Data Intelligence & Discovery

Practice — 01

Data Intelligence & Discovery

You cannot protect, retain, delete or feed to AI what you have never counted. This practice produces the single most valuable artefact in governance: a complete, classified, continuously refreshed account of everything your organisation holds.

What the practice delivers

The full estate, one inventory

SharePoint, Teams, OneDrive and Exchange; Snowflake, Fabric, Dataverse and the data lakes; Box, Confluence, file shares and archive media. One inventory, one taxonomy, one owner of the truth.

Classification that holds up

Sensitivity and record types assigned by layered methods — pattern, context, trainable classifiers and human adjudication for the contested residue — then sampled and scored so accuracy is a published number, not a hope.

Risk, ranked

Not a heat map for a slide deck. A ranked register of concrete exposures — this repository, this data class, this audience — ordered by consequence, ready to hand to the Exposure & Access Control practice.

AI usage, surfaced

Which copilots, agents and models are already reading your data — sanctioned or otherwise — and what they can currently reach. Most clients are surprised. None regret knowing.

How an engagement runs

  1. Survey (weeks 1–2). Connectors deployed across the estate; first-pass discovery establishes scale, spread and the obvious concentrations of sensitive material.
  2. Classify (weeks 3–6). Taxonomy agreed with legal, privacy and records stakeholders; automated classification tuned against sampled ground truth until precision targets are met.
  3. Rank (weeks 6–8). The exposure register is built and priced in business terms: regulatory reach, breach consequence, discovery burden, AI readiness.
  4. Sustain (ongoing). The inventory refreshes continuously under Stewardship, so the map never again drifts from the territory.

Questions boards ask us

We already ran a discovery exercise two years ago. Why again?

A discovery exercise is a photograph; your estate is a river. Two years of sharing, migration and AI adoption make a static report an artefact of history. We build the inventory as living infrastructure, not a deliverable.

Does this require moving our data?

No. Discovery and classification run in place, inside your tenancy and your security boundary. Nothing is copied out to us — a point your privacy office will appreciate.

How does this relate to Microsoft Purview?

Purview supplies the classification engine wherever it reaches; our tooling and method extend the same taxonomy to the platforms it does not. You get one answer, estate-wide.

Start with the map

Shopping Basket