eDiscovery Daily Blog
Getting the Most out of Your Keyword Searches
Though a more basic searching technique, keyword searches allow professionals to identify one or two specific words from multiple documents. Nowadays, keyword searches are considered inferior to the successor, predictive coding (TAR). In comparison to TAR, the “outdated” search method is more expensive and time-consuming. Keyword searches are also less predictable; when filtering through the same data set, keyword searches yield fewer results. Based on these flaws, some would argue that keyword searches are a dying technique. So, why bother talking about them at all? Though keyword searches have their flaws, they are far from obsolete. Some legal teams prefer to utilize manual review, recognizing it as a tried-and-true method. For example, the defendants in Coventry Capital U.S., LLC v. EEA Life Settlements, Inc. attempted to use TAR in 2020 to resolve the fraud case, but they argued the process was “protracted and contentious.” Thus, Judge Sarah L. Cave declined to compel the inclusion of TAR. [1] Similar outcomes occurred in cases such as Hyles v. New York City (2016) and In re Viagra (Sildenafil Citrate) Prods. Liab. Lit. (2016). In both cases, the court refused to mandate the usage of TAR when the responding party demonstrated a clear preference for keyword searching. [2] With this knowledge in mind, it’s important to recognize that keyword searches are still effective when done right.
Five Tips for Effective Keyword Searches
- Good communication is crucial.
Consult your custodians before running your searches. Use the conversations to identify any specific terms or abbreviations that may be relevant to your review. If necessary, you may also need to speak with an experienced advisor. Through their expertise, they can assist you with the sampling and testing process. Advisors are a great way to save time and money for everyone involved.
- Create and test your initial set of terms.
Everyone has to start somewhere. Your initial search terms don’t have to be perfect. While constructing your list, estimate how many results you expect each term to yield. Once you’ve run your test, evaluate how the search results compare to your expectations. If you received significantly fewer results than anticipated, adjust the search terms as needed. You may have to refine your search list multiple times. Anticipate this possibility to avoid missing any deadlines. [3]
- Limit searches that include wildcards and/or numbers.
When searching for words with slight differences, it’s better to search for each variation rather than use wildcards. For example, you should set up individual searches for “email” and “emails” instead of using “email*” as a search term. Numbers can also be a problem if not done correctly (i.e. searching for the number 10 will show results for 100, 1 000, etc.). Make sure to place the number in quotes to avoid this issue.
- Count the characters.
Search terms with four or fewer characters are likely to yield false hits. Short words or abbreviations like HR or IT may be identified in longer, unrelated results. Filtering out the false hits requires extra review time and money.
- Know how to search for names properly.
Avoid searching for custodian names. Their name will most likely be attached to more documents and hits than expected or desired. When searching for non-custodians, place “w/2” between their first and last name. Doing so will show all variations of the full name. Finally, consider searching for nicknames to get even more results. Ask the client what nicknames they respond to before making your search term list. [4]
[1] Doug Austin, “Court Rules for Defendant on TAR and (Mostly) Custodian Disputes: eDiscovery Case Law,” eDiscovery Today, January 12, 2021.
[2] “How Courts Treat ‘Technology Assisted Review’ in Discovery,” Rivkin Radler, March 13, 2019.
[3] “Improving the effectiveness of keyword search terms,” E-discovery Consulting, November 11, 2021.
[4] Kathryn Cole, “Key Word Searching – What Is It? And How Do I Do It (Well)?,” All About eDiscovery, December 9, 2016.