Language

Purview

Label the right documents with the right label

Purview labels what it can parse. Data & More classifies the rest, scores how well your labels actually protect it, and writes the correct label back into Microsoft 365.

Book a demo See the benchmark

Labelled in Purview

3.2%

of assessed documents, against the 4.5% that hold sensitive data

ClassifyDecide what the data actually is.
VerifyLet the person who owns it confirm.
Act (this page)Block, strip, relabel or remove.

The Purview dashboard Scope tab: documents monitored, assessed, classified by Data & More and labelled in Purview, above the label mapping that pairs each privacy classification with a sensitivity label — The gap, in four numbers. 1,923,774 documents monitored, 84,769 that Data & More classified as sensitive, 61,797 carrying a Purview label. Below it, the mapping: every privacy classification against the MS sensitivity label it should carry.

Purview 2.0

Scope, measure, repair

A dashboard that stops being a measurement and starts being a fix.

Scope

How many documents are monitored, how many have been assessed, how many Data & More has classified, and how many carry a Purview label. The gap between the last two numbers is the work.

Set business logic

Map each Data & More classification to the sensitivity label it should carry, with optional alternative labels and defaults for files and emails holding nothing sensitive. Nest labels to rank them, and mark the ones that genuinely enforce protection through encryption or rights management.

Measure

A Purview protection score next to the optimal score a labelling policy could reach, alongside a data protection analysis. One number that says how much of your protection is real.

Detailed analysis

Every assessed item lands in one bucket: correctly labelled, over-labelled, under-labelled, not labelled, not mapped, or not labellable by Purview, broken down by label and by classification.

Repair

Write the right label back. A repair rule either stamps the label into the file's metadata or asks the Purview API to apply it, can be limited to a set of data owners, and sends every fix to an append-only log.

Purview

Awaiting capture

The Purview dashboard showing the protection score against the optimal score

public/product/purview-measure.webp

The protection score, next to the best a labelling policy could reach.

Proof

Measured on 89,521 files

89,521,

files benchmarked

99.8%

of passports Purview misclassified

91%

of payment cards Purview misclassified

languages Data & More classifies

Purview

Awaiting capture

Detailed analysis breaking items into correctly, over and under-labelled

public/product/purview-analysis.webp

Every assessed item in one bucket: correct, over, under, missing, unmapped.

Capability

What it adds to Purview

Classification

The categories Purview does not have

Health, union membership, political views, sexual orientation, tax, salary, CVs and consents, classified in 28 languages against local regulation.

Accuracy

Semantics, not pattern matching

Purview's out-of-the-box classifiers are largely regex-based. Ours combine taxonomies, NLP, OCR and face and ID recognition, at a false positive rate of 0.003277%.

Over time

Change over time

Coverage and protection followed through a daily snapshot per workspace, so a labelling programme can be shown to be working rather than asserted to be.

Sources

More than Microsoft

The shared-data chart filters by source: email, OneDrive, SharePoint, Teams, Google Drive or Slack.

Reports

Labels in the report

MS sensitivity labels appear as a report column, next to the classification, the data owner and the location.

Writer

Two ways to write a label

Stamp the file, or go through the Purview API. Legacy Office formats, locked files and API refusals fall back to stamping, and email is always stamped.

Purview

Awaiting capture

The repair panel with the label writer and owner scope

public/product/purview-repair.webp

Repair: stamp the file or call the Purview API, scoped to the owners you pick.

Questions about Purview

Does this replace Purview?+

No. Keep your sensitivity labels, auto-labelling, DLP and retention. This makes the labels correct, which is what the rest of Purview depends on.

What does Repair actually change?+

It applies the mapped label to items carrying the wrong one, either by stamping the file or through the Purview API. It is created for review and does not run until it is approved.

What do we need on the Microsoft side?+

Sensitivity labels are read through your Microsoft 365 app registration. Writing labels through the Purview API needs an app registration Microsoft bills; stamping does not. If the registration is missing a permission, the dashboard says which one.

Where does the benchmark come from?+

A public dataset of 89,521 files with manually validated ground truth. Method, numbers and the dataset itself are on the benchmark page.

Score your own labelling

We run the measurement on your tenant and show you the gap between the labels you have and the labels the data needs.