In-store analytics measures what happens inside the space using sensors rather than tills: how many people entered, which zones they visited, where they stopped, and how that changes by hour or by display.
The value is the denominator problem again. Sales tell you what was bought; sensing tells you how many people were near it and did not buy, which is the number that distinguishes a merchandising problem from a range problem.
The caveat is that sensing accuracy is far lower than the dashboards imply. Overhead counters miscount groups and pushchairs, wifi-based measurement now sees only a fraction of devices because MAC randomisation is default on both major mobile platforms, and staff movement inflates everything unless deliberately excluded. Treat the figures as directional, and compare a site with itself over time rather than with another site.
The privacy line is the same as everywhere else in this category: aggregate counting at the device is defensible; storing anything that re-identifies a visitor across visits is a different legal obligation under GDPR and KVKK, and needs a basis you probably do not have.