Muckle LLP

Muckle LLP, a commercial law firm based in Newcastle upon Tyne, faced a breach of warranty dispute for an aerospace client involving roughly one million documents and a looming production deadline. The dense mix of complex contracts, financial models, and spreadsheets threatened to overwhelm manual review without compromising quality or straining the client's resources.

Muckle engaged Advanced Discovery's eReview service to run Technology Assisted Review (TAR/predictive coding) on the Relativity platform. Advanced Discovery first culled the data set using near-duplicate analysis, domain parsing, and concept clustering to strip out non-responsive files, then trained the TAR model through iterative rounds of Muckle attorney coding decisions, validated by quality-control checks.

After culling, the working set stood at 660,000 documents; TAR identified approximately 35,000 of them as likely responsive, meaning 95% of the set was defensibly excluded from manual review. The Muckle team completed disclosure preparation two weeks ahead of its deadline.

• Predictive coding, paired with iterative QC validation rounds, can defensibly eliminate the vast majority of a document set from manual review without sacrificing quality.
• Pre-TAR culling (near-duplicate analysis, domain parsing, concept clustering) removes non-responsive volume before predictive coding training even starts.
• TAR can compress eDiscovery timelines enough to beat court deadlines by weeks, not just reduce review cost.