Robert Ross, Individually, and on behalf of all others similarly situated, Plaintiff, v. Abercrombie &amp; Fitch Company, et al., Defendants Consolidated Case No. 2:05-cv-0819, No. 2:05-cv-0848., 2:05-cv-0860, 2:05-cv-0879, 2:05-cv-0893, 2:05-cv-0913, 2:05-cv-0959, 2:05-cv-0964, 2:05-cv-0998, 2:05-cv-1084 United States District Court, S.D. Ohio, Eastern Division October 27, 2008 Kemp, Terence P., United States Magistrate Judge OPINION AND ORDER These securities cases are before the Court to consider a discovery dispute that has arisen concerning the retrieval of electronically-stored information, or ESI. Both parties filed briefs on the issue on October 14, 2008. Plaintiff is asking the Court to order Abercrombie to produce an additional 95,000 documents which were "hit" by a keyword search performed with terms provided by plaintiff. For the following reasons, the Court concludes that plaintiff is not entitled to the exact relief it seeks, but that it is free to pursue additional discovery of ESI. I. Here is the background. As might be expected, Abercrombie creates and stores much of its information electronically. That includes emails and information kept on individual employees' hard drives. For any given time frame, there is quite a large volume of ESI generated. At the outset of discovery, the parties discussed how best to access this information. They finally agreed to allow Abercrombie to do keyword searches to try to identify emails and other electronic documents that might be relevant to the issues here. Plaintiff prepared a list of the key words it wanted to be used and sent it to Abercrombie in January of this year. The first list contained around 120 key words. After deciding which databases to search (something plaintiffs had input into as well), and converting certain files to a format that would allow them to be searched in this fashion, Abercrombie ran the search. As expected, that search "hit" a large number of items. In an effort to focus the search, the original search terms were revised in March, and a new list of 123 terms was created. Another search was run using those terms. Once the results came in, the parties agreed that some data sets would be reviewed and produced. That has happened and Abercrombie has produced more than one million pages of documents. The parties also agreed that further refinements of the search needed to be done in order to cut down on the number of documents remaining. That was when the problem occurred. John Lowther, one of plaintiff's attorneys, wrote a letter on April 16, 2008, to John Kulewicz, one of Abercrombie's lawyers. It contained a list of "Search Term Revisions" which consisted of six search terms (Albany, Quarter, brand, distribution center, sale, and Ruehl) to be searched for their proximity to a number of other terms like "denim" or "sale" or "stock" or "SEC." The rest of the letter reads in its entirety as follows: As the parties discussed, Lead Plaintiff proposed the following search term revisions in an effort to reduce the total number of hits. As before, we expressly reserve the right to obtain additional or further searches, including, without limitations, adding or revising search terms. Abercrombie ran a new search using these six terms. It did cut down on the number of hits. The parties then worked on getting additional documents produced. At some point, however, the parties came to learn that they were operating under different assumptions about how this reduced set of documents was identified. Abercrombie assumed from the April 16, 2008 letter that it was to run an entirely new search using just the six terms (and, of course, all of the terms they were to be searched in proximity to) set out in the letter. Plaintiff apparently believed it had asked for a new search consisting of not just these six terms, but its entire previous list of 123 terms, which included these six. In other words, plaintiff viewed the letter as modifying the prior list, not replacing it. Once they realized that a miscommunication had happened, the parties agreed to have a search run with the earlier 123 terms as revised on April 16. That search "hit" almost 100,000 documents. According to Abercrombie, since the average document produced so far is over 13 pages long, this is an additional 1.3 million pages, more or less. Plaintiff wants all these documents, but is willing to look at them unreviewed if Abercrombie doesn't want to spend the time needed to screen them for privilege, and turn back any privileged documents that it might come across. Abercrombie isn't willing to agree to that procedure and is also unwilling to spend the time and money to review another million-plus pages of documents, most of which, in its view, are completely irrelevant to any issue in this case. The parties have asked the Court to resolve this impasse, which will have an impact on the progress of other discovery, like depositions, until it is resolved. II. It is important, in terms of deciding this dispute, to pick the appropriate starting point for the analysis. Plaintiff's argument (at least the way the Court reads it) is fairly simple. But for an unreasonable mistake in the way it read Mr. Lowther's letter, says plaintiff, Abercrombie would have searched these documents properly back in April, and given the relatively small size of the two sets of documents at issue (about 20,000 in one set, and about 75,000 in the other), the documents probably would have been produced by now. It would be unfair to allow Abercrombie to withhold them now just because it made a mistake back then. Abercrombie's position is a bit more nuanced. It does argue that its interpretation of Mr. Lowther's letter was entirely reasonable. However, even if it were not, and had this search been run in April rather than August, that would not change several important facts. First, Abercrombie has already reviewed and produced more than one million pages of documents. Second, a great many of the documents already produced are completely irrelevant. Third, that same problem plagues the documents that have been "hit" by the August search. Thus, the problem here is not so much that the parties' misunderstanding is what prevented these documents from being produced — Abercrombie would not have reviewed and produced them back in April, or in the succeeding months, precisely for the same reasons: volume and likely irrelevance. Abercrombie has the better of this argument. At some point, given the large amount of time and money which has been spent so far on document production, it becomes the plaintiff's burden to show that the cost to review and produce even more documents is outweighed by their likely relevance to the key issues in this case. That showing has not been made. The only things that appear likely true, from the briefs filed by both sides, are that either set of keyword searches produced a very large number of irrelevant documents, that Abercrombie has already reviewed and produced a very large number of documents, and that it will take a lot more time and money to review the documents still in dispute. Without some showing that the plaintiff is likely to get at least a moderate benefit from that process, the Court is not inclined to order it. That is not to say, however, that the plaintiff must abandon all hope at seeing some of these remaining documents. One might reasonably suspect that a person conversant with both key word searching and the types of documents produced to date could make an educated guess about why these two searches have turned up so many irrelevant communications. Armed with that information, the search could then be refined further, and perhaps tested on a sample group of documents, to see if the refined search actually both reduces the total number of documents that are "hit" and increases the percentage of those documents that have relevant information in them. If so, it would be reasonable for the plaintiff to ask to have that search run on the 95,000 documents at issue here, and to have the parties discuss how they might quickly and economically review whatever new subset of documents results from the search. In the meantime, it appears (at least from what the parties have submitted) that there ought to be a number of Abercrombie witnesses who can be deposed based on the documents that have already been produced. If a handful of new documents turns up later that might prompt additional questioning of these witnesses, the parties ought to be able to agree on a way to get that done without incurring significant additional expense. III. Based on the foregoing, the plaintiff's motion for an order requiring Abercrombie &amp; Fitch to produce the remaining e-mails and hard drive documents at issue (#263 in Case No. 2:05-cv-819) is denied. Any party may, within ten (10) days after this Order is filed, file and serve on the opposing party a motion for reconsideration by a District Judge. 28 U.S.C. §636(b)(1)(A), Rule 72(a), Fed. R. Civ. P.; Eastern Division Order No. 91-3, pt. I., F., 5. The motion must specifically designate the order or part in question and the basis for any objection. Responses to objections are due ten days after objections are filed and replies by the objecting party are due seven days thereafter. The District Judge, upon consideration of the motion, shall set aside any part of this Order found to be clearly erroneous or contrary to law. This order is in full force and effect, notwithstanding the filing of any objections, unless stayed by the Magistrate Judge or District Judge. S.D. Ohio L.R. 72.4.