How do I compare performance across several locations fairly?

Use the same definitions and time periods, then normalize for factors that make locations genuinely different, such as size, traffic, staffing, or operating days. Compare both totals and rates so larger sites do not automatically look better or worse.