E-Discovery Practice Pointer: Don’t Stipulate To Search Terms Until You Count Your Hits

February 8, 2012

Download   

Share Page

Are you eager to review 95 million pages worth of electronically stored information (ESI) for privileged material before producing it to your opponent in litigation? Neither was the plaintiff in I-Med Pharma Inc. v. Biomatrix, Inc., Case No. 03-3677 (D.N.J. Dec. 11, 2011). But it came perilously close to having to do just that before the court relieved it of the burden of a discovery stipulation that it “should have known better than to agree to…”

What Happened in The Case

Embroiled in a contract dispute with its former supplier, biochemical firm I-Med Pharma, Inc. entered into a formal discovery stipulation agreeing to search its entire computer system for 58 keywords and produce whatever ESI the search returned. I-Med never expected these terms to generate 64,382,929 hits, comprising upwards of 95 million pages of documents in its computer system’s unallocated space alone. Unallocated space is space that is outside designated saved files and is not write-protected – in other words, it consists of the broken up fragments of old files that have been deleted, but not yet over-written.

I-Med asked to be relieved from the overwhelming burden of reviewing this mountain of ESI, and the court ultimately agreed to let it off the hook. I-Med’s opponent, Biomatrix, argued that it had spent thousands of dollars on the search and that it should get the resulting material unless I-Med could show “exceptional circumstances” and “manifest injustice.” But the court pointed out that Biomatrix could get the cost of search back under the terms of its order, and this was no reason to waste millions more dollars on a privilege review of material that Biomatrix had not demonstrated was likely to be relevant. Biomatrix had not put on any evidence that relevant material had been deleted and was, therefore, likely to be in unallocated computer space.

The court also joined the growing number of courts explicitly considering proportionality in e-discovery, helpfully reminding litigants to consider “whether the number of results obtained could be practically reviewed given the economics of the case and the amount of money at issue.”

What We Can Learn From It

I-Med got lucky. They found themselves in an e-discovery savvy court that applied some forgiving good sense to the situation. Future litigants should not rely on that kind of luck, but instead should heed the good advice the court gave.

Don’t Stipulate Until You Run Your Search. I-Med stipulated to produce the entire universe of search results for 58 keywords before running the search. Agreeing with your opponents to use certain search terms is entirely appropriate, but agreeing in advance to produce everything they return is not. Run a preliminary search to determine how many hits there are. If there are an overwhelming number, confer again to modify the search terms or narrow the scope of the search. Review a randomly selected sample of returned documents to check the ratio of relevant to irrelevant documents. Even if the parties cannot agree, and have to get the court to help them decide on an appropriate way to cut down the results, that is a better position to be in than having to ask the court to relieve you from a stipulation you unwisely agreed to.

Watch For Overbroad Terms in Your Search. I-Med should have expected its search would turn up a mountain of irrelevant results, because it included relatively generic terms like “credit”, “manufactur*”, and “contract*”. As the court aptly noted, a better search would have used Boolean operators like “and”, “not”, or “near” to tether these generic terms to terms like “proper names, uncommon abbreviations, or other terms unlikely to occur in irrelevant documents.”

Search Only Places Likely To Have Relevant Material. I-Med agreed to search its entire computer system, even unallocated space, even though there was no evidence of spoliation. This was a recipe for disaster. As the court pointed out, it would have been much better to restrict the search to specific computers, file systems, document custodians, and a range of dates to make sure the search turned up more wheat and less chafe.