Reducing data volumes in eDiscovery is crucial for controlling costs, improving efficiency, and streamlining the review process. Here's a comprehensive overview of strategies and techniques for achieving this:

I. The EDRM Model and Data Reduction

The Electronic Discovery Reference Model (EDRM) provides a framework for eDiscovery. Data reduction strategies are relevant across multiple stages:

• Information Governance: Proper information governance policies are the most proactive way to reduce data volumes. Implement policies for:


• Data Retention Schedules: Define how long different types of data need to be kept based on legal and business requirements. Regularly dispose of data that is no longer needed.


• Data Mapping: Understand where critical information resides within the organization.


• Data Minimization: Reduce the amount of data created and stored.


• Employee Training: Educate employees on data management best practices.


• Identification: Quickly locate and identify relevant data sources.


• Preservation: Ensure that potentially relevant data is not altered or destroyed.


• Collection: Collect only the data that is reasonably likely to be relevant and proportional to the needs of the case.


• Processing: This is a major phase for data reduction. Techniques employed include deduplication, deNISTing, email threading, filtering by date ranges, and keyword searches.


• Review: Reduced data volumes directly translate to faster and more cost-effective review.


• Analysis: The reduced and processed data is easier to analyze.


• Production: Preparing the final data for production, including redacting and formatting.

II. Key Data Reduction Techniques

Here's a detailed breakdown of common data reduction techniques:

A. Early Case Assessment (ECA)

• Purpose: Evaluate the merits of a case early on to determine if it's worth pursuing or to inform settlement negotiations. ECA helps define the scope of discovery and identify key custodians and data sources.


• Techniques:


• Keyword Searching: Quickly identify and assess documents containing relevant keywords.


• Conceptual Search: Find documents similar in concept or theme to known relevant documents.


• Custodian Interviews: Gather information from key custodians about data locations and document types.


• Data Sampling: Examine a representative sample of the data to understand its nature and relevance.


• Visualization Tools: Use charts and graphs to identify patterns and trends in the data.

B. Data Deduplication

• Purpose: Remove duplicate files across the entire data set.


• Mechanism: Hashing algorithms (e.g., MD5, SHA-1) are used to create unique identifiers for each file. Files with identical hashes are considered duplicates and only one copy is retained.


• Types:


• Global Deduplication: Removes duplicates across all custodians and data sources.


• Custodian-Level Deduplication: Removes duplicates within each custodian's data.


• Benefits: Significant reduction in data volume, especially in environments with shared files.


• Considerations: Ensure the deduplication process is defensible. Document the methodology used.

C. DeNISTing (Operating System File Filtering)

• Purpose: Exclude known system files, program files, and other non-relevant files from the data set.


• Mechanism: Compares file hashes against the National Institute of Standards and Technology (NIST) National Software Reference Library (NSRL) database. Files identified as known system files are excluded.


• Benefits: Removes a large volume of irrelevant data, especially operating system files.


• Considerations: Verify the accuracy and currency of the NIST database. Consider whether to exclude all NIST files, as some might be relevant in certain cases (e.g., malware analysis).

D. Email Threading and Near-Duplicate Analysis

• Purpose: Organize and group email messages into conversations (threads) and identify near-duplicate documents.


• Email Threading:


• Mechanism: Analyzes email headers (To, From, Subject, Date, etc.) and content to identify email chains. The most inclusive email in a thread (the "inclusive email") contains all prior emails in the conversation.


• Benefits: Reduces the number of emails requiring review by focusing on the inclusive email. Allows reviewers to see the entire conversation history.


• Considerations: Complex email threading algorithms are required for accuracy.


• Near-Duplicate Analysis:


• Mechanism: Identifies documents that are substantially similar, even if not exact duplicates. Algorithms analyze text and metadata to determine similarity.


• Benefits: Reduces review effort by identifying and prioritizing the most relevant version of a near-duplicate document.


• Considerations: Set appropriate similarity thresholds to balance accuracy and efficiency.

E. Date Range Filtering

• Purpose: Limit the data set to a specific date range relevant to the case.


• Mechanism: Filters documents based on their creation date, modification date, or other date fields.


• Benefits: Eliminates data outside the relevant timeframe.


• Considerations: Carefully define the date range based on the facts of the case.

F. Keyword Searching and Filtering

• Purpose: Identify and extract documents containing specific keywords or terms relevant to the case.


• Mechanism: Uses search algorithms to locate documents containing the specified keywords.


• Benefits: Narrows the focus to the most relevant documents.


• Considerations: Develop a comprehensive keyword list in consultation with legal counsel and subject matter experts. Consider using Boolean operators (AND, OR, NOT) and proximity searches to refine search results.

G. File Type Filtering

• Purpose: Exclude certain file types that are unlikely to be relevant (e.g., system files, program files, image files, audio files).


• Mechanism: Filters documents based on their file extension (e.g., .exe, .dll, .jpg, .mp3).


• Benefits: Removes non-essential data.


• Considerations: Be cautious about excluding file types, as they may contain relevant information in some cases (e.g., embedded images in documents).

H. Custodian Filtering

• Purpose: Focus collection and processing on data from specific custodians identified as key players in the case.


• Mechanism: Collect data only from designated custodians.


• Benefits: Reduces the volume of data to be processed and reviewed.


• Considerations: Carefully select custodians based on the facts of the case and legal requirements.

I. Managed Early Data Assessment (MEDA)

• Purpose: Using technology assisted review tools earlier in the eDiscovery process to assess the relevance of documents before collection and processing.


• Mechanism: This is done by using TAR tools to sample, search, and analyze the data, helping legal teams make informed decisions about which data to collect and process.


• Benefits: Significant cost savings by reducing the amount of data that needs to be collected, processed, and reviewed.


• Considerations: Requires expertise in TAR and a well-defined workflow to ensure accuracy and defensibility.

J. Technology Assisted Review (TAR) / Predictive Coding

• Purpose: Use machine learning algorithms to prioritize and classify documents based on their relevance to the case.


• Mechanism: A subject matter expert (SME) or attorney manually reviews a sample of documents and identifies those that are relevant ("seed set"). The TAR system learns from this training set and then predicts the relevance of the remaining documents.


• Benefits: Significantly reduces the number of documents requiring manual review. Improves the accuracy and consistency of review.


• Considerations: Requires a well-defined workflow, a representative training set, and ongoing monitoring of the system's performance. Requires expertise in TAR.

III. Strategic Considerations and Best Practices

• Proportionality: Ensure that the scope of discovery is proportional to the needs of the case.


• Defensibility: Document all data reduction techniques used and ensure that the process is defensible in court.


• Transparency: Communicate with opposing counsel about the data reduction techniques being used.


• Collaboration: Involve legal counsel, IT professionals, and eDiscovery experts in the data reduction process.


• Testing and Validation: Thoroughly test and validate data reduction techniques to ensure accuracy and completeness.


• Software Selection: Choose eDiscovery software with robust data reduction capabilities. Ensure the software is compatible with the organization's IT infrastructure and data sources.


• Ongoing Monitoring: Monitor the effectiveness of data reduction techniques and make adjustments as needed.


• Legal Hold: Implement a legal hold policy to prevent the destruction of potentially relevant data. Suspend data retention schedules for custodians involved in litigation.


• Data Security: Implement appropriate security measures to protect data during collection, processing, and review.


• Iterative Approach: Data reduction is not a one-time process. It is an iterative process that should be revisited throughout the eDiscovery lifecycle. As you learn more about the case, you may need to refine your data reduction strategies.

IV. The Impact of Data Volume on Costs

Reducing data volumes directly translates into cost savings in several areas:

• Collection: Less data to collect means lower collection costs.


• Processing: Reduced processing costs due to less data to process.


• Storage: Lower storage costs due to smaller data sets.


• Review: The largest cost component of eDiscovery. Less data to review significantly reduces review costs.


• Production: Lower production costs due to less data to produce.

V. Examples of Cost Savings

• Deduplication: Deduplication can often reduce data volumes by 20-50%, resulting in significant cost savings in processing and review.


• DeNISTing: DeNISTing can remove up to 30% of the data, further reducing processing and storage costs.


• Technology Assisted Review (TAR): TAR can significantly reduce the number of documents requiring manual review, often by 50-90%, resulting in substantial cost savings.

By implementing these strategies, organizations can effectively reduce data volumes in eDiscovery, control costs, improve efficiency, and streamline the review process. Remember to always prioritize defensibility and proportionality in all data reduction efforts.