Issue link: https://epubs.iltanet.org/i/43390
narrow the scope and number of custodians as they progress through the matter and learn more about what information is likely to be relevant to the dispute. At the outset of the matter, when the organization first has knowledge of the actual or potential dispute and becomes subject to an obligation to preserve evidence, the scope of the dispute may not be fully understood. The universe of custodians subject to preservation should be defined fairly broadly, so that potentially relevant evidence is not lost. But at the time documents are actually collected, when presumably the organization and its counsel know more about the matter, the scope of custodians collected from should be narrowed if possible. When the collected data is sent for processing and review, parties should attempt to reduce the number of custodians even further. This approach requires a constant, strategic evaluation of the case as the investigation and defense or prosecution of the matter unfolds. It also requires that counsel knowledgeable about the substance of the case communicate regularly with those who can provide technical knowledge of the scope and size of the potentially relevant dataset. Without that knowledge transfer, decisions about how the document population is processed and reviewed for production will not effectively reflect the current scope of the case and cannot maximize cost-reduction. Also, litigants should actively engage and coordinate with the opposing side about which custodians are subject to discovery. Getting agreement on which custodians are in play is the easiest way to manage the number of custodians. If the litigation forum requires meet-and-confer on these issues (as federal courts do), parties and counsel should treat that process as a significant opportunity to lower costs. VALIDATE THE SEARCH TERMS USED IN THE PROCESS A fairly common strategy for bringing down the cost of document discovery is to reduce the size of the dataset for review by applying search terms to the data and reviewing only those documents that come up as potentially relevant. However, the use of search terms must meet two criteria. • The terms must be effective in reducing the size of the dataset by returning only those documents likely to be relevant (i.e., the number of "false positives" must be minimized). • The terms must be accurate enough that potentially relevant documents are not missed. In other words, the concepts of "recall" (the fraction of relevant documents in the dataset identified by a search) and "precision" (the fraction of documents identified by a search that are in fact relevant) should be applied to the use of search terms. To validate that proposed search terms have sufficient recall and precision, they should be tested. In most cases, the search terms will be refined multiple times. As false positives in the search results are analyzed, the searches should be adjusted and several iterations of the searches run to increase the precision. Testing recall by making sure relevant documents are not being missed by the search terms is more difficult, but equally important. One method of testing recall is to identify some number of definitely relevant www.iltanet.org Corporate Law Departments 33