KPMG Achieves 90%+ Accuracy in Contract Review with Relativity Contracts

How did they do it?

• Replaced manual, pattern-based extraction with custom AI prompts
• Used confidence scoring and rationale statements to strengthen quality control
• Streamlined review to improve accuracy and free up team resources

A Contract Conundrum

A multinational firm and KPMG client faced a familiar problem: mountains of agreements, all formatted differently, all demanding precise extraction. Typically, a project like this would call for eyes-on review and manual data extraction – a laborious and costly process that would not only take far too long; it would limit the firm’s reporting and insights.

Fortunately, KPMG had a better tool for the job: Relativity Contracts, an AI-powered solution designed for complex, high-stakes scenarios like this one.

With confidence scoring, rationale statements, and flexible prompting layered into their workflow, the team built a quality control process that was not only stronger, but smarter, than the manual approach. Relativity Contracts combined AI with rule-based checks and human oversight, allowing the team to adapt quickly to new document types and challenges.

“The turning point came when we transitioned to custom AI prompts in Relativity Contracts. The AI extractions worked significantly better than the custom RegEx extractions,” said Sean Hora, director In the Forensic Technology Practice at KPMG US.

Faster Reviews, Higher Accuracy

Once KPMG’s client embraced the AI-driven workflows, the pace of work accelerated dramatically. What once required slow, manual checks became an iterative, collaborative process between people and technology.

With Relativity Contracts, KPMG replaced the painstaking task of writing regular expressions per field and instead created prompts using natural language. And, because Relativity Contracts provides detailed rationale behind its decisions, the team was able to identify any incorrect assumptions the AI made in its decisions. In doing so, KPMG scoped down the universe of documents needing to be reviewed during second pass.

The impact was clear:

• Accuracy improved dramatically, with some areas reaching near-perfect levels.
• Review effort dropped significantly, freeing resources for higher-value work.
• AI reliably handled structured tables and variable formats that previous methods struggled to identify and extract.
• Client adoption and trust increased, thanks to side-by-side comparisons of manual versus AI-assisted review.

“After implementing Relativity Contracts’ custom AI prompt capabilities, we saw a dramatic improvement in first-pass extraction accuracy – in some cases jumping from 25% to 90%,” Sean said.

Beyond this project, Relativity Contracts has enabled KPMG to transform contract review from a tedious burden into a fast, scalable, and modern solution.

“With Relativity Contracts at the core of our workflow, we can now evolve our review processes faster, deliver insights sooner, and confidently address increasingly complex contract analysis needs,” Sean said. “The long-term impact is a more agile, data-driven, and future-ready review model for our teams and the clients we support.”