Data-driven content is content whose existence is decided by data, not merely populated by it. The distinction matters: a template filled with today's price is dynamic; a system that decides which product to promote on this screen because of stock, weather and yesterday's sales is data-driven.
The second is where the compounding value sits. It removes the weekly meeting about what to promote, and it makes decisions per screen that no central team could make at that granularity.
The caveat is that the rules encode assumptions nobody revisits. A rule promoting whatever is overstocked will, over a year, quietly train the network to advertise the least popular products, which is exactly what the data said and precisely the wrong outcome. Automated decisions need periodic review against results, not just against whether they fired correctly.
Start with one rule that has a clearly better-than-random answer, measure it against a control, and add rules only when the previous one is proven. A system with twenty interacting rules and no measurement is unfalsifiable, and usually worse than a human choosing.