> ## Documentation Index
> Fetch the complete documentation index at: https://irisdocs.prescientlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Scoping a Diagnostic

> Define a bounded business question, process scope and evidence plan before the engagement begins.

Poor scope creates poor diagnostics. “Find every AI opportunity in the company” sounds attractive but usually produces shallow evidence, weak recommendations and uncontrolled delivery effort.

## Start with the business question

Good questions include:

* Where is the order-to-cash process losing time or capacity?
* Which parts of proposal production create avoidable rework?
* Where could automation improve customer-service throughput without reducing control?

The question should connect a process to a business outcome.

## Define scope explicitly

Document:

* process start and end points;
* teams and locations included;
* relevant systems;
* stakeholder groups;
* evidence sources;
* time period where structured data is used;
* outputs;
* exclusions.

## Set success criteria

Examples:

* validated current-state map;
* agreement on the top three constraints;
* prioritised opportunity backlog;
* quantified assumptions for the top opportunities;
* decision on the first implementation project.

## Identify client responsibilities

The client may need to provide:

* sponsor and process owner;
* stakeholder access;
* operational exports;
* policy or process documentation;
* security approvals;
* timely review of findings.

## Scope control question

Before accepting a new request during delivery, ask:

> Does this help answer the agreed business question, or is it a new piece of work?

If it is new, trade scope, timeline or fee rather than silently absorbing it.

<Tip>
  A narrow diagnostic that produces a strong implementation decision is usually more valuable than a broad diagnostic that produces twenty generic ideas.
</Tip>