A whale descending from the surface into dark water

Inside the investigation

The surface tells you almost nothing.You have to go down.

An examination is not a report writing exercise with data attached. It is a sequence of six stages, each of which can end the work if the evidence does not hold.

Footage: Chen EdisoN / Pexels

Method

Connect. Analyze. Investigate. Substantiate. Quantify. Correct.

01

Connect

We take custody of the records that matter and preserve every original exactly as received, with an integrity value recorded on intake.

02

Analyze

We reconcile systems against one another over the same period to surface combinations that cannot all be true.

03

Investigate

Analysts examine every candidate, test alternative explanations, and discard what the evidence does not support.

04

Substantiate

A finding exists only when a named analyst can trace it to source records and an independent reviewer agrees.

05

Quantify

Exposure is computed deterministically, with stated inputs, calculation versions and a duplicate exposure check.

06

Correct

We map the control failures that permitted it and set out what recovery and remediation require.

Deep underwater scene far below the surface

Most of what matters

is below the last place anyone looked.

Footage: Pavel Danilyuk / Pexels

Difference

How this differs from an audit.

Conventional reviewPERISCOPE examination
Samples a population to estimate a rateExamines the full population to name specific records
Tests whether systems reconcileTests whether the combined account could have happened
Reports issues and observationsSubstantiates findings a named analyst traced to source
Estimates impactComputes exposure deterministically, reproducible to the cent
Numbers cited without lineageEvery material statement traceable to preserved evidence
Findings assessed by the reviewer aloneIndependent review, classification and certification before release

Discipline

What we refuse to do.

  • We do not substantiate anything an alternative explanation still accounts for.
  • We do not report an estimate as a confirmed figure.
  • We do not allow a model to calculate exposure or approve a finding.
  • We do not count the same records twice across findings.
  • We do not assert motive or intent from records alone.
  • We do not publish detection logic, thresholds, weighting or rules.

If the numbers have stopped agreeing with each other, that disagreement is evidence.

Request an examination