: a score of 30% and a rank at the 85th percentile. One number answers 'how much of the test did I get right?', the other answers 'how many people did I beat?'.
To find a percentile in a data set, sort the values and count. In 20 sorted scores, the 75th percentile sits about three-quarters of the way along, around the 15th value. Textbooks and calculators interpolate between neighbouring values in slightly different ways, so a small data set can give answers that differ a little; with large data sets the methods agree closely.
Quartiles and the median are percentiles you already know. The 50th percentile is the median, the middle position, not the mean. For skewed data such as incomes, the two can be far apart.
Think of it like this. A percentile is your place in a queue sorted by score, read as the share of people standing behind you. Swap in a different crowd and your place moves even though your score did not. The picture is sharp for large groups; with ties or only a handful of people, 'your place' gets fuzzy.
Core idea. A percentile says what share of a group falls below a value, so it measures standing relative to that group, and it changes when the group changes even if the value does not.