MEMORANDUM AND ORDER
Too often, discovery is not just about uncovering the truth, but also about how much of the truth the parties can afford to disinter. As this case illustrates, discovery expenses frequently escalate when information is stored in electronic form.
The plaintiffs are black concert promoters who contend that they have been frozen out of the market for promoting events with white bands by the discriminatory and anti-competitive practices of the defendants. The defendants fall into two categories: some are booking agencies that represent white artists and allegedly steer their clients away from the plaintiffs; others are promoters like the plaintiffs, but they purportedly collude with the booking agency defendants to monopolize the concert industry.
During discovery in this action, each defendant has responded to the plaintiffs’ requests by permitting inspection of its concert files, which contain documents relating to the promotion of concerts. Four sets of defendants, however, have now moved pursuant to Rules 26(b)(2)(iii) and 26(e) of the Federal Rules of Civil Procedure for a protective order relieving them of the obligation of producing electronic mail, commonly known as e-mail, that may be responsive to the plaintiffs’ discovery requests.
*424 Background
A. The Discovery Demands
The plaintiffs’ document demands are sweeping. For example, they demand production of all documents concerning any communication between any defendants relating to the selection of concert promoters and bids to promote concerts. (Plaintiffs’ First Request for Documents (“PI. Doc. Req.”), attached as Exh. A to Declaration of Richard G. Primoff dated Oct. 26, 2001 (“Primoff Dec!.”), Request No. 1). Similarly, the plaintiffs have requested “[a]ll documents concerning the selection of concert promoters, and the solicitation, and bidding processes relating to concert promotions.” (PL Doc. Req., Request No. 7). They have also demanded “[a]ll documents concerning market shares, market share values, market conditions, or geographic boundaries in which any ... concert promoter operates.” (Pl. Doc. Req., Request No. 12). These are but three examples of the thirty-five requests made in the plaintiffs’ first document demand.
B. The Defendants ’ Motions
• Each of the moving defendants contends that it should be relieved of the obligation of producing e-mail responsive to the plaintiffs’ requests because the burden and expense involved would far outweigh any possible benefit in terms of discovery of additional information. If production is nevertheless required, the defendants ask that the plaintiffs bear the cost. Because the burden to each defendant depends upon the specific structure of its e-mail retention and on the related means for retrieving and producing responsive e-mails, each defendant’s position will be outlined separately.
1. William Moms Agency
The William Morris Agency, Inc. (“WMA”), one of the booking agency defendants, argues that the chances are small that a search of its e-mails would turn up responsive documents. According to its Senior Vice President, WMA’s music agents have historically conducted business by telephone and fax and have been slow to utilize e-mail. (Affidavit of David Snyder dated Sept. 20, 2001 (“Snyder Aff.”), H 2). Moreover, to the extent an e-mail was deemed important, it would likely have been printed and saved in the appropriate concert file — files that have been produced for inspection by the plaintiffs. (Snyder Aff. H 2).
. WMA further contends that the production of its e-mails would be exorbitantly expensive and, to some extent, a technical impossibility. Prior to May 1998, WMA’s Music Division had utilized Quickmail for Macintosh for its e-mail communications. (Affidavit of John F. Porter dated Sept. 20, 2001 (“Porter Aff.”), 116). The e-mail files were backed up using a software program called Retrospect that is no longer commercially available. (Porter Aff. H13). Consequently, WMA has neither the computer hardware nor the software to read these tapes. (Porter Aff. 1( 14). And, although WMA has given them to an outside vendor for examination, that vendor has thus far been unable to recover enough data to determine what is stored on the tapes. (Porter Aff. 1115).
In May 1998, WMA’s Music Department converted to Lotus Notes for e-mail communications. (Porter Aff. 116). It backs up its e-mail files along with other electronic files such as word processing and spreadsheet documents five times each week, using a software program called Arcserve (Porter Aff. 11118, 9).
The plaintiffs agreed, at least as a first cut, to limit their discovery demands to e-mail generated or received by 56 WMA employees located in the defendant’s New York and Beverly Hills offices. (Porter Aff. 114). Likewise, they proposed an initial search limited to a sample of one back-up session for each quarter of 1998 and 1999, for a total of eight sessions. (Affidavit of Sandra C. McCallion dated Sept. 19, 2001, H10). According to WMA, in order to comply with this request, it must engage in a three-step process: cataloguing, restoring, and processing. (Porter Aff. 1118). Cataloguing involves identifying the tapes that contain the mailbox files of the designated employees and marking them for restoration. (Porter Aff. UK 18, 19). Restoration consists of saving all emails from the identified files to a master database and then removing all duplicates. *425(Porter Aff. 1121). Finally, each file must be processed so that it is not only readable on a computer screen, but also may be printed and Bates-stamped. (Porter Aff. U 24). Where an e-mail contains an attached file such as a word processing document, WMA proposes converting the attachment into a Tagged Image File Format or “TIFF” file. (Porter Aff. H 27). According to WMA, this would be necessary in order to make any redactions. (Reply Declaration of Sandra C. McCallion dated Nov. 12, 2001, U12).
WMA obtained an estimate from Fios, Inc., a computer consultant, of the cost of such an undertaking. (Porter Aff., Exh. 1). Fios projects that the cost for eight selected back up sessions would include $7,864 for cataloguing, $8,960 for restoration, and $379,120 for processing, for a total of $395,944. (Porter Aff. H 28 & Exh. 1). If the e-mails on all of the back-up tapes were produced instead of a sample of eight sessions, the total cost would mushroom to almost $9,750,000. (Porter Aff. 1128).
2. Monterey Peninsula Artists
Like WMA, Monterey Peninsula Artists, Inc. (“Monterey”) is a booking agency. It currently employs nine agents, each of whom is supplied with a personal Macintosh computer. (Declaration of Dan Weiner dated Sept. 20, 2001 (“Weiner Deck”), 113). The agents’ personal computers use a variety of different e-mail programs, so that all files cannot be reviewed by a single search program. (Declaration of Tary McConnell dated Sept. 19, 2001 (“McConnell Deck”), If 4). In total, there are almost 200,000 e-mails stored on the hard drives of these computers. (McConnell Deck 115).
According to the President of Monterey, the company still does ninety percent of its business by means other than e-mail, including telephone and fax. (Weiner Deck 114). It is estimated that fifty to seventy percent of the use of e-mail consists of alerting agents about telephone calls received by a receptionist. (Weiner Deck 116). In addition, some portion of the e-mail relates to advice of counsel. (Weiner Deck If 8). Significant e-mails may be printed as hard copy and placed in the relevant files. (Weiner Deck 117).
Monterey Computer Corporation (“MCC”) has provided three alternative cost estimates for producing the e-mail of Monterey’s agents. First, a computer operator could be employed to retrieve and print all e-mails from each personal computer. (McConnell Deck 117). This is a labor-intensive option estimated to cost $84,060. (McConnell Deck 11118-11). Alternatively, a programmer could archive all e-mails on the computer in which they are located and then print them more efficiently in “batch” mode. (McConnell Deck fill 12, 15). MCC projects that this option would entail approximately $57,860 in costs. (McConnell Deck Iflf 14-16). Finally, an information systems analyst could import all of the agents’ e-mail into a single common format, creating a single database. The entire database could then be reviewed using one search engine. (McConnell Deck 1117). Assuming that all of the e-mails would nevertheless be printed, MCC estimates the cost of this third option to be $43,110. (McConnell Deck 111119-22). None of the alternatives proposed by MCC require the creation of TIFF files. (Reply Declaration of Tary McConnell dated Nov. 8, 2001,116).
Monterey has also proffered an estimate of the cost involved in reviewing the e-mails for privilege. By using paralegals to review the production at the rate of two e-mails per minute at fees of $150 per hour, it is estimated that the privilege analysis would cost approximately $247,000. (Declaration of Matthew P. Kanny dated Sept. 20, 2001, 118).
3. Creative Artist Agency
The third moving defendant, Creative Artist Agency, LLC (“CAA”), is also a booking agency. It currently employs approximately 50 music agents and assistants. (Declaration of Michael Keithley dated Oct. 1, 2001 (“Keithley Deck”), attached as Exh. B to Creative Artist Agency, LLC’s Notice of Motion for a Protective Order (“CAA Notice of Motion”), H 10(c)). These personnel use Microsoft Exchange software for e-mail. (Keithley Deck 118). Since January 1999, CAA has backed up its corporate electronic data including e-mails on a daily basis, using *426either Backup Exec or Omniback II software. (Keithley Decl. 1114, 7). It has accumulated a total of 523 back-up tapes for this period, of which 261 have been catalogued and contain e-mails. The remaining 262, ten of which may predate 1999, have not been catalogued and may or may not include emails. (Keithley Dec. 115, 6).
Prior to 1999, CAA’s computer system went through a series of transformations. From 1993 through March 1996, CAA used a Mac OS-based system. (Keithley Deck 115). From March 1996 through August 1997, the company converted to a Windows NT 4.0 system. (Keithley Deck 116). CAA first acquired access to the Internet and external e-mail capacity in February 1996. (Keithley Deck 120). Until January 1999, CAA backed up its corporate data on a daily basis but only saved the last week of each month for archival purposes. (Keithley Deck 121). CAA has located 376 back-up tapes from its Mac system and 64 digital audio tapes (“DAT’s”) with pre-1999 back up data, of which 18 contain e-mails. (Keithley Deck 1122, 23, 26, 29). The remaining 46 DAT’s and the other back-up tapes have not been catalogued and may or may not contain email data. (Keithley Deck 1123, 26, 29).
As with WMA, production of the e-mail would require cataloguing the tapes for which this has not been done, then restoring and retrieving the e-mails. (Keithley Deck 110). CAA’s Director of Information Technology estimates for the period from 1999 on, this process would cost $187,500 to $237,500 for labor and $48,000 for necessary hardware and software and would take two and one-half years to complete. (Keithley Deck 1111,12).
With respect to the pre-1999 period, it would cost $25,000-$31,000 to retrieve the emails from the DAT’s and $78,500 to restore them from the Mac back-up tapes. (Keithley Deck 1132-33). CAA thus estimates that its total cost would be a minimum of $395,000. (Keithley Deck 134).
In addition, like Monterey, CAA has calculated the cost of conducting a privilege review. It estimates that one employee earning $60,000 per year would take two years to complete the task, for a total cost of $120,000. (Declaration of Michael Rubel dated Sept. 20, 2001, attached as Exh. C to CAA Notice of Motion, 18).
4. SFX and QBQ
The last group of moving defendants consists of QBQ Entertainment, a booking agency, along with twenty-two concert promoters, collectively referred to as “SFX.”1 The emails for this group are contained in SFX’s exchange server, 23 back-up tapes for SFX’s exchange server, 24 back-up tapes for the exchange server of Contemporary Productions, Inc., and the personal computers of 126 e-mail users. (Declaration of Kevin Crawford dated Oct. 2, 2001,14).
Ontrack Data International, Inc. (“On-track”) provided an estimate of the cost of retrieving the e-mail from these sources. The total estimate, including printing costs, management fees, and travel expenses, comes to over $403,000. (Declaration of Deborah Hagen dated Oct. 2, 2001 (“Hagen Deck”), 17). Of this, approximately $126,000 is attributable to the cost of creating TIFF files, based on the assumption that 10,000 - pages would be retrieved for each of the 126 e-mail users. (Hagen Deck 116(e), 7).
C. The Plaintiffs ’ Response
The plaintiffs argue first that the discovery of e-mail is critical to their case. They dispute the assumptions that little business is conducted by the defendants through e-mail and that important e-mails would have been reduced to hard copy in any event. (Declaration of Richard G. Primoff dated Oct. 26, *4272001 (“Primoff Decl”), UU 20-25). Furthermore, the plaintiffs maintain that the paper discovery taken to date supports their claims that the defendants have engaged in discriminatory and anti-competitive practices. For example a CAA memorandum refers to how the head of its music department kept the “black promoter issue” “under control” in connection with a tour by Janet Jackson in 1998. (Primoff Decl., Exh. K). The plaintiffs interpret this to mean that CAA deterred black promoters from participating in the tour by imposing on them more onerous financial terms than were offered to white promoters. (Primoff Decl. UU11-14 & Exhs. G, H, I, J). Similarly, the plaintiffs arg-ue that a letter by Beaver Productions, a white-owned promotion company, advocating that the City of Memphis should not enter into an exclusive contract with a black promoter, illustrates the collusive relationship between white promoters and white booking agencies. (Primoff Decl. If 16,17 & Exh. M).
The plaintiffs also contend that the defendants’ cost estimates are wildly inflated. In general terms, the plaintiffs propose to reduce the expense of e-mail production by: (1) identifying key personnel rather than retrieving the e-mail of all employees; (2) restoring only a portion of the archival tapes, based on date restrictions and sampling; (3) producing e-mail in electronic rather than paper form; and (4) conducting automated searches for privilege and responsiveness. (Primoff Decl. U26). The plaintiffs would also forego the use of TIFF files, which they argue is unnecessary and unduly expensive. (Primoff Decl. 1129). According to one of the plaintiffs’ consultants, Charles R. Kellner of Electronic Evidence Discovery, Inc. (“EED”), TIFF files are not themselves searchable, since they are merely graphic depictions of electronic text. (Affidavit of Charles R. Kellner dated Oct. 26, 2001 (“Kellner Aff.”), 111124, 35). Because they are not searchable, they do not facilitate the removal of duplicate documents from a file. (Kellner Aff. HU 29, 30). And, as an image of an electronic document, the TIFF version contains less information than the original document. (Kellner Aff. U 33). These principles underlie the plaintiffs’ cost estimates with respect to production by each of the moving defendants.
1. WMA
The plaintiffs agree to limit production of WMA’s e-mails to a sample of eight sessions of the archive tapes for 30 e-mail accounts. (Primoff Decl. U 34(d); Kellner Aff. U 63). By also eliminating conversion to TIFF files, they estimate that WMA could provide responsive e-mails at a cost ranging from $24,000 to $87,000 depending upon assumptions about the volume of e-mail and the amount of duplication. (Primoff Decl. U 34(d); Kellner Aff. U 62).
2. Monterey
The plaintiffs’ second expert, Andrew S. Rosen of ASR Data Acquisition & Analysis, LLC (“ASR”), proposes to create a “mirror image” of the hard drive of each personal computer at Monterey. (Affidavit of Andrew S. Rosen dated Oct. 18, 2001 (“Rosen Aff.”), U10). The plaintiffs would then identify potentially responsive directories and files, convert them to a common format, and create a single database that could be searched for privilege and responsiveness. (Rosen Aff. UU 10, 11). ASR estimates the cost for this project to be $10,000-$15,000. (Rosen Aff. UU 10,17).
3. CAA
By engaging again in a sampling process, the plaintiffs estimate that they could reduce the cost of production of CAA’s pre-1999 email to $20,000. (Kellner Aff. U 46). Likewise, by limiting production for 1999 to quarterly samples for 30 e-mail accounts, they project costs of $40,000-$50,000. (Kellner Aff. U 43). Thus, the plaintiffs’ total cost estimate for production by CAA is $60,000-$70,000. (Primoff Decl. U 34).
4. SFX and QBQ
Apparently, the plaintiffs are willing to forego any review of the back-up tapes of SFX and QBQ. Instead, they propose to work from the hard drives of approximately 60 users who store e-mail on their individual computers. (Kellner Aff. UU 17,19). As with Monterey, the plaintiffs would take an image *428of the hard drive and create a database of potentially responsive files. They estimate that.this would cost $64,000. (Kellner Aff. H 19; Primoff Decl. U 34).
Discussion
A. Discoverability and Relevance
The plaintiffs have successfully demonstrated that the discovery they seek is generally relevant. Although the defendants vigorously contest the plaintiffs’ interpretation of the documents that have already been produced (Declaration of Robert Light dated Nov. 9, 2001, UK 4-37; Declaration of Carole Kinzel dated Nov. 9, 2001 (“Kinzel Decl.”), H1119-23; Reply Declaration of Matthew P. Kanny dated Nov. 8, 2001, 11 6 & Exh. C), those documents are plainly pertinent to the plaintiffs’ claims. To the extent that the defendants’ e-mails contain similar information, they are equally discoverable. Electronic documents are no less subject to disclosure than paper records. See Simon Property Group L.P. v. mySimon, Inc., 194 F.R.D. 639, 640 (S.D.Ind.2000); Playboy Enterprises, Inc. v. Welles, 60 F.Supp.2d 1050, 1053 (S.D.Cal.1999); Daewoo Electronics Co. v. United States, 650 F.Supp. 1003, 1006 (CIT 1986); Bills v. Kennecott Corp., 108 F.R.D. 459, 461 (D.Utah 1985).
Nor are the defendants’ claims that the email is unlikely to yield relevant information persuasive. General representations by WMA and Monterey that their employees do little business by e-mail are undocumented and are contradicted by data proffered by these same defendants. Monterey, for example, estimates that its eight computers contain 198,000 e-mail messages (McConnell Decl. H 5), while WMA’s figures lead to the conclusion that each of its agents sent or received about 43 e-mails per day. (Kellner Aff. HIT 72, 73). It is probable that some significant portion of this traffic related to the conduct of business.
Furthermore, the supposition that important e-mails have been printed in hard copy form is likewise unsupported. In general, nearly one-third of all electronically stored data is never printed out. See Corinne L. Giacobbe, “Allocating Discovery Costs in the Computer Age: Deciding Who Should Bear the Costs of Discovery of Electronically Stored Data,” 57 Wash. & Lee L.Rev. 257, 259 (2000). Here, the defendants have not alleged that they had any corporate policy defining which e-mail messages should be reduced to hard copy because they are “important.” Finally, to the extent that any employee of the defendants was engaged in discriminatory or anti-competitive practices, it is less likely that communications about such activities would be memorialized in an easily accessible form such as a filed paper document.
The defendants’ concern about privacy is also unavailing. To the extent that the corporate defendants’ own privacy interests are at issue, the are adequately protected by the confidentiality order in this case. To the degree the defendants seek to assert the privacy concerns of their employees, those interests are severely limited. Although personal communications of employees may be appear in hard copy as well as in electronic documents (see, e.g., Kinzel Decl. H12), the defendants made no effort to exclude personal messages from the search of paper records conducted by plaintiffs’ counsel. Moreover, an employee who uses his or her employer’s computer for personal communications assumes some risk that they will be accessed by the employer or by others.
Thus, there is no justification for a blanket order precluding discovery of the defendants’ e-mails on the ground that such discovery is unlikely to provide relevant information or will invade the privacy of non-parties.
B. Cost-Shifting
The more difficult issue is the extent to which each party should pay the costs of production. “Under [the discovery] rules, the presumption is that the responding party must bear the expense of complying with discovery requests[.]” Oppenheimer Fund, Inc. v. Sanders, 437 U.S. 340, 358, 98 S.Ct. 2380, 57 L.Ed.2d 253 (1978). Nevertheless, a court may protect the responding party from “undue burden or expense” by shifting some or all of the costs of production to the requesting party. Id. (citing Fed.R.Civ.P. 26(c)). Here, the expense of locating and extracting responsive e-mails is substantial, *429even if the more modest estimates of the plaintiffs are credited. Therefore, it is appropriate to determine which, if any, of these costs, are “undue,” thus justifying allocation of those expenses to the plaintiffs.
1. Production
One line of argument, adopted by the plaintiffs, holds that the responding party should bear the 'costs of producing electronic data since “if a party chooses an electronic storage method, the necessity for a retrieval program or method is an ordinary and foreseeable risk.” In re Brand Name Prescription Drugs Antitrust Litigation, Nos. 94 C 897, MDL 997, 1995 WL 360526, at *2 (N.D. Ill. June 15, 1995); see also Daewoo, 650 F.Supp. at 1006 (“The normal and reasonable translation of electronic data into a form usable by the discovering party should be the ordinary and foreseeable burden of a respondent in the absence of a showing of extraordinary hardship.”). But even if this principle is unassailable in the context of paper records, it does not translate well into the realm of electronic data. The underlying assumption is that the party retaining information does so because that information is useful to it, as demonstrated by the fact that it is willing to bear the costs of retention. That party may therefore be expected to locate specific data, whether for its own needs or in response to a discovery request. With electronic media, however, the syllogism breaks down because the costs of storage are virtually nil. Information is retained not because it is expected to be used, but because there is no compelling reason to discard it. And, even if data is retained for limited purposes, it is not necessarily amenable to discovery. Back-up tapes, for example,
are not archives from which documents may easily be retrieved. The data on a backup tape are not organized for retrieval of individual documents or files, but for wholesale, emergency uploading onto a computer system. Therefore, the organization of the data mirrors the computer’s structure, not the human records management structure, if there is one.
Kenneth J. Withers, “Computer-Based Discovery in Federal Civil Litigation,” SF97 ALI-ABA 1079, 1085 (2001); see also McPeek v. Ashcroft, 202 F.R.D. 31, 32 (D.D.C.2001) (“The purpose of having a backup system and retaining the tapes was to permit recovery from a disaster, not archival preservation.”). Thus, it is not enough to say that because a party retained electronic information, it should necessarily bear the cost of producing it.
The contrary argument is that the requesting party should bear the burden since, when the costs of discovery are internalized, that party can perform a cost-benefit analysis and decide whether the effort is justified. See Marnie H. Pulver, “Electronic Media Discovery: The Economic Benefit of Pay-Per-View,” 21 Cardozo L.Rev. 1379, 1424 (2000). Yet, this “market” approach has two shortcomings. First, it flies in the face of the well-established legal principle, cited above, that the responding party will pay the expenses of production. Second, it places a price on justice that will not always be acceptable: it would result in the abandonment of meritorious claims by litigants too poor to pay for necessary discovery.
Because of the shortcomings of either bright-line rule, courts have adopted a balancing approach taking into consideration such factors as: (1) the specificity of the discovery requests; (2) the likelihood of discovering critical information; (3) the availability of such information from other sources; (4) the purposes for which the responding party maintains the requested data (5) the relative benefit to the parties of obtaining the information; (6) the total cost associated with production; (7) the relative ability of each party to control costs and its incentive to do so; and (8) the resources available to each party. Each of these factors is relevant in determining whether discovery costs should be shifted in this case.
a. Specificity of Bequests
The less specific the requesting party’s discovery demands, the more appropriate it is to shift the costs of production to that party. See In re General Instrument Corp. Securities Litigation, No. 96 C 1129, 1999 WL 1072507, at *6 (N.D.Ill. Nov. 18, 1999) (denying motion to compel production *430of e-mails where requesting parties “have not identified any specific factual issue for which additional discovery would help them prove their case”). Where a party multiplies litigation costs by seeking expansive rather than targeted discovery, that party should bear the expense.
As noted above, the plaintiffs’ demands in this case are extremely broad. They stand in sharp contrast to cases such as Daewoo where the court declined to shift costs. There the plaintiff sought only specific data sets that were utilized in the administrative review that resulted in the challenged governmental order. Daewoo, 650 F.Supp. at 1004-05. Similarly, in McPeek the court did not shift costs but required production of only the e-mails of specific persons who purportedly retaliated against the plaintiff. 202 F.R.D. at 33. Here, the plaintiffs’ requests are far more nebulous, a factor that favors shifting the costs of discovery to them.
b. Likelihood of a Successful Search
In McPeek, the court utilized the concept of marginal utility in determining whether to shift costs:
The more likely it is that the backup tape contains information that is relevant to a claim or defense, the fairer it is that the [responding party] search at its own expense. The less likely it is, the more unjust it would be to make [that party] search at its own expense. The difference is “at the margin.”
202 F.R.D. at 34. Here, there is a high enough probability that a broad search of the defendants’ e-mails will elicit some relevant information that the search should not be precluded altogether. However, there has certainly been no showing that the e-mails are likely to be a gold mine. No witness has testified, for example, about any e-mail communications that allegedly reflect discriminatory or anti-competitive practices. Thus, the marginal value of searching the e-mails is modest at best, and this factor, too, militates in favor of imposing the costs of discovery on the plaintiffs.
C. Availability From Other Sources
Some cases that have denied discovery of electronic evidence or have shifted costs to the requesting party have done so because equivalent information either has already been made available or is accessible in a different format at less expense. In Anti-Monopoly, Inc. v. Hasbro, Inc., No. 94 Civ. 2120, 1996 WL 22976, at *1 (S.D.N.Y. Jan. 23, 1996), the defendant had already produced the requested data in hard copy. However, the plaintiff sought the same information in electronic form, presumably to facilitate computerized analysis. While recognizing that prior production in one form did not foreclose the plaintiffs demand, the court held that “[if] plaintiff wants the computerized information, it will have to pay defendants’ reasonable costs of creating computer programs to extract the requested data from defendants’ computers.” Id. Similarly, in Williams v. El. du Pont de Nemours & Co., 119 F.R.D. 648, 649-50 (W.D.Ky.1987), the defendant had originally supplied data to the EEOC, a plaintiff in the action, in hard copy form, and then demanded disclosure of the computer files created by the agency from that data. The court granted the discovery, but required the defendant to pay a reasonable share of the costs that the EEOC had incurred in formulating the database. Id. at 651.
In the instant case there has been no showing that the defendants’ e-mails are generally available other than by a search of the defendants’ hard drives or back-up tapes. The representations that “important” e-mails were probably printed out are entirely speculative. Accordingly, this consideration favors requiring the defendants to produce the emails at their own expense.
d. Purposes of Retention
If a party maintains electronic data for the purpose of utilizing it in connection with current activities, it may be expected to respond to discovery requests at its own expense. Under such circumstances, the guiding principle is that “information which is stored, used, or transmitted in new forms should be available through discovery with the same openness as traditional forms.” *431 Daewoo, 650 F.Supp. at 1006. A party that expects to be able to access information for business purposes will be obligated to produce that same information in discovery.
Conversely, however, a party that happens to retain vestigial data for no current business purposes, but only in case of an emergency or simply because it has neglected to discard it, should not be put to the expense of producing it. In this case, the back-up tapes clearly fall into this category. There is no evidence that the defendants themselves ever search these tapes for information or even have a means for doing so. Cost-shifting is therefore warranted with respect to the back-up tapes. See Withers, supra, at 1085 (“retrieving documents from backup tapes may be conditioned on the requesting party paying some or all of the costs”).
The same is true of e-mails which, although deleted from the user’s active files, remain on the hard drive. “Aside from the occasional practice of ‘dumpster diving,’ the discovery of deleted computer documents does not have a close analogue in conventional, paper-based discovery.” Id. Just as a party would not be required to sort through its trash to resurrect discarded paper documents, so it should not be obligated to pay the cost of retrieving deleted e-mails.2 Thus, since there has been no showing that the defendants access either their back-up tapes or their deleted e-mails in the normal course of business, this factors tips in favor of shifting the costs of discovery to the plaintiffs.
e. Benefit to the Parties
Where the responding party itself benefits from the production, there is less rationale for shifting costs to the requesting party. See Bills, 108 F.R.D. at 464. Such a benefit could come in one of two forms. First, the process of production could have collateral benefits for the responding party’s business. For example, a computer program created to conduct a search for purposes of discovery could also be useful in the regular activities of the business. Second, the responding party may benefit in litigation from the review of its own records.
Neither circumstance is present here. Since the computer data at issue is not regularly used by the defendants, cataloguing or searching it would have little business value to them. Similarly, recovery of e-mail will not benefit the defendants in this litigation since the e-mails are not relevant to any issue on which the defendants bear the burden of proof. The situation might be different if there were a universe of data that either party might be able to use to support its position: for example, hiring or promotion statistics in an employment discrimination case. But that is not the case here, and the absence of any benefit to the defendants makes cost-shifting more appropriate.
f. Total Costs
If the total cost of the requested discovery is not substantial, then there is no cause to deviate from the presumption that the responding party will bear the expense. See Oppenheimer Fund, 437 U.S. at 361-62, 98 S.Ct. 2380; Anti-Monopoly, 1996 WL 22976, at *2; Bills, 108 F.R.D. at 464. In Oppenheimer Fund, the Supreme Court found that “a threshold expense of $16,000 ... hardly can be viewed as an insubstantial burden,” even for a defendant with assets exceeding one-half billion dollars. 437 U.S. at 361-62, 98 S.Ct. 2380. In Anti-Monopoly, the court considered costs of $1,680 to one defendant and $5,000-6,000 to another to be sufficiently significant to warrant cost-shifting. 1996 WL 22976, at *2. Here, the costs of the proposed discovery would be substantial by any definition. Even the plaintiffs project that the costs for WMA would be between $24,000 and $87,000, for Monterey between $10,000 and $15,000, for CAA between $60,000 and $70,000, and for SFX and QBQ approximately $64,000. The magnitude of these expenses favors cost-shifting.
g. Ability to Control Costs
The plaintiffs have professed an ability to limit the costs of discovery of e-mails to a *432much greater extent than defendants. Of course, this factor alone does not dictate cost-shifting; the defendants could be required to pay the bill for the less expensive methodologies proposed by the plaintiffs. However, where the discovery process is going to be incremental, it is more efficient to place the burden on the party that will decide how expansive the discovery will be. Cf. Bills, 108 F.R.D. at 464 (declining to shift costs where “[t]he relative expense and bur-' den in obtaining the data would be substantially greater to the requesting party as compared with the responding party”). The plaintiffs here will be able to calibrate their discovery based on the information obtained from initial sampling. They are in the best position to decide whether further searches would be justified. See McPeek, 202 F.R.D. at 33-34 (requiring producing party to pay all costs is disincentive to requesting party to narrow its demands). This consideration, then, also militates in favor of cost-shifting.
h. The Parties’ Resources
Finally, the ability of each party to bear the costs of discovery may be an appropriate consideration. See Oppenheimer Fund, 437 U.S. at 361-62, 98 S.Ct. 2380; Bills, 108 F.R.D. at 464. In some cases, the cost, even if modest in absolute terms, might outstrip the resources of one of the parties, justifying an allocation of those expenses to the other. But in this case, all parties have sufficient resources to conduct this litigation. Although the plaintiffs argue that the defendants are some of the most powerful players in the concert promotion business, the plaintiffs purport to be able to compete with them in the marketplace. The relative financial strength of the parties, then, is at most a neutral factor.
The relevant factors thus tip heavily in favor of shifting to the plaintiffs the costs of obtaining discovery of e-mails in this case. The protocol to be followed will be addressed below.
2. Privileged and Confidential Documents
Beyond the cost of isolating and producing the requested e-mails, the defendants argue that the expense of reviewing these documents for privilege and confidentiality would be enormous. Monterey, for example, estimates that it would expend nearly $250,000, while CAA projects costs of $120,000 and two years of work. However, the sanctity of the defendants’ documents can be adequately preserved at little cost by enforcement of the confidentiality order and by the additional elements of the protocol described below, including requirements that the e-mails be reviewed on an attorneys’eyes-only basis and that review of attorney-client documents shall not be deemed a waiver of the privilege. See Playboy, 60 F.Supp.2d at 1054-55 (use of protective order and provision precluding waiver of attorney-client privilege sufficient). But see Anti-Monopoly, 1996 WL 22976, at *2 (notwithstanding confidentiality order, business data should not be viewed by competitors).
Even with such protections, however, the disclosure of privileged documents cannot be compelled. See In re Dow Corning Corp., 261 F.3d 280, 284 (2d Cir.2001). Therefore, it must be determined who should bear the costs if, notwithstanding the recommended precautions, the defendants still choose to conduct a complete review prior to production. Apparently, the defendants retained privileged or confidential documents in electronic form but failed to designate them to specific files. This situation is analogous to one in which a company fails to shred its confidential paper documents and instead leaves them intermingled with non-confidential, discoverable papers. The expense of sorting such documents is properly borne by the responding party, and the same principle applies to electronic data. Accordingly, if any defendant elects to conduct a full privilege review of its e-mails prior to production, it shall do so at its own expense.
C. Protocol
The protocol set forth here for the defendants’ production of e-mails is necessarily only a set of guidelines, and the parties are free to add detail and otherwise modify the protocol by agreement.
*433Initially, the plaintiffs shall designate one or more experts who shall be responsible for isolating each defendant’s e-mails and preparing them for review. The defendants shall have the opportunity to object to any expert so designated. The expert shall be bound by the terms of this order as well as any confidentiality order entered in the case.
With the assistance and cooperation of the defendants’ technical personnel, the plaintiffs’ expert shall then obtain a mirror image of any hard drive containing e-mails as well as a copy of any back-up tape. The plaintiffs may choose to review a sample of hard drives and tapes in lieu of all such devices.
Plaintiffs’ counsel shall formulate a search procedure for identifying responsive e-mails and shall notify each defendant’s counsel of the procedure chosen, including any specific word searches. Defendants’ counsel may object to any search proposed by the plaintiffs.
Once an appropriate search method has been established, it shall be implemented by the plaintiffs’ expert. Plaintiffs’ counsel may then review the documents elicited by the search on an attorneys’-eyes-only basis. The plaintiffs may choose the format for this review; they may, for example, view the documents on a computer screen or print out hard copy. Once plaintiffs’ counsel have identified those e-mails they consider material to this litigation, however, they shall provide those documents to defendants’ counsel in hard copy form with Bates stamps. The plaintiffs shall bear all costs associated with the production described thus far. However, the defendants shall pay for any procedures beyond those adopted by the plaintiffs, such as the creation of TIFF files.
Defendants’ counsel shall then have the opportunity to review the documents produced in order to designate those that are confidential and assert any privilege. Any purportedly confidential or privileged document shall be retained on an attorneys’-eyes-only basis until any dispute about the designation is resolved. The fact that such a document has been reviewed by counsel or by the expert shall not constitute a waiver of any claim of privilege or confidentiality.
Should any defendant elect to review its database prior to production, it shall do so at its own expense. In that event, the defendant shall review those hard drives and backup tapes selected by the plaintiffs and shall create copies from which privileged or confidential and unresponsive material has been deleted. The defendant shall then provide plaintiffs’ counsel with each “redacted” hard drive or tape, together with a privilege log identifying the documents removed. The process would then continue as described above.
Conclusion
For the reasons discussed here, the defendants’ motion for a protective order is denied insofar as it sought to preclude altogether the discovery of e-mail. It is granted to the extent that the plaintiffs shall bear the costs of production, though the defendants shall continue to be responsible for the expense of any review for privileged or confidential material. Production shall be accomplished in accordance with the protocol outlined above.
SO ORDERED.