Confidence Score

Learning

A confidence score is a measure of how much evidence actually supports a finding, so a solid, well-backed flag can be told apart from a weaker, more uncertain one.

Not every inventory problem is equally certain. Some things can be proven outright from a single check. Others are more of a "this looks a bit odd" situation that deserves attention, but isn't definite.

A confidence score is how a system tells those two apart, instead of treating every flag as equally urgent.

Why this matters more than it sounds

A tool that flags everything at the same volume trains you to ignore it. If "definitely broken" and "might be worth a look" both show up as identical red alerts, you stop trusting either one.

How Ghost Stock uses it

Each finding Ghost Stock surfaces carries a confidence level, built from how much real evidence backs it and how complete the underlying data was at scan time. A single snapshot of your store can only ever say so much. As Ghost Stock observes a store over more time, that confidence has more to actually stand on.

It's a deliberate choice not to oversell certainty we don't have. Some things genuinely deserve a lower confidence label, and we'd rather say so than round everything up to "urgent."