On The E-Discovery Horizon: Voicemail


E-Discovery Alert

March 11, 2009

As modern technology marches inexorably forward, companies are presented with more and more options for generating, receiving, storing, retrieving and disposing of electronic business communications. Perhaps nowhere is the progression of technology more evident (and the body of case law so sparse) as in the context of voicemail messages. Where voicemail messages were once stored on analog tapes, many companies now utilize digital technology, and some have opted for “unified” technology in which a company’s telephone and computer systems are integrated and voicemail messages are received and capable of playback from a user’s e-mail inbox.

Not surprisingly, these advances in voicemail technology raise a number of e-discovery issues. A company considering implementation of new voicemail technology must evaluate the effect, if any, that implementation will have on its obligations to preserve, search for, and disclose relevant voicemail messages. Given the dearth of case-law directly on point, this proves to be a difficult task. But as technology blurs the distinctions between voicemail and e-mail (an electronic communication for which there is a developed body of case law), the writing on the wall becomes clearer. The purpose of this article is to provide a brief overview of e-discovery issues of which companies should be aware when considering implementation of various digital voicemail arrangements, and to give some insight as to how courts will likely rule on issues concerning the discoverability of digital voicemail messages.

Digital vs. Analog Voicemail

Before dedicated server digital systems, a voicemail message would either reside on the phone company’s server (such as is still the case with cell phone voicemail), or directly on the recipient’s phone in the form of a local recording on a magnetic tape. The phone company had exclusive control over the length of time voicemail was retained, as well as access to those voicemail files. With respect to a magnetic tape, the end user would listen to the recording and record over old messages as new messages were received.

Digital systems represent a significant departure from this simple arrangement. These modern systems store the messages on a central server, from which, if not routinely deleted, they can be retrieved. Some systems amplify the risk by creating additional data points, such as e-mail “message received” notification, or e-mail delivered “pointers” to the messages which allow the messages to be retrieved from the server after clicking on a link in the e-mail. These notification systems sometimes utilize e-mail notices that contain certain data which identify the caller, the caller’s telephone number and the duration of the call (in addition, of course, to the date and time of the call). All of this information provides a litigation adversary with the ability to trace the employees’ receipt of messages, and can thus lead to additional burdensome discovery demands. Depending on the system in place (more on this below), the manner in which such data is retained will likely be under the direct control of the company and its IT professionals, as opposed to the phone company. Further, such digital messages and their accompanying e-mails will likely be subject to a company or client’s backup and retention policies. Unlike traditional voicemail messages, digital messages may prove difficult to delete. Instead, as is common with e-mails, redundant backup systems and personal end user habits may ensure that additional copies continue to persist in many locations.

“Unified” messaging systems – in which the voicemail message appears as an audio file (most commonly a “.wav” file ) in the end user’s email box, accompanied by a transmittal e-mail – pose the greatest risk and potential for increased discovery costs. There are many reasons for this. First, the .wav files create an additional point of easily retrievable data that can exist for extended periods of time in an end user’s e-mail folders. Second, .wav files are typically “large” packets of data – the presence of many such files in the e-mail folders of key employees can exponentially increase the amount of data that must be collected and ultimately reviewed in a litigation setting. Third, the production and review of these audio files is far more complex and time consuming — and therefore expensive — than simply locating and printing out a series of e-mails. The standard methods of reviewing audio to locate information are to listen to the audio file or to have it transcribed and searched as text. Companies in possession of large amounts of electronically stored audio data may find hiring contract attorneys to review and transcribe the audio data to be cost prohibitive. Finally, voicemail recordings, even more so than e-mails, represent evidence of very candid and sometimes damaging communications, complete with voice tone and inflection, most akin to live testimony and more likely than any written material to command the interest of a fact finder.

Discoverability of Unified System Voice Messages

Digital voicemail recordings accompanied by e-mail notification share more in common with e-mails than they do with traditional voicemail, and will likely be subject to identical production obligations during discovery. Although digital voicemail e-discovery disputes have yet to play a prominent role in reported decisions, previous decisions regarding the production of recorded phone conversations are likely to guide the courts’ reasoning when these disputes do arise. For example, in E*Trade Secs. LLC v. Deutsche Bank AG, 230 F.R.D. 582 (D. Minn. 2005), the defendants kept digital recordings of broker/trader conversations on two rewritable DVD’s which were recorded over as new calls were completed. The court found that the defendants had engaged in spoliation of evidence for failing to preserve recorded conversations. Accordingly, the defendants were sanctioned for their conduct. Similarly, in In Re: Seroquel Products Liability Litigation, 244 F.R.D. 650 (M.D. Fl. 2007), the defendant’s failure to produce any voicemails, despite the fact that defendant’s unified system delivered voicemail messages to employees’ Outlook inboxes, was one of several discovery failures that factored into the court’s decision to sanction the defendant. Expert e-discovery commentators, including the Sedona Conference, have indicated that digital voicemail recordings will soon become standard targets of discovery. All point out that Fed. R. Civ. P. 34 has contemplated the production of “sound recordings” even before the Rule amendments in December 2006. See Fed. R. Civ. P. 34(a), 1970 Advisory Committee Notes (revising “the inclusive description of ‘documents’ … to accord with changing technology” and making clear “that Rule 34 applies to electronics data compilations….” ).

Indeed, as judges become more familiar with electronic discovery and electronic storage techniques, they have become increasingly skeptical of conclusory assertions by counsel that all sources of electronically stored information have been searched. Rather, parties are increasingly required to provide detailed descriptions of what was searched, and the parameters under which the search was executed. A party’s obligation to produce electronically stored data, and the expense likely to be incurred in compliance with such obligations, hinges upon whether the data sought is stored in an “accessible” or “inaccessible” format. As a general rule, a party is required to produce data that is deemed accessible at its own expense, but may be able to employ cost shifting if the burdensome and often expensive production of inaccessible data is requested. Zubulake v. UBS Warburg, 217 F.R.D. 309 (S.D.N.Y. 2003), provides an in-depth discussion of the issue. The Zubulake court ranked the following data storage systems in descending order of accessibility: on-line data (active), near-line data, offline storage/archives, backup tapes and finally, erased, fragmented or damaged data. The court noted that the first three types of storage systems constituted accessible sources of electronically stored data, defined as, “data [that] does not need to be restored or otherwise manipulated to be usable.” The remaining two systems constituted inaccessible sources of electronically stored data, which the court defined as “not readily usable” because “[b]ackup tapes must be restored . . . fragmented data must be de-fragmented, and erased data must be reconstructed, all before the data is usable.” Importantly, the Sedona Conference has recently issued a formal comment which proposes a significant expansion to the definition of “inaccessible,” noting that factors relating not only to the type of media (back-up tapes, etc.), but to the complexity of the data and of the processes needed to render the data usable, should be considered.1 Future litigants are likely to invoke this expanded definition when looking to block production of certain stored voicemail recordings, although it is far from clear whether courts will agree with this position. What is more predictable is that e-mail-stored .wav message files – which can be simply copied from a user mailbox to a disc and produced – are not likely to be considered by any court to be “inaccessible,” even under the expanded Sedona definition.

Comparison of Digital Arrangements

A company’s duty to preserve and produce voicemail messages will depend largely on the digital voice mail arrangement in place and the capabilities of that arrangement. We compare below four basic arrangements currently available on the market, and analyze how each arrangement might affect a company’s preservation and production obligations.2

Option 1 – “Basic Digital Voicemail”

A digital arrangement essentially similar to an analog system in that notification of voice messages would only be available to the user via the message light on the telephone and a message advisory when a user calls in to check voice messages remotely. Voice messages which have been received and which have not yet been deleted pursuant to the company’s retention policy will be digitally stored on a voice message server. We assume for purposes of this article that the voice message server is separate and apart from the company’s e-mail servers and will contain the sole copy of each stored message. All voice messages would be stored on a central voicemail server.

Option 2 – “E-mail Notice”

Similar to Option 1, but with the addition of an e-mail alert notifying users of a voice message. As contemplated for this discussion, the e-mail would contain no information other than advising that a voice message has been received. As with the Option 1, voice messages will be exclusively stored on a central voice message server, and users would be required to access voice messages via their office phone or remote call-in process. The added notification benefit to this arrangement would be that remote e-mail users would have immediate notification of a message received without having to be at their workstations to visualize a phone light. If VPN access is provided, a further enhancement could allow a user to play the message through a local pc and speakers utilizing a web application. In this case, the user is still playing the message from the central voicemail server. The user would not, for purposes of this discussion, be able to manipulate, copy or forward the message.

Option 3 – “Link”

A “link system” (sometimes referred to as “Integrated Messaging”) provides a separate folder within the mail client program. Voicemail messages in the user’s integrated mailbox are visible in this folder as “links” to messages on the central voicemail server. While the messages can be reviewed and played by accessing these links and the messages are not physically stored in the mail system, for this discussion we presume an arrangement in which these messages could also be copied to the local client or e-mail system as .wav files which could then be forwarded as e-mail attachments or saved in the end user’s computer. This option is often chosen by companies that are concerned about clogging their e-mail server with voice messages.

Option 4 – “Unified Messaging”

In this arrangement, the full functionality of the unified system is enabled allowing e-mail notification, and delivery of a voice message containing an attached audio file of the message is sent to users. A phone (for example a Blackberry phone) can then be used to tap into the user’s inbox to retrieve the messages, or, of course, the message can be played by most media players by simply clicking on the file (including on most PDAs). Here, the message is stored on the e-mail server along with all other standard e-mail, and would be subject to the company’s e-mail auto deletion and back-up schemes. As with standard e-mail, users could also control retention or deletion of the messages by, for example, “double deleting” the files which would permanently remove them from the server and the local computer, or, conversely, storing the messages in e-mail subfolders that are not subject to the auto deletion scheme or copying the files to c-drive locations on the end user’s hard drive. Also, absent restrictions placed on the system, a user may forward the e-mail with copy of the audio file to others, thus allowing for copies of the voice message to reside in multiple locations.

Comparison and Discussion of the Various Arrangements

A company looking to implement one of the above arrangements (or some hybrid thereof) needs to determine the effect, if any, that implementation of the above arrangements will have on its obligation to preserve and produce voicemail messages.

As Sedona explains:

The duty to preserve relevant information arises when litigation is “reasonably anticipated.” The duty to preserve relevant information is certainly triggered when a complaint is served or a governmental proceeding is initiated or a subpoena is received. However, the duty to preserve could well arise before a complaint is served or a subpoena is received and regardless of whether the organization is bringing the action, is the target of the action or is a third party possessing relevant evidence. The touchstone is “reasonable anticipation.” “Reasonable anticipation” of litigation arises when an organization is on notice of a credible threat it will become involved in litigation or anticipates taking action to initiate litigation.3

Once triggered, the duty to preserve applies to all relevant existing or subsequently created data. It would not apply, of course, to data properly discarded in the normal course prior to the triggering of the hold. This would include all voicemails discarded in the normal course on an existing retention schedule.4 Thus, once triggered, the company’s obligation to preserve voicemail messages and/or e-mails reflecting same would likely encompass messages and related e-mails already received and retained pursuant to the company’s voice message retention policy, and messages and e-mails received on an going forward basis.

Option 1 – “Digital Voicemail”

Of the four arrangements, Option 1 would impose the least onerous voicemail preservation obligation on a company. Although preservation of voice messages on a simple digital voicemail system is possible, identification of relevant voice messages which have already been received and are within the retention period would be difficult. In theory, upon issuance of a litigation hold notice or on receipt of a discovery request, a company’s IT department would have to suspend its retention policy and halt destruction of potentially all voice messages on the voicemail server. Then, in order to identify the relevant messages for preservation, each and every message would have to be reviewed by the Legal Department for a relevance determination on a message by message basis. This could be prohibitively time consuming, expensive and generally burdensome enough to allow for a strong argument that the information is not “reasonably accessible” and thus need not be preserved, restored and produced.

Notably, however, since the voice messages are digitally stored and might be identifiable through metadata fields, such as sender/recipient, date received, etc., voice message preservation may be considered feasible in a case where preservation efforts can reasonably be limited to voicemails received by certain individuals or groups within the company. The case law, however, indicates that the obligation to undertake this effort might only be triggered in the unusual circumstance that the producing party is aware (or reasonably should have been aware) of the existence of the message and its high degree of relevance to the litigation (e.g., the central issue in the case turns on the phone message left). This would be akin to the court’s holding in Columbia Pictures v. Bunnell, in which the data, stored in RAM, was extremely ephemeral, yet worthy of preservation given its central relevance to the issues at bar. See Columbia Pictures v. Bunnell, 2007 U.S. Dist. LEXIS 46364 (C.D. Cal. 2007).

Preservation of voice messages received on an ongoing basis after implementation of the litigation hold can be accomplished by instructing any employees subject to the litigation hold notice to notify the Legal Department on receipt of any potentially relevant voice message so that the message can be reviewed and, if necessary, preserved. It is likely that courts would impose such an ongoing obligation on a company notwithstanding the nature of this type of system, because once the hold notice is disseminated, the custodians and the company are aware of the need to sequester the post-hold created information for possible production.

By far the most favorable attribute of the Option 1 arrangement is the lack of any e-mail or other automatic written notification of a voice message. Since voice messages will be stored solely on a central voice message server and not on an e-mail server or any individual’s computer, individual employees will have little or no involvement with identification and preservation of voice messages already received and stored and deletion schedules can be much more carefully controlled and carried out by the company’s IT department.5 This also simplifies and centralizes the litigation hold process. By limiting or eliminating any written trail of a voice message, the inadvertent destruction of the message will be much more difficult to detect by adversaries.

Option 2 – “E-mail Notice”

A company’s preservation obligations under an E-mail Notice arrangement would will be similar to its obligations under the Option 1 arrangement. However, an E-mail Notice set up will create an additional data point within the end user’s e-mail system that, if not regularly and permanently deleted, may point an adversary to the fact that a voicemail message was received by the end user, on a certain date and at a certain time. A court could hold that if the particular date, time and custodian are highly relevant to the litigation issues, a company would have an obligation to preserve the voicemail message. However, this scenario becomes unlikely if the company maintains a strict 15 day deletion schedule, as most relevant voicemails would likely have been created prior to 15 days before the hold trigger. Again, because the actual voicemail messages here would continue to reside solely on the voicemail server, IT could maintain strict control over deletion schedules. Ideally, in this system, the e-mail advisory would contain no information which might be used to identify a voice message by content or sender. This would further minimize the risk of discovery and a triggering of the obligation to preserve.

Option 3 – “Link”

Identifying and preserving voice messages using a Link, or Integrated Messaging, arrangement would be similar to the Option 1 and 2 in that voice messages reside in one location, the voicemail server. The degree of risk involved with implementation of this option will depend greatly on the information contained in the advisory e-mail containing the message link and whether the web interface option is permitted which would allow copying and manipulation of the voice message. If the e-mail identifies the sender of the voice message, an adversary reviewing the e-mail may be able to argue the relevance of the voice message based on the context of the metadata regarding the sender, recipient and date/time received. Additionally, even though the actual voice message file resides on a central server, not as an attachment to an e-mail, if the recipient forwards the e-mail containing the sender information, or accesses the message through the web interface and copies it, the risk of an adversary discovering the existence of the voice message increases. Also, depending on the security implemented on the message links, forwarding an e-mail link may allow users to share the content of a voice message with ease. As with any document or record, the more people that know about the existence of a voice message, the greater the likelihood of the discovery of the message by an adversary. If this type of arrangement is implemented by a company, users need to be educated about the risks of disseminating voice message notification e-mails to unintended or unnecessary recipients and copying messages (if the web interface option is permitted).

Option 4 – “Unified Messaging”

As noted above, the fully unified messaging system poses the greatest risk relating to identification and preservation of voice messages. In this type of system, voice messages, generally in the form of .wav files, are forwarded directly to the end user through the company’s e-mail system. Like standard e-mail, the voice message will reside on the e-mail server but can be duplicated, preserved and forwarded with ease by the end user. Although a fully unified system is the most attractive in terms of user convenience, several risks and cost concerns are present here:

  • Because the messages will enter the end users inbox, and frequently remain in folders or hard drive locations, the volume of searchable data and the likelihood of discovery will increase significantly.
  • The company will lose much of its ability to centrally control the retention of the data; like e-mails, messages will be kept on individual hard drive locations by custodians, indefinitely in some cases.
  • The company will lose much of its ability to control the dissemination of the messages. Unified voicemails are as easy to send to e-mail recipients as forwarding an e-mail. Moreover, many custodians will (or will at least have the ability to) store these messages at off-site locations. These factors increase the overall scope and cost of the company’s preservation obligation.
  • The messages will reside on the company’s e-mail server(s), thereby making their retention possibly subject to periods longer than the typically shorter retention for voicemail messages. In fact, courts will likely treat voice messages similar to e-mails for identification and preservation purposes. See, e.g., In Re: Seroquel Products Liability Litigation, 244 F.R.D. 650, 661 (M.D. Fl. 2007) (defendant’s failure to produce any voicemails, despite the fact that defendant’s unified system delivered same to Outlook inboxes, was one of several discovery failures argued by plaintiff and listed in the court’s decision granting sanctions against defendant).
  • Unified messages typically have identifying information in their e-mail “vehicles” (particularly if they are sent from one system user to another). For instance, the transmitting e-mail may identify the telephone number or internal extension from where the call originated. This additional data would make it far more difficult for a company to argue that it is unduly burdensome and costly to search all relevant custodians e-mail for the existence of messages containing relevant phone numbers or names. Most courts would likely analyze such requests no differently than now standard requests for e-mail searches using search term protocols.
  • A unified system with full functionality enabled may significantly increase the cost of e-discovery processing at every stage because it exponentially increases the total volume of data to be processed, such data requiring and requires the costly procedure of full message review (listening, which takes more time than reading), or implementation of new and expensive voicemail transcription technologies.


In sum, much like any other recent technological advance, a digital voicemail implementation plan should be carefully reviewed with an eye towards potential business and legal consequences. Although modern voicemail technology has increased the convenience with which users may receive, store, search and play back voicemail recordings, companies need to be aware of the risks and costs. In light of Zubulake and many similar decisions, and the Sedona Principles and commentary, it is important to prepare for the possibility that litigants will routinely demand, and courts will compel, that all relevant and available digital voicemail messages be produced. Unless carefully constructed and strictly enforced use and retention protocols are enacted along with the new system, digital voicemail systems will likely lead to the creation of large volumes of discoverable electronically stored information. 

1 The Sedona Conference® Commentary on Preservation, Management and Identification of Sources of Information that are Not Reasonably Accessible (July 2008).
2 Although the four digital arrangements are by no means the only ones available on the market, they present basic options in convenience and system capability, and are thus instructive.
3 The Sedona Commentary On Legal Holds: The Trigger And The Process (August 2007); see also, e.g., Treppel v. Biovail Corp., 233 F.R.D. 363 (S.D.N.Y 2006) Cache la Poudre Fees v. Land O’Lakes,, 244 F.R.D. 614, 621 (D. Colo. 2007) (letter referencing potential “exposure” but not threatening litigation did not trigger obligation to preserve).
4 For companies seeking to reduce the discoverable data pool, unless there is a business or legal need to the contrary, voice messages ideally should not be regularly retained for any longer than 30 days, and preferably shorter, e.g. 15 days. Of course, the shorter the retention period, the shorter the amount of time a company whose duty to preserve has been triggered will have to take steps to ensure that potentially relevant messages are preserved before they are auto-deleted.
5 Of course, to the extent that an individual employee has saved or is aware of a relevant voice message received within the retention period, that employee should be instructed in the litigation hold notice to contact the Legal Department for possible preservation. (See below.)