Retail analytics measures what happens inside the store rather than only what reaches the till. Footfall, dwell by zone, conversion rate against traffic, and the gap between people who passed a display and people who bought.
The value is in the denominators. Sales figures alone cannot distinguish a poor product from a good product nobody walked past, and that distinction changes what you do next: merchandising versus range.
The caveat is that in-store data is far noisier than web analytics and gets treated with the same confidence. Sensors miscount groups, staff movement pollutes footfall, and weather moves numbers more than most interventions. Treat a single week's change as noise; look for effects that survive across sites and seasons.
The privacy caveat applies as it does to any in-store sensing: aggregate counting is defensible, re-identifying an individual across visits is a different legal obligation. Count, do not recognise.