# MSL insight analysis
> Your MSL notes, scored, structured and turned into a decision.
Source: https://www.visfo.health/projects/msl-insight-analysis
Tags: Sprint, AI-assisted

_Your MSL notes, scored, structured and turned into a decision._

A data science project on your real note corpus. Every insight scored against a rubric for insight quality and documentation quality, tagged, clustered and compliance-flagged, then handed back as findings, a scored dataset and coaching recommendations.

## What it is

A data science project on the notes your MSLs have already written. We ingest the corpus, score every insight against an agreed rubric, tag and cluster it, and hand back what the field is actually telling the organisation, alongside an honest read on how much of it was captured well enough to be usable.

Two questions get answered at once. What is in the insights, themes, objections, evidence gaps, sentiment, by territory and by team. And how good the insights are, note by note, so coaching goes to the people and habits that need it rather than to everyone equally.

Insight quality is scored on educational need captured, clinical context, actionability, organisational value and evidence linkage. Documentation quality is scored separately on structure, objectivity and completeness, because a well-written note about nothing and a badly written note about something are different problems with different fixes.

## How we run it

1. **Frame** — Agree the corpus, the therapy area and the strategic questions the insights are meant to answer. Calibrate the rubric with medical and MSL leadership on a sample, so the scores mean what your team thinks they mean before we score at volume.

2. **Ingest** — Pull the note corpus from CRM export, Veeva or flat file. Normalise, de-duplicate and date-audit it, so the analysis runs on what was actually captured rather than on a partial or double-counted extract.

3. **Score and structure** — Every note scored on the five insight-quality dimensions and the three documentation-quality dimensions, with written reasoning per dimension. Compliance status flagged pass, flag or fail. Tags applied against a canonical taxonomy covering theme, topic, evidence gap, HCP sentiment, suggested action and therapeutic context.

4. **Analyse** — Quality distribution by team, territory and author, not just an average. Theme and evidence-gap clusters across the corpus, with the underlying notes attached. Where the strong signal is concentrated, and where whole territories are producing volume without value.

5. **Deliver** — Findings and recommendations, the scored dataset, worked before-and-after rewrites for coaching, and the rubric and taxonomy handed over so the scoring can keep running after we leave.

## What you get

- Scored dataset with dimension scores, written reasoning and tags on every note

- Insight quality baseline by team, territory and author, showing the distribution not just the mean

- Theme and evidence-gap clusters across the corpus, with the notes behind each

- Compliance flag register for notes needing review

- Coaching recommendations per group, with worked before-and-after rewrites

- The rubric and tag taxonomy, documented and yours to keep running