How can we tell a business task contains a sufficiency judgement worth testing separately?

Use these signs to spot tasks where an agent must judge whether the paperwork is enough before it acts.

- A written policy lists required evidence that varies by product, amount or customer type - Staff can legitimately pause a case with a status such as 'awaiting information' - Files arrive in several parts, from different senders, at different times - Documents have validity rules such as expiry, signature, issuer or date range - Substitutes or waivers are allowed under stated conditions - Outbound requests to customers or colleagues are part of the task - Earlier decisions must be remembered so satisfied items are not requested twice - Proceeding on a weak file creates cost later, such as rework, loss or a compliance review

If three or more of these hold, a sufficiency check is probably a distinct step worth testing on its own. Isolating it lets you see whether failures come from reading documents, applying policy or writing the request, which an end-to-end task outcome cannot show. It also lets you build cases for each branch of the decision, including the quiet ones where the correct answer is simply to proceed without asking for anything further.