From 'what happened' to 'what to do': prescriptive analytics, honestly

Analytics vendors love the ladder: descriptive (what happened), diagnostic (why), predictive (what next), prescriptive (what to do). It is a genuinely useful map. It is also the most inflated territory in software marketing — most tools sold as prescriptive are predictive tools with imperative sentences bolted on. "Stock is trending low" does not become a prescription by rephrasing it as "Reorder now!".

What real prescription requires

A recommendation worthy of the name has four load-bearing parts:

A quantified action. Not "consider reordering" but "order 40 units by Thursday" — specific enough to be wrong, which is the price of being useful.

Consequences of alternatives. Prescription is choosing between futures, so it must show them: order now versus wait a week, in money — like the markdown maths of acting early versus late.

Constraints respected. A reorder suggestion that ignores your cash position, or a rota fix that ignores contracted hours, is a maths demo. Real prescription knows your contribution floors, your budgets, your rules — and shows its reasoning against them.

An accuracy trail. If the tool prescribed 40 units, what happened? A prescriptive system that never scores its own past advice is asking for permanent faith. Decline.

The demo test that sorts the market

Ask one question of any prescriptive tool, ours included: "Show me a recommendation you made, the reasoning it carried, the constraint it respected, and the measured outcome." Genuine systems answer with a record. Dressed-up dashboards answer with a roadmap.

And one boundary worth paying for rather than avoiding: the best prescriptive systems *recommend and stop*. The tool that quantifies the action, argues its case, and leaves the deciding to a named human is not a lesser version of one that executes — it is the correct version. We built Kauzio as exactly that: prescription with its homework attached, and a human hand on every send button. That is the honest top of the ladder — not software that runs your shop, but software that has done the thinking by the time you get there.