72.00 and 88.00 differ by 20.00%, regardless of which one you call "first."
Percentage difference expresses how far apart two values are, as a percentage of their average, with no assumption about which one is the "correct" or "original" figure. Swap the two inputs and the result stays exactly the same.
Use it whenever you're comparing two independent values that don't have a natural before-and-after relationship: two competing prices, two survey results, two measurements of the same thing taken by different instruments.
Because the numerator already takes an absolute value, and the denominator is a genuine average rather than either input alone, this formula gives identical results no matter which value you call "first", the property that makes it symmetric.
Four steps, and unlike a decrease or change calculation, there's no "which one goes first" decision to get wrong.
A price comparison and a survey comparison, two situations with genuinely no natural starting value.
This is the single most important distinction on this page: percentage decrease is directional, it has a defined original value and a defined new value, and swapping them changes the answer and its meaning entirely. Percentage difference is symmetric, it has no original value, and swapping the two inputs never changes the result. If your two numbers have a genuine "before" and "after," you want percentage decrease (or percentage change), reach for percentage difference only when neither value is privileged as the starting point.
Comparing two current listings for the same product, neither one is the "original" price, so percentage difference, not percentage decrease, is the honest way to describe the gap.
Two independent polls on the same question are compared symmetrically, since neither poll is more "correct" than the other by default.
When two labs or two instruments measure the same sample with no known true value to check against, percentage difference reports how closely their results agree with each other.