Why most pilots never produce ROI.

Walk into almost any mid-market or enterprise operation today and you will find a graveyard of pilots. Each one worked. Each one demonstrated something. And almost none of them produced a result anyone could point to on the P&L twelve months later. The pattern is so consistent that it has become the single most common reason executives grow sceptical of transformation altogether.

The instinctive explanation is technical — the technology wasn't ready, the data was messy, integration was harder than expected. Occasionally that is true. Far more often, the pilot failed for an economic reason baked in before a single line of work began: it was chosen for feasibility, not for economic weight.

A pilot chosen because it is easy to demonstrate will, reliably, be easy to ignore.

The demo trap

Pilots are usually selected because they are convenient to build and impressive to show. That selection criterion is precisely backwards. The problems that are easiest to demonstrate are rarely the ones carrying meaningful cost, and a result that doesn't move a number no executive was watching is, functionally, invisible.

Value that survives the P&L

In practice, the opportunities whose value survives contact with the P&L share four traits. Large — the cost or time being addressed is material on its own. Concentrated — the saving lands in one place you can see. Recurring — the work happens continuously, so the benefit compounds. Attributable — you can measure the baseline before and the result after, and defend the difference.

Miss any one of these and the return tends to disappear — not because it wasn't real, but because it could not be captured, concentrated, or proven.

The Stansbury view

Choosing well means starting from the economics and working toward the problem — not the reverse. Size the prize in money first. The technology question comes last, and it is almost never the one that decides whether the return arrives.