What the spread tells you that the average does not
An average hides the thing that hurts you. Five sites averaging 79% could be five sites at 79, or it could be a 92 and a 64. The first is a business with a standard. The second is five businesses sharing a logo, and the customer who walks into the 64 does not know they got the bad one, they just know they are not coming back.
The spread is the number to manage. Closing it is worth more than lifting the average, because the average improves on its own once the bottom comes up.
Why the bottom site is usually not a people problem
The instinct is that the worst site has the worst manager. Sometimes true. More often that site is the one where the standard was never written down in a form anybody could follow, so the team invented their own, and it drifted.
The test is simple: if you removed the manager of your best site and dropped them into the worst, would the score move? If yes, you have a people problem. If no, you have a process problem, and replacing the manager just resets the clock.
Score the same way at every site or do not bother
A scorecard only ranks sites fairly if every site was scored with the same questions, the same evidence and the same tolerance for a maybe. Two auditors with different standards produce a league table that measures the auditors.
Photograph anything that fails. A score with no evidence behind it is an opinion, and opinions do not survive the conversation with the site manager who disagrees.