Tools and Techniques for Defensibly Processing Electronic Data

Section One of this article provides an overview of data processing. It outlines the basic goals of processing, identifies potential difficulties that may occur with processing, offers various considerations organizations should keep in mind when selecting appropriate vendors, and offers suggestions on how to approach processing in a quality-controlled and legally defensible manner. Section Two examines some tools and technologies available for processing data.

Section One: Data Collection

After data collection has occurred, the files that have been collected and culled need to be processed and sorted. The ultimate goal of processing is to create a data set that can be reviewed by attorneys utilizing the selected reviewing platforms. In order to achieve this goal, the data must be put into consistent file formats and organized into fully searchable data sets that will enable attorneys to review and define the documents that must be produced. Some of the primary goals of processing are to: 1) identify exactly what data is contained; 2) record metadata as it existed prior to processing; and 3) reduce the amount of data in a defensible way by selecting only the appropriate data that should go on to be reviewed. Processing Guide, Electronic Discovery Reference Model available at www.edrm.net/resources/guides/edrm-framework-guides/processing. Maintaining proper documentation and transparency is critical at the processing stage. Jason R. Baron & Macyl A. Burke, The Sedona Conference Commentary on Achieving Quality in the E-Discovery Process, 10 Sedona Conf. J. 299, 306-7 (Fall 2009).

When data first arrives, it can come in many formats, and often requires further processing before it can be reviewed. Many files are now collected and reviewed in their native file formats; this is especially true of structured data (such as number-intensive databases). Not all collected data will be capable of review without further processing, however. For example, some data may be collected from backup tapes, and may then have to be restored before further work can be done. Other files and e-mails may need to be extracted, such as zip files, while yet other files such as legacy file formats, may first need to be converted.

Processing Basics

Processing can be subdivided into four parts: Assessment, Preparation, Selection, and Output. Processing Guide, Electronic Discovery Reference Model. Each of these sub-processes allows for opportunities to reduce the amount of items that will move forward to the reviewing stage. Quality control methods should be employed at every step, to verify that all the processes were performed as expected, and that all of the data was properly identified, itemized, and transformed.

During Assessment, parties should be determining which types of data should be moved on to processing. In making this determination, parties should align the processing phase with the goals and strategies of the overall electronic discovery, and develop and implement a methodology that will yield the desired results in terms of the data output, time, and costs.

During Preparation, the data chosen in the Assessment phase may need to be restored, converted, extracted, catalogued, itemized, indexed and checked for de-duplication and similarity.

In Selection, parties have the opportunity to reduce the amount of data that should move forward to the review stage, and to determine which data should be left behind. De-duplication can be employed to avoid redundancies in the data to be reviewed, and other search terms and parameters can be applied to further filter the data, ensuring that only potentially relevant data is classified for further review. Common filtering techniques used at the Selection phase include keyword searches and date filters.

It is important to recognize that simply running a filtering tool and trusting the results is typically not sufficient. Parties should evaluate the outcomes of these filters and searches and apply appropriate metrics, such as numbering the amount of included or excluded documents, for an indication of whether the filtering parameters were too narrow or too broad, and to help identify potential errors in how the filtering rules were applied. Ideally, the parties should agree on filtering and searching methods they wish to employ, including search terms and concepts, and should in their agreement allow for flexibility, should an issue develop. The Sedona Principles: Best Practices Recommendations & Principles for Addressing Electronic Document Production (2d ed.), Comment 11.a, (June 2007), available at https://thesedonaconference.org/publications (last viewed Nov. 16, 2012). If not, the parties may find themselves in a position where the court must order additional searches, which can become costly for the parties involved.

In the Output phase of processing, the data that has been selected for review is then transformed into the proper formats, corresponding with the needs of the review stage, so that it can be reviewed using the review platforms selected.

Potential Difficulties with Processing

There can be some major challenges at the processing stage of discovery. One significant challenge during processing is maintaining the proper relationships between e-mail messages and their attachments, and accurately accounting for which people saw what, and when they saw it. Donald Stever, Sedona Conference Database Principles Addressing Preservation & Production of Databases & Database Information, ST051 ALI-ABA 101, Sedona Principle 12 (March 2011). Another challenge may occur if much of the data collected is made up of archived files, which can be difficult, time-intensive, and costly to decompress. For organizations doing business in countries outside the United States, there might be data collected in languages other than English, and correctly processing and capturing this information can be complicated. As new types of media continually emerge, there can also be challenges in creating reviewable data sets from the data collected off of new types of media such as iPads. It can also be difficult to properly estimate the cost of properly processing files, especially early on in the discovery process.

Another challenge can occur with processing files to be produced in the manner in which it is ordinarily maintained (the “native format”). While production of files in native format is a popular option, a production in the native format of a database may often render the information less useable than an alternative production format. Donald Stever, at Principle 12. A true production of a native format database may only be accessible to someone who has a licensed copy of the correct software.

Vendor Considerations for Processing

Applications for processing electronic data can be purchased by companies or law firms, although the high initial expense of the hardware, application, licensing fees, as well as the recurring usage fees, makes it an impractical approach for many. Jerry Thompson, The Evolution of Litigation Support, 45 Spg Ark. Law, 20, 22 (Spring 2010). For this reason, many companies and law firms prefer to use an outside vendor for data processing.

It is becoming increasingly important for parties to be able to quickly analyze documents soon after they have been collected, in order to develop a basic understanding of the documents and how those documents relate to the issues in the matter. John H. Jessen, An Overview of ESI Storage & Retrieval, 11 Sedona Conf. J. 237 (Fall 2010). There are now early case assessment technologies that allow for the opportunity to analyze native file collections before they are processed. These tools can provide valuable insight into the content of the data collections, and can help legal teams develop their strategies for appropriately and accurately reducing and filtering the volume of data. 45 Spg Ark. Law, 21-22. Parties with short timelines, in particular, should consider employing early case assessment technologies, and discuss them with potential vendors. There are also other considerations that come into play when choosing vendors to assist with processing.

First, organizations seeking the proper vendor should consider the state of their own information management processes. If the organization has a significant document management process in place, and it is facing relatively narrow discovery obligations, there may not be a need for a vendor. Second, organizations should consider their time constraints. If a large volume of electronic data must be processed in a short time frame, then a vendor with sufficient capacity and resources to accomplish this task must be selected. Third, organizations should consider the quality control procedures that the vendor utilizes, as described below. Fourth, organizations should consider the pricing mechanisms used by the vendor. Different pricing structures may be more or less advantageous for a party, depending on its discovery needs. Fifth, organizations should consider the data security methods employed by the vendor, particularly in instances where the data being processed is subject to privacy regulations such as HIPAA.

Quality Control and Defensibility

Processing errors can be costly for litigants, and once they occur, they can be very difficult to correct. When processing data, parties should be careful to strictly adhere to “process auditing; quality control; analysis and validation, and chain of custody considerations.” EDRM Processing Guide. Quality control is very important, not only because of the risk of sanctions, but also because failure to employ a quality electronic discovery process can have a variety of undesirable consequences. In addition to risking court sanctions, failure to implement a quality process could result in the failure to disclose key evidence, or it could result in the inadvertent production of privileged or confidential information. 10 Sedona Conf. J., 309.

Employing a process that measures quality, offers transparency, and documents any actions taken can contribute greatly to defensibility by providing metrics and allowing for the possibility of corrections. Without proper documentation of the processing efforts, it may be difficult for a producing party to defensibly represent that the production is complete. 10 Sedona Conf. J., 313. Maintaining simple metrics, such as how and what type of data has been collected and processed can be very helpful in detecting potential processing issues. 10 Sedona Conf. J., 316. Parties should keep a record of any files that were not produced because of difficulty processing the files, as with corrupted or encrypted data, for example. Another useful approach can be sampling from the files deemed unresponsive, testing to confirm results, and inspecting to make sure there are no discrepancies or inconsistencies. 10 Sedona Conf. J., 303.

Section Two: Tools and Technologies

It has been estimated that 80% of the time and cost of electronic discovery can be spent in the stages of processing, review, and analysis, which means having the right workflow tools to manage the processing can contribute significantly to improving efficiency and reducing costs. Thomas Allman, Ashish Prasad, Anthony Diana & Matthew Rooney, Electronic Discovery Deskbook, 13:9, Practicing Law Institute, 2009. This section examines some of the tools and technologies associated with data processing. Discovery processing applications seem diverse in their offerings, but most are very similar in functionality and capabilities. They may provide different names to their processes and/or focus on one specific aspect of their capabilities, but in the end, the applications essentially provide a legally defensible process that can assess, prepare, select, and output data for use in legal review. Our discussion of specific tools and technologies should not be taken as an endorsement thereof. The following section summarizes the Gartner Group's independent analysis of the leaders, challengers, niche players, and visionaries in all categories of electronic discovery. www.gartner.com/technology/research/methodologies/magicQuadrants.jsp.

One of the most widely used discovery processing vendors, according to the Gartner Group, is LexisNexis' LegalAccessWare (LAW) PreDiscovery application. LAW exceeds the requirements, laid out in Section 1 above, for processing applications, by providing the ability to quickly ingest multiple file formats, properly identify those formats regardless of visible extension, extract most archive containers, search/filter/mark data for removal early in the process, and to produce in almost every conceivable production format. Another player in the processing arena is Nuix. This originally started as an application that could quickly index and ingest large amounts of native data. It has now grown to be a significant player in all aspects of the processing arena, having added extensive filtering and production capabilities to its offerings.

As processing applications have evolved, early case assessment tools have arisen to provide earlier and easier insight into data. One of the early application providers of this service was Symantec's Clearwell. Clearwell began as an e-mail indexing software but then grew into a tool that was used to quickly index data; it was an early tool to provide “transparent search,” the ability to use the data itself to provide insight into the proper keyword and concept searches to use. This ability allowed law firms and corporations to get information inside the “black box” ' insight into processing by service providers that wasn't available previously until the final delivery of data. Finally, companies such as Recommind, kCura, and Catalyst, which traditionally started in the web review and analysis arena, are introducing new processing engines into their applications to allow clients to process and host the data in a single step. Some of these providers have already existing dynamic search capabilities that offer similar functionality to that of the early case assessment technologies that are being offered by providers in that area.

Conclusion

There are many different aspects to the data processing phase of discovery that parties should keep in mind. Parties should understand the basic goals and principles of processing, and should understand the different stages involved in processing. While processing can provide a good opportunity to reduce the volume of data that goes onto the review phase, parties must be aware of the potential complications and intricacies involved in filtering data, so as to avoid potentially costly errors.

Parties should carefully analyze their needs and the different services and pricing structures offered by different vendors, in order to make the best determination of which vendor to use. Above all else, parties should take appropriate measures to maintain quality control and defensibility at the processing stage, and should carefully document all of their efforts. Various tools and technologies can be helpful to parties in the processing stage. Careful consideration, planning, and execution at the processing phase can help parties to better control costs, and to avoid liabilities and pitfalls associated with electronic discovery.

Jason Fliegel is Director and Senior Counsel, eDiscovery and Records Management at Abbott Laboratories, Ashish Prasad is the Founder and CEO of Discovery Services LLC, and Todd Haley is the Vice President, eDiscovery, of eTera Consulting. The views expressed in this article are those of the authors and do not necessarily reflect the views of their respective employers.

Pursuant to the Federal Rules of Civil Procedure, if the parties to litigation fail to take reasonable steps to assure that relevant, non-privileged data is produced to opposing parties in response to discovery requests, courts may impose sanctions and may instruct juries to draw an adverse inference. Consequently, parties to litigation must take special care to adhere to defensible standards and practices when processing electronic data, especially since the overwhelming majority of information generated by companies is produced and stored electronically. Because the matter of what constitutes these defensible standards and practices is the subject of so much debate, and the risks of error weigh so heavily in the balance, the need for corporate counsel develop proper procedures is a matter of critical and growing need.

Data Processing

Data processing is a key stage in the efforts of parties to comply with their discovery obligations with respect to electronically stored information (ESI). Once ESI is collected, that data must be sorted and some method of identification should be applied in order to segregate the potentially relevant data from the other data. The potentially relevant data that is defined by counsel as appropriate for review by attorneys for relevance, privilege and other factors must then be put into a reviewable format so that such review can take place. If the data is not sorted and identified properly after collection, then there can be no confidence that the correct data will be defined for review by attorneys, and that parties' compliance with discovery obligations will be achieved.

Section One of this article provides an overview of data processing. It outlines the basic goals of processing, identifies potential difficulties that may occur with processing, offers various considerations organizations should keep in mind when selecting appropriate vendors, and offers suggestions on how to approach processing in a quality-controlled and legally defensible manner. Section Two examines some tools and technologies available for processing data.

Section One: Data Collection

After data collection has occurred, the files that have been collected and culled need to be processed and sorted. The ultimate goal of processing is to create a data set that can be reviewed by attorneys utilizing the selected reviewing platforms. In order to achieve this goal, the data must be put into consistent file formats and organized into fully searchable data sets that will enable attorneys to review and define the documents that must be produced. Some of the primary goals of processing are to: 1) identify exactly what data is contained; 2) record metadata as it existed prior to processing; and 3) reduce the amount of data in a defensible way by selecting only the appropriate data that should go on to be reviewed. Processing Guide, Electronic Discovery Reference Model available at www.edrm.net/resources/guides/edrm-framework-guides/processing. Maintaining proper documentation and transparency is critical at the processing stage. Jason R. Baron & Macyl A. Burke, The Sedona Conference Commentary on Achieving Quality in the E-Discovery Process, 10 Sedona Conf. J. 299, 306-7 (Fall 2009).

When data first arrives, it can come in many formats, and often requires further processing before it can be reviewed. Many files are now collected and reviewed in their native file formats; this is especially true of structured data (such as number-intensive databases). Not all collected data will be capable of review without further processing, however. For example, some data may be collected from backup tapes, and may then have to be restored before further work can be done. Other files and e-mails may need to be extracted, such as zip files, while yet other files such as legacy file formats, may first need to be converted.

Processing Basics

Processing can be subdivided into four parts: Assessment, Preparation, Selection, and Output. Processing Guide, Electronic Discovery Reference Model. Each of these sub-processes allows for opportunities to reduce the amount of items that will move forward to the reviewing stage. Quality control methods should be employed at every step, to verify that all the processes were performed as expected, and that all of the data was properly identified, itemized, and transformed.

During Assessment, parties should be determining which types of data should be moved on to processing. In making this determination, parties should align the processing phase with the goals and strategies of the overall electronic discovery, and develop and implement a methodology that will yield the desired results in terms of the data output, time, and costs.

During Preparation, the data chosen in the Assessment phase may need to be restored, converted, extracted, catalogued, itemized, indexed and checked for de-duplication and similarity.

In Selection, parties have the opportunity to reduce the amount of data that should move forward to the review stage, and to determine which data should be left behind. De-duplication can be employed to avoid redundancies in the data to be reviewed, and other search terms and parameters can be applied to further filter the data, ensuring that only potentially relevant data is classified for further review. Common filtering techniques used at the Selection phase include keyword searches and date filters.

It is important to recognize that simply running a filtering tool and trusting the results is typically not sufficient. Parties should evaluate the outcomes of these filters and searches and apply appropriate metrics, such as numbering the amount of included or excluded documents, for an indication of whether the filtering parameters were too narrow or too broad, and to help identify potential errors in how the filtering rules were applied. Ideally, the parties should agree on filtering and searching methods they wish to employ, including search terms and concepts, and should in their agreement allow for flexibility, should an issue develop. The Sedona Principles: Best Practices Recommendations & Principles for Addressing Electronic Document Production (2d ed.), Comment 11.a, (June 2007), available at https://thesedonaconference.org/publications (last viewed Nov. 16, 2012). If not, the parties may find themselves in a position where the court must order additional searches, which can become costly for the parties involved.

In the Output phase of processing, the data that has been selected for review is then transformed into the proper formats, corresponding with the needs of the review stage, so that it can be reviewed using the review platforms selected.

Potential Difficulties with Processing

There can be some major challenges at the processing stage of discovery. One significant challenge during processing is maintaining the proper relationships between e-mail messages and their attachments, and accurately accounting for which people saw what, and when they saw it. Donald Stever, Sedona Conference Database Principles Addressing Preservation & Production of Databases & Database Information, ST051 ALI-ABA 101, Sedona Principle 12 (March 2011). Another challenge may occur if much of the data collected is made up of archived files, which can be difficult, time-intensive, and costly to decompress. For organizations doing business in countries outside the United States, there might be data collected in languages other than English, and correctly processing and capturing this information can be complicated. As new types of media continually emerge, there can also be challenges in creating reviewable data sets from the data collected off of new types of media such as iPads. It can also be difficult to properly estimate the cost of properly processing files, especially early on in the discovery process.

Another challenge can occur with processing files to be produced in the manner in which it is ordinarily maintained (the “native format”). While production of files in native format is a popular option, a production in the native format of a database may often render the information less useable than an alternative production format. Donald Stever, at Principle 12. A true production of a native format database may only be accessible to someone who has a licensed copy of the correct software.

Vendor Considerations for Processing

Applications for processing electronic data can be purchased by companies or law firms, although the high initial expense of the hardware, application, licensing fees, as well as the recurring usage fees, makes it an impractical approach for many. Jerry Thompson, The Evolution of Litigation Support, 45 Spg Ark. Law, 20, 22 (Spring 2010). For this reason, many companies and law firms prefer to use an outside vendor for data processing.

It is becoming increasingly important for parties to be able to quickly analyze documents soon after they have been collected, in order to develop a basic understanding of the documents and how those documents relate to the issues in the matter. John H. Jessen, An Overview of ESI Storage & Retrieval, 11 Sedona Conf. J. 237 (Fall 2010). There are now early case assessment technologies that allow for the opportunity to analyze native file collections before they are processed. These tools can provide valuable insight into the content of the data collections, and can help legal teams develop their strategies for appropriately and accurately reducing and filtering the volume of data. 45 Spg Ark. Law, 21-22. Parties with short timelines, in particular, should consider employing early case assessment technologies, and discuss them with potential vendors. There are also other considerations that come into play when choosing vendors to assist with processing.

First, organizations seeking the proper vendor should consider the state of their own information management processes. If the organization has a significant document management process in place, and it is facing relatively narrow discovery obligations, there may not be a need for a vendor. Second, organizations should consider their time constraints. If a large volume of electronic data must be processed in a short time frame, then a vendor with sufficient capacity and resources to accomplish this task must be selected. Third, organizations should consider the quality control procedures that the vendor utilizes, as described below. Fourth, organizations should consider the pricing mechanisms used by the vendor. Different pricing structures may be more or less advantageous for a party, depending on its discovery needs. Fifth, organizations should consider the data security methods employed by the vendor, particularly in instances where the data being processed is subject to privacy regulations such as HIPAA.

Quality Control and Defensibility

Processing errors can be costly for litigants, and once they occur, they can be very difficult to correct. When processing data, parties should be careful to strictly adhere to “process auditing; quality control; analysis and validation, and chain of custody considerations.” EDRM Processing Guide. Quality control is very important, not only because of the risk of sanctions, but also because failure to employ a quality electronic discovery process can have a variety of undesirable consequences. In addition to risking court sanctions, failure to implement a quality process could result in the failure to disclose key evidence, or it could result in the inadvertent production of privileged or confidential information. 10 Sedona Conf. J., 309.

Employing a process that measures quality, offers transparency, and documents any actions taken can contribute greatly to defensibility by providing metrics and allowing for the possibility of corrections. Without proper documentation of the processing efforts, it may be difficult for a producing party to defensibly represent that the production is complete. 10 Sedona Conf. J., 313. Maintaining simple metrics, such as how and what type of data has been collected and processed can be very helpful in detecting potential processing issues. 10 Sedona Conf. J., 316. Parties should keep a record of any files that were not produced because of difficulty processing the files, as with corrupted or encrypted data, for example. Another useful approach can be sampling from the files deemed unresponsive, testing to confirm results, and inspecting to make sure there are no discrepancies or inconsistencies. 10 Sedona Conf. J., 303.

Section Two: Tools and Technologies

It has been estimated that 80% of the time and cost of electronic discovery can be spent in the stages of processing, review, and analysis, which means having the right workflow tools to manage the processing can contribute significantly to improving efficiency and reducing costs. Thomas Allman, Ashish Prasad, Anthony Diana & Matthew Rooney, Electronic Discovery Deskbook, 13:9, Practicing Law Institute, 2009. This section examines some of the tools and technologies associated with data processing. Discovery processing applications seem diverse in their offerings, but most are very similar in functionality and capabilities. They may provide different names to their processes and/or focus on one specific aspect of their capabilities, but in the end, the applications essentially provide a legally defensible process that can assess, prepare, select, and output data for use in legal review. Our discussion of specific tools and technologies should not be taken as an endorsement thereof. The following section summarizes the Gartner Group's independent analysis of the leaders, challengers, niche players, and visionaries in all categories of electronic discovery. www.gartner.com/technology/research/methodologies/magicQuadrants.jsp.

One of the most widely used discovery processing vendors, according to the Gartner Group, is LexisNexis' LegalAccessWare (LAW) PreDiscovery application. LAW exceeds the requirements, laid out in Section 1 above, for processing applications, by providing the ability to quickly ingest multiple file formats, properly identify those formats regardless of visible extension, extract most archive containers, search/filter/mark data for removal early in the process, and to produce in almost every conceivable production format. Another player in the processing arena is Nuix. This originally started as an application that could quickly index and ingest large amounts of native data. It has now grown to be a significant player in all aspects of the processing arena, having added extensive filtering and production capabilities to its offerings.

As processing applications have evolved, early case assessment tools have arisen to provide earlier and easier insight into data. One of the early application providers of this service was Symantec's Clearwell. Clearwell began as an e-mail indexing software but then grew into a tool that was used to quickly index data; it was an early tool to provide “transparent search,” the ability to use the data itself to provide insight into the proper keyword and concept searches to use. This ability allowed law firms and corporations to get information inside the “black box” ' insight into processing by service providers that wasn't available previously until the final delivery of data. Finally, companies such as Recommind, kCura, and Catalyst, which traditionally started in the web review and analysis arena, are introducing new processing engines into their applications to allow clients to process and host the data in a single step. Some of these providers have already existing dynamic search capabilities that offer similar functionality to that of the early case assessment technologies that are being offered by providers in that area.

Conclusion

There are many different aspects to the data processing phase of discovery that parties should keep in mind. Parties should understand the basic goals and principles of processing, and should understand the different stages involved in processing. While processing can provide a good opportunity to reduce the volume of data that goes onto the review phase, parties must be aware of the potential complications and intricacies involved in filtering data, so as to avoid potentially costly errors.

Parties should carefully analyze their needs and the different services and pricing structures offered by different vendors, in order to make the best determination of which vendor to use. Above all else, parties should take appropriate measures to maintain quality control and defensibility at the processing stage, and should carefully document all of their efforts. Various tools and technologies can be helpful to parties in the processing stage. Careful consideration, planning, and execution at the processing phase can help parties to better control costs, and to avoid liabilities and pitfalls associated with electronic discovery.

Jason Fliegel is Director and Senior Counsel, eDiscovery and Records Management at Abbott Laboratories, Ashish Prasad is the Founder and CEO of Discovery Services LLC, and Todd Haley is the Vice President, eDiscovery, of eTera Consulting. The views expressed in this article are those of the authors and do not necessarily reflect the views of their respective employers.