Discover · Early Case Assessment

Know the shape of the case before anyone reviews a document

Custodian ranking, activity timelines, external-domain exposure, off-hours anomalies, and a communication map — computed straight over the archive with no index to build, and every figure clickable through to the documents behind it.

Where it starts

A matter opens and you know almost nothing: who matters, when it happened, how much there is, what review will cost. This is the screen to open before Search — it sits above it in the console for that reason.

How it runs

01
Profile the corpus
Totals, custodians, conversations, near-duplicate families, unique content, attachment volume, external share — and the derived numbers that set expectations: deduplication reduction, review-set reduction, thread reduction.
02
Rank the custodians
Messages, threads, external traffic, average length, first and last activity — the prioritisation signal for collection order and interview lists. Click a row and you are in Search, scoped to that custodian.
03
Read the timeline and the edges
Activity by month with active-custodian counts; external domains ranked by traffic — where 'outside' is decided by your verified domains, not a guess. With none configured, the screen says so instead of inventing an answer.
04
Look where behaviour changes
Off-hours and weekend activity by custodian, largest threads, and a force-directed communication map where edge weight is message count. The help text says what a vendor usually won't: volume is not significance — use this to decide where to look, never as a finding in itself.
05
Project the matter
The ECA report separates transferable measured ratios (dedup retention, review-set ratio) from stated assumptions (average message size), names the assumption in words, and prices the storage line items.

Why it holds up

Every panel states its own noise floor on screen — 'domains with at least 20 messages' — and the floors scale with corpus size, so a small tenant never reads an empty table as a clean answer.

The network panel refuses to be misread: its response says in words that weights are summed over the heaviest edges, not the whole archive.

Every figure is a drill-through: the number carries you to the documents that produced it, with the scope already set.

Runs on DuckDB directly over the lakehouse — no index build, no search cluster, available the day the data lands.

What you hand the regulator

A workflow that ends on a screen isn’t finished. This one ends in a document.

The ECA report

A single report file carrying the whole assessment — the input to your collection plan, your budget, and your proportionality argument.

corpus profile: totals, dedup & review-set reduction, external share
custodian ranking with first/last activity
monthly timeline · external domains · off-hours by custodian
top threads and the communication network
cost projection with measured ratios separated from stated assumptions

Adjacent workflows

Search & Term Hit Reports TAR / Continuous Active Learning Matters & Legal Holds

See it run on your data scenario

The demo form asks which workflows you want to see — name this one and we’ll stage it.

Book a Demo All workflows

How it runs

Profile the corpus

Totals, custodians, conversations, near-duplicate families, unique content, attachment volume, external share — and the derived numbers that set expectations: deduplication reduction, review-set reduction, thread reduction.

Rank the custodians

Messages, threads, external traffic, average length, first and last activity — the prioritisation signal for collection order and interview lists. Click a row and you are in Search, scoped to that custodian.

Read the timeline and the edges

Activity by month with active-custodian counts; external domains ranked by traffic — where 'outside' is decided by your verified domains, not a guess. With none configured, the screen says so instead of inventing an answer.

Look where behaviour changes

Off-hours and weekend activity by custodian, largest threads, and a force-directed communication map where edge weight is message count. The help text says what a vendor usually won't: volume is not significance — use this to decide where to look, never as a finding in itself.

Project the matter

The ECA report separates transferable measured ratios (dedup retention, review-set ratio) from stated assumptions (average message size), names the assumption in words, and prices the storage line items.