> ## 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.

# Synthesising Findings

> Turn raw evidence into validated findings without collapsing assumptions, symptoms and root causes into the same thing.

Synthesis is where a diagnostic becomes consulting rather than transcription.

## Separate four layers

### Observation

What the evidence directly shows.

> Sales administrators re-enter customer details from email into the CRM.

### Consequence

What the observation causes.

> This adds delay, increases error risk and consumes capacity during peak periods.

### Root-cause hypothesis

Why the condition may exist.

> The intake process has no structured capture step and the current integration does not parse inbound requests.

### Recommendation

What should be changed.

> Introduce structured intake and automate transfer into the CRM, with human review for incomplete submissions.

Do not skip from observation directly to recommendation without validating the root cause.

## Find patterns across evidence

Look for:

* repeated handoffs;
* queues;
* duplicate entry;
* inconsistent decision rules;
* frequent exceptions;
* rework loops;
* system switching;
* manual checks;
* unclear ownership;
* known problems that persist because nobody has authority or incentive to change them.

## Distinguish technical and operating-model causes

The same symptom may have multiple causes. A delay may be caused by missing integration, unclear ownership, policy design, staffing, incentives or approval structure.

Do not recommend technology when the real constraint is organisational.

## Validate before finalising

Use a client validation session to confirm:

* whether the reconstructed process is recognisable;
* which findings are factual;
* which are assumptions;
* where evidence is incomplete;
* whether the proposed causes are credible.

The goal is not unanimous agreement. It is a transparent, defensible evidence base.