Skip to content

English court approves predictive coding for 3 million documents

In February 2016 Master Matthews approved predictive coding for over 3 million documents. He found no evidence that it gives less accurate disclosure and judged it far cheaper than manual review.

  • Not rated an organization

    The review had not run when the court approved it, so the judgment reports no measured saving.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Intended benefit.
Evidence

found, the grade of this recordstudiedreportedclaimed

Two independent court documents support the benefit.

Compared with

A full manual review of over 3 million documents. Assumed without a measurement.

The system’s part
Not isolated. The court approved the use before the review ran. Its accuracy finding draws on US and Irish cases, and the costs are estimates. The judgment describes software that learns from a sample that a lawyer reviews.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Not yet checked.
Benefit type
Reduced Task Cost
System
Predictive coding software
Made by
Not named in the judgment
Used by
MWB Business Exchange Ltd (second claimant), with the agreement of all parties
Where
London, United Kingdom
Stage
Proposed
Date
, when the source was published
Kind
Benefit created
Broad benefit
Processing Information, Getting Information
Collections
Scrutiny and findings
Last revised
31 October 2026

Sources

  1. Court document, by authors with no tie to the system.

    High Court of Justice, Chancery Division (Master Matthews). Pyrrho Investments Ltd & Anor v MWB Property Ltd & Ors [2016] EWHC 256 (Ch), approved judgment of 16 February 2016. Find Case Law copy, a scan of the approved judgment.

    There is no evidence to show that the use of predictive coding software leads to less accurate disclosure being given than, say, manual review alone or keyword searches and manual review combined, and indeed there is some evidence (referred to in the US and Irish cases to which I referred above) to the contrary.
  2. Court document, by authors with no tie to the system.

    High Court of Justice, Chancery Division, Companies Court (Mr Registrar Jones). Brown v BCA Trading Ltd & Ors [2016] EWHC 1464 (Ch), 17 May 2016, para. 10, quoting Pyrrho para. 33 with its cost estimates.

    the estimates given in this case vary between £181,988 plus monthly hosting costs of £15,717, to £469,049 plus monthly hosting costs of£20,820

Details

AI system

Workflow

Removing duplicates cut more than 17.6 million restored files to some 3.1 million. All parties agreed on using the software and on how to use it. A senior lawyer reviews a sample of documents to train the software.

Follow-up

Conditions for replication

The number of documents is large enough to make a full manual review unreasonable. The estimated costs stay proportionate to a claim worth tens of millions of pounds. There is time before trial to use other methods if the software fails.

Entry details

Sector

Legal services / e-discovery.

Purpose

A vendor that the judgment does not name built the predictive coding software to sort documents as relevant or not, after learning from a sample that a senior lawyer reviews.

Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). English court approves predictive coding for 3 million documents (AIBD-0830). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0830.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions