Conditions and Cost Shifting

Conditions and Cost Shifting

Rule 26(b)(2)(B) ends with a short sentence that gives a court real room to work: when it orders discovery from a source that is not reasonably accessible, it “may specify conditions for the discovery.”¹ That sentence follows two showings. The responding party has proved that the source is not reasonably accessible because of undue burden or cost, and the requesting party has shown good cause to reach it anyway. The court need not choose between ordering everything and ordering nothing. It can order the discovery and attach terms.

Two conditions recur in the cases. The first is a sample, in which the responding party restores and searches a small part of the source so that the court can decide the rest of the dispute on evidence. The second is an allocation of expense, and Rule 26(c)(1)(B) now says in terms that a protective order may specify “the allocation of expenses” for discovery.² Allocation departs from the presumption that each party bears its own discovery expenses, so a court that shifts cost must make explicit findings supported by evidence the parties provide.³

The subsections below take the conditions in the order a court reaches them: the sample, which produces the facts; the presumption against shifting; the factors courts weigh when they do shift; and a limit on a party’s power to shift a burden it created. Zubulake (2003), Wiginton (2004), and Quinby (2006) all predate the 2015 amendments to Rule 26, so some of their outcomes might differ under the current text.⁴

Sampling Before Deciding

The dispute in Zubulake began with a large estimate and an empty record. UBS Warburg said that restoring the e-mail that survived only on backup and archival media would cost about $175,000, not counting attorney time to review it, and Zubulake moved to compel production at UBS’s expense.⁵ UBS could price the work. No one could say what the tapes held, because no one had looked.

That gap puts a requesting party in a Catch-22: the requester seems to need to show the likely importance of costly e-discovery in order to get the e-discovery.⁶ The court refused to resolve the problem against the requester. It found that demanding proof that the discovery would hit paydirt was inconsistent with the lenient federal presumption of discovery and the relevancy standard.⁷ Such proof, it added, “will rarely exist in advance of obtaining the requested discovery,” and requiring a plaintiff to show that electronic discovery will yield a “gold mine” contradicts Rule 26(b)(1), which permits discovery of “any matter” relevant to a claim or defense.⁸

The court found the best solution to that problem in McPeek.⁹ It limited the discovery UBS had to provide in the first instance to a sample of just five tapes, which would give the parties and the court a basis for deciding whether more e-discovery, and more cost, would be likely to turn up relevant evidence.¹⁰ In the court’s words, “by requiring a sample restoration of backup tapes, the entire cost-shifting analysis can be grounded in fact rather than guesswork.”¹¹

The order in Zubulake I shows how the method runs. UBS had to produce all responsive e-mail on its optical disks and active servers at its own expense, and also, at its own expense, responsive e-mail from any five backup tapes selected by Zubulake. It then had to prepare an affidavit detailing the results of its search and the time and money spent, after which the court would review the tapes’ contents and UBS’s certification and conduct the cost-shifting analysis.¹⁵ Two choices stand out. Zubulake selected the tapes, so the party resisting the search did not control what the sample would show. And the sample reached only the source that raised the cost problem: the usable optical disks and active servers were produced in full, with no test run.¹⁶

The numbers a sample generates drive the next decision. Wiginton v. CB Richard Ellis, Inc. came soon after Zubulake and presented a similar problem about recovering e-mail from backup media. A test of three backup tapes showed that between 1.64 and 6.5 percent of the e-mails were responsive, and the court had an estimate of $249,000 for full restoration.¹⁷ The device also works outside backup tapes. In Oxbow, the responding party first opposed the defendants’ request on grounds of undue burden and cost, then agreed to analyze a random sample of Koch’s digital records using agreed-upon search terms, so that the dispute would rest on concrete numbers about responsiveness and the parties would have a basis for negotiating new search terms if necessary. The sample cost $57,198 and showed that 11.67 percent of the hits were responsive. Oxbow estimated that processing, reviewing, and producing the rest would cost about $85,000, for a total near $142,000, and when it declined to go further, the defendants renewed their motion to compel and the Magistrate Judge conducted a proportionality analysis.¹⁸ A sample therefore serves a proportionality dispute as well as a dispute over an inaccessible source.

A sample does not replace the requester’s own showing. The court observed that its decision in Zubulake was easier because she had already found one smoking-gun e-mail, urging that she be fired “ASAP” after her EEOC complaint in order to deprive her of year-end bonuses.¹⁹ A requester with no comparable lead is likely to find the good-cause showing harder. The same commentary explains that today’s rule fits these facts: UBS would have to show the undue cost of accessing the tapes, Zubulake could still try to show good cause, the court could use a cost-benefit analysis to specify conditions, and Rule 26(c)(1)(B) expressly authorizes allocating expenses, so Zubulake I would likely come out the same way today.²⁰

Presumption and Cost Shifting

Rule 26(c)(1)(B) is the provision that lets a court move the expense of discovery from one party to the other. It authorizes a protective order, for good cause, to protect a party from undue burden or expense by “specifying terms, including time and place or the allocation of expenses, for the disclosure or discovery.”²² The 2015 amendments made that authority explicit, and courts have enormous discretion to shape protective orders to the disputes before them.²³

The authority operates against a default. In the Supreme Court’s formulation, the responding party ordinarily bears the expense of complying with discovery requests, but it may invoke the court’s discretion under Rule 26(c) to obtain an order protecting it from undue burden or expense, including an order conditioning discovery on the requesting party’s payment of the costs.²⁴ Zubulake I applied that default to electronic evidence. Because electronic evidence is no less discoverable than paper evidence, “any principled approach to electronic evidence must respect this presumption.”²⁵ The 2015 Committee Note adds a caution: “[r]ecognizing the authority does not imply that cost-shifting should become a common practice.”²⁶ Shifting departs from the long-standing rule, so a court must base it on explicit findings supported by evidence the parties provide.²⁷

Electronic evidence does not change the default, because it is often cheaper to produce than paper: it can be searched automatically, keywords can be run for privilege checks, and production in electronic form avoids mass photocopying.²⁸ What does matter is accessibility. Whether production is unduly burdensome or expensive turns primarily on whether the data sits in an accessible or an inaccessible format, a distinction that tracks the expense of production closely, and paper is accessible when it is readily available in a usable format and reasonably indexed.²⁹ Data kept in an accessible format therefore falls under the usual rules, and the responding party pays. A court should consider cost shifting only when electronic data is relatively inaccessible, such as backup tapes.³⁰

An exception exists because the only other responses to a costly request are all or nothing. A court can hold the responder to full production or deny the discovery. Shifting offers a third path. As one court explained, when a responding party asks for protection from undue burden or expense, the court may shift the costs to the other side instead of just disallowing the discovery, and the 1970 Advisory Committee note to Rule 34 describes the court’s power to protect a respondent “either by restricting discovery or requiring that the discovering party pay costs.”³¹ Narrowing and shifting can work together, because narrowing a request under Rule 26(b)(2) does not stop the court from also granting a protective order that shifts the cost of the documents it still orders produced.³² For inaccessible ESI, Rule 26(b)(2)(B) supplies the textual hook, because it gives the court discretion to order discovery and specify cost shifting, while Rule 26(b)(2)(C) separately requires a court to limit discovery whose burden or expense outweighs its likely benefit.³³

Shifting is not automatic, and the responding party must start it. On a motion to compel or for a protective order, the party from whom discovery is sought must show that the source is not reasonably accessible because of undue burden or cost.³⁴ More generally, the party resisting production bears the burden of establishing lack of relevancy or undue burden.³⁵ A bare assertion that discovery is unduly burdensome is “ordinarily insufficient to bar production,” and one court was “not persuaded” by an objection that gave no estimate of the hours or money compliance would take.³⁶ UBS, by contrast, put a figure on restoration, about $175,000, though that figure excluded attorney review time.³⁷ The responding party holds the facts about hours, vendor charges, and systems, so the rules ask it to supply them.

Shifting is also the last of several disciplines on expensive requests. A lawyer’s signature on a discovery request certifies that it is neither unreasonable nor unduly burdensome or expensive, considering the needs of the case, prior discovery, the amount in controversy, and the importance of the issues, and Rule 26(b)(2)(C) requires a court to limit discovery whose burden or expense outweighs its likely benefit.³⁸ Courts have criticized kneejerk requests served without regard to cost, and they have urged lawyers to confer before discovery about the amount in controversy and how much, what type, and in what sequence discovery should proceed.³⁹

Allocation is easiest before the money is spent. In CBT Flint Partners, the court noted that the Northern District of Georgia’s local rule required a joint preliminary report and discovery plan addressing limits on electronic discovery, so the parties should have discussed cost shifting and burdensome discovery then and could have agreed to limit discovery or shift certain costs. Agreements to allocate costs, the court added, are permissible and enforceable.⁴⁰ CBT Flint Partners, LLC v. Return Path, Inc. also shows what late action looks like. The responding party, Cisco, sought fee shifting late, and the district court found certain documents not reasonably accessible and ordered Cisco to produce them on the condition that CBT pay $300,000 for the cost of a privilege review. CBT then chose not to pursue those documents. The court observed that Cisco could have turned to these options, and to the court, when CBT made its initial requests, which Cisco described as asking for “in effect, every document in the company.”⁴¹

The Zubulake Factors

Neither Rule 26(b)(2)(B) nor Rule 26(c)(1)(B) says when a court should move part of the cost to the party who asked for the data. The first lets the court specify conditions, and the second lets it specify terms, including the allocation of expenses.⁴³ The working standard comes from case law, and the case is Zubulake v. UBS Warburg LLC, in which a former employee sought backup e-mail that her employer said would cost about $175,000 to restore.⁴⁴ The court answered with a seven-factor test.⁴⁵ That test revises the eight-factor test from Rowe Entertainment, Inc. v. William Morris Agency, Inc., which both parties agreed should govern.⁴⁶

The Rowe test asked about the specificity of the discovery requests, the likelihood of discovering critical information, the availability of that information from other sources, the purposes for which the responding party maintains the data, the relative benefits to the parties of obtaining it, the total cost of production, each party’s ability to control costs and its incentive to do so, and each party’s resources.⁴⁷ The Zubulake court wanted the analysis neutral, with close calls resolved in favor of the presumption that the responding party pays, and it concluded that the Rowe factors, as applied, undercut that presumption for three reasons. The first was that the test is incomplete.⁴⁸ The opinion’s response has three parts: a revised list, an order of importance, and a sample to supply the facts. The sample is covered above. The list comes first.

The seven factors, in the order the court gave them, are these:⁴⁹

1. The extent to which the request is specifically tailored to discover relevant information;
2. The availability of such information from other sources;
3. The total cost of production, compared to the amount in controversy;
4. The total cost of production, compared to the resources available to each party;
5. The relative ability of each party to control costs and its incentive to do so;
6. The importance of the issues at stake in the litigation; and
7. The relative benefits to the parties of obtaining the information.

Comparing the two lists shows what changed. Cost now appears twice, once against the amount in controversy and once against each party’s resources. The importance of the issues is a new factor. The Rowe factors on specificity and on the likelihood of finding critical information give way to a single question about how specifically the request is tailored to relevant information. The factor on why the responding party keeps the data is gone.⁵⁰ The court called the importance of the issues a critical consideration, even one that will rarely be invoked: a case with the potential for broad public impact, such as a toxic tort class action, an environmental action, social reform litigation, a case involving criminal conduct, or one implicating important legal or constitutional questions, calls for extensive discovery as a matter of public policy.⁵¹ The same opinion also insists that accessible data must be produced even when the party kept it for no business reason, as when data that should have been erased under a retention policy was inadvertently retained.⁵²

The order of the list is deliberate, because the court refused to treat the factors as a checklist. A multifactor test invites a court to count the checks in each column and rule for the longer one, but “we do not just add up the factors.”⁵³ The central question is whether the request imposes an undue burden or expense on the responding party, or, put another way, how important the sought-after evidence is in comparison to the cost of producing it. The factors help answer that question, but a court that applies them mechanically risks losing sight of their purpose.⁵⁴

The first two factors, tailoring and availability from other sources, form what the court called the marginal utility test, and they are the most important.⁵⁵ The test asks, in effect, what the search will add. The cost factors come next, and the importance of the issues follows them. The court noted that its first three groups correspond to the three explicit considerations of Rule 26(b)(2)(iii) as that rule then read.⁵⁶ The relative benefits of production come last. It is fair to presume that a response generally benefits the requester, so this factor rarely distinguishes one case from another, but when production also gives the responding party a tangible or strategic benefit, that fact may weigh against shifting costs.⁵⁷

Courts have shifted most of the expense when a sample shows a poor return on a large bill. In Wiginton v. CB Richard Ellis, Inc., the court considered the Rowe and Zubulake factors but chose factors closer to the then-current wording of Rule 26(b)(2)(iii), adding a new factor of its own. It found that the factors favored cost shifting and put 75 percent of the cost on the requesting party.⁵⁸ The three-tape test had found only 1.64 to 6.5 percent of the e-mail responsive, and low yield of that kind gives the marginal utility factors little to credit.

Quinby v. WestLB AG shows how the cost factors compare expense to the stakes. The employer there said it had spent $226,266.60 restoring and searching the inaccessible e-mail of six former employees. The plaintiff gave no estimate of her damages, but she said her salary in 2000 was $750,000 and would probably have been higher in 2005, the year after she was terminated for the second time, and the court concluded that her salary showed the potential for a multimillion-dollar recovery.⁵⁹ Quinby also added a factor of its own, concerning data a party moved to backup tapes after it was on notice of litigation, and the next subsection takes it up.⁶⁰

The seven factors are guidance, and they are not the rule’s text. The opinions that built and applied them predate the 2015 amendments, and the amended Rule 26(b)(1) now sets out six proportionality considerations: the importance of the issues at stake, the amount in controversy, the parties’ relative access to relevant information, the parties’ resources, the importance of the discovery in resolving the issues, and whether the burden or expense outweighs the likely benefit.⁶¹ The Zubulake cost comparisons and the importance of the issues line up with three of those considerations, and the tailoring, other-source, and relative-benefit factors bear on the importance of the discovery and the balance of burden against benefit. Rule 26(b)(1) adds relative access to information, which the Zubulake list lacks, and it assigns no weights, so the ranking is Zubulake’s own contribution.⁶²

Self-Created Inaccessibility

Rule 26(b)(2)(B) lets a party decline to search sources it identifies as not reasonably accessible because of undue burden or cost, and a court that orders discovery anyway may specify conditions.⁶⁵ Shifting restoration cost is one such condition, but the reason for shifting assumes that the burden comes from the condition of the data. Quinby v. WestLB AG tested that assumption against a responding party that may have created the condition itself. Quinby was an employment discrimination case in which the defendant asked to shift the cost of retrieving data that had been moved to backup tapes after its duty to preserve was triggered. The court held that the cost would not be shifted to the requesting plaintiff if the defendant was on reasonable notice, at the time of the transfer, that the data would potentially be relevant to discovery, and it added a factor to the Zubulake list to capture the point.⁶⁶ The text of Rule 26(b)(2)(B) contains no such limit, so present it as a principle courts draw from the logic of shifting.⁶⁷

The logic is short. Shifting compensates a responder for expense that the data’s inaccessibility imposes. A party that moved data into an inaccessible form when it should have expected to be asked for it chose that expense, and passing the expense to the requester would reward the choice.

Quinby sorted the data custodian by custodian. The employer claimed to have spent $226,266.60 restoring and searching the inaccessible e-mail of six former employees.⁶⁸ The court found that the employer should reasonably have anticipated having to produce the e-mail of every one of the six except Barron, and it applied the Zubulake cost-shifting factors only to the cost of restoring and searching Barron’s e-mail.⁶⁹ The $226,266.60 figure covers all six custodians, but only the share attributable to one of them was open to shifting.

The preservation duty supplies the measure of foreseeability. It arises when a party has notice that evidence is relevant to litigation, or when the party should have known that the evidence may be relevant to future litigation.⁷⁰ The question in a self-created inaccessibility dispute is narrower than whether litigation was coming. It asks whether, when the party moved particular data, it should reasonably have anticipated having to produce that data. The inquiry therefore runs source by source and as of the time of the transfer, as Quinby’s treatment of Barron apart from the other five shows.⁷¹

The measure has limits. Quinby quoted the question whether a corporation, upon recognizing the threat of litigation, must preserve every shred of paper, every e-mail or electronic document, and every backup tape. The answer was “clearly, ‘no,’” because such a rule would cripple large corporations that are almost always involved in litigation.⁷² Victor Stanley, Inc. v. Creative Pipe, Inc. tied preservation to proportionality. The scope of discovery under Rule 26(b) has a proportionality component for ESI, because Rule 26(b)(2)(B) lets a party refuse ESI that is not reasonably accessible without undue burden and expense, and Rule 26(g)(1)(B)(iii) requires a party seeking discovery to certify that the request is not unreasonable or unduly burdensome or expensive. Reasonableness and proportionality should therefore be at the forefront of any inquiry into whether a party met its duty to preserve.⁷³ Guidelines quoted in one opinion point the same way. Absent a showing of special need and relevance, a responding party should not be required to preserve, review, or produce deleted, shadowed, fragmented, or residual data (Principle 9), and unless it is material to resolving the dispute, there is no obligation to preserve and produce metadata absent agreement of the parties or order of the court (Principle 12).⁷⁴

Backup tapes mark the boundary most clearly. Counsel must undertake a reasonable investigation to identify and preserve relevant materials, but that investigation does not automatically include backup tapes. As a general rule, a litigation hold does not apply to inaccessible backup tapes, which may continue to be recycled on the schedule in the company’s policy.⁷⁵ Zubulake IV drew the same line. It treated backup tapes held solely for disaster recovery as inaccessible and tapes accessed in the regular course of business as accessible. It held that parties need not suspend their normal overwriting of disaster-recovery tapes even after the duty to preserve arises, unless they know or should know that the ESI on them is potentially relevant and not duplicated in other accessible sources, and it created an exception for the backup tapes of key players when their data was unavailable from a more accessible source and likely to contain relevant information.⁷⁶ The 2006 amendment to Rule 26(b)(2) then adopted the distinction between accessible and not reasonably accessible sources.⁷⁷ Recycling tapes that only duplicate accessible data removes nothing the requester could have used, while recycling the only copy of a key player’s relevant data is what the exception forbids.

One court found a database not reasonably accessible after the parties agreed that it was no longer in a searchable format and that restoring it would cost at least $124,000, with a monthly storage cost of $27,823. Because of that cost, and because the database would have to be restored from original sources, the court found it not “at present” reasonably accessible. The defendants countered that accessibility should be determined from a particular point in time.⁷⁸ The finding is a snapshot, which is why the moment of measurement, and what the responding party did before it, matter.

The Land O’Lakes dispute shows a routine practice colliding with a hold. Land O’Lakes wiped clean the computer hard drives of former employees who worked on the PROFILE project, and by doing so it eliminated a readily accessible source of potentially relevant information.⁷⁹ Counsel knew that the routine was eliminating an accessible source, and the court found the practice all the more questionable because Land O’Lakes kept no backup tapes of those drives for more than ten days. Once a litigation hold is in place, a party cannot continue a routine procedure that effectively ensures that potentially relevant and readily available information is no longer “reasonably accessible” under Rule 26(b)(2)(B).⁸⁰ A routine adopted for ordinary business reasons becomes a choice once the hold arrives. The decision concerns preservation and allocates no restoration cost, so it supplies the first half of the Quinby reasoning: the duty that makes the loss of accessibility foreseeable.