A global chemical company found themselves embroiled in a contract dispute. As discovery progressed, the company needed to sift through a large corpus quickly and in a cost-efficient manner.
Working with an AM Law 100 firm, the end-client engaged Sandline Global to reduce the number of documents for review, prioritize those documents during review, and keep discovery costs down. Using search terms and early case assessment, including search term consultation, email threading, and file type analysis, Sandline narrowed the potential review population from over 5 million documents to fifty-one thousand. Sandline then employed NexLP continuous active learning. Counsel then conducted the seed set training, reviewing eight-hundred documents to provide relevance scores to the model.
Sandline then engaged a small review team to review the remaining documents. Using CAL for prioritization, the most likely to be responsive documents were reviewed first and brought to counsel’s attention early. Once the team reviewed half of the documents, an elusion test was run on 400 documents with a result of 1.3%, meaning out of the ~400 documents sampled, only 1.3% of them were responsive. Since there were ~24,500 documents left to be reviewed, we deduced that there were about ~320 documents likely to be responsive.
At our suggestion and in consultation with the law firm, it was determined that it would not be reasonable or proportionate to review the remaining 50% of the document population. It would be unduly burdensome, and therefore, these documents would not need to be reviewed. This saved approximately four-hundred review hours or two and a half weeks, resulting in a cost savings of approximately $19,000.