Interpretation
When averages lie about stores
Portfolio averages comfort executives. They also smother the two stores that have been quietly leaking margin since last quarter. If your multi-location board leads with a single network mean, you are practicing optimism as a default.
Pair every average with a dispersion cue
We favor a compact strip: mean, median, and count of locations below a negotiated floor. The strip is not a statistics lecture—it is a tripwire. When median and mean drift apart during a promo week, someone must open the branch list.
Rankings are optional; absences are not
Instead of naming “bottom five” every morning (which breeds defensive theater), show which locations failed to report. Missingness is often the earliest signal that a store’s average is synthetic.
A short drill
Take last week’s “healthy” region chart. Remove the top two sites. Recalculate. If the story flips, your visual hierarchy was celebrating outliers, not describing the network. This drill appears in our Visual Escalation Clinic because it changes what leaders ask for next.