Anyone involved in eDiscovery has been on either side of productions and they can present significant challenges in agreements on format, execution, and quality control.
Production falls after the process, review and analyze portion of the Electronic Discovery Reference Model (EDRM). They are at the end of the eDiscovery process, where documents are lower in volume and higher in relevance. This means that a series of human or human-directed computer decisions have been made in order to narrow a document set.
The most common type of production contains load files, images, natives and text files. The .dat file contains all the metadata from the original files, the .opt contains the paths to the image files and the .lst contains the paths to the text. The images can be in many formats, such as tiffs, jpgs, or pdfs. They can be single or multipage files and they can be in color or black & white. Native productions are also an option but should be approached cautiously since documents in their original state are subject to alteration and may be difficult to use in legal proceedings. Native file productions are generally used for file categories that do not image well, cannot be imaged at all like video/audio files, or documents that have different image options such as track changes, notes or hidden data.
The ideal type of production for use with modern review tools are the following:
• Load file with all available derived metadata
• Multipage PDFs
• Native files (for certain file categories such as Excel, PowerPoint, or other specific file types)
Productions should contain placeholders for select file categories that indicate further processing needed or errors such as empty files, encrypted documents, compressed archives, etc. Text should be extracted, OCR, or redacted OCR text depending on the document.
The worst types of productions may be one or more of the following:
• A single or several PDFs that contains multiple documents in one file (They are difficult to separate into documents.)
• No load files, only Bates-stamped PDFs or images (they are missing metadata and parent-child relationships)
• No native files at all (There was no consideration for poor imaging types such as Excel files with formulas, track changes, corrupt documents or exceptions)
• Documents without any text (this can be suspicious for many file categories)
There are many other issues that are often seen with productions but the above are difficult to fix even by skilled litigation support professionals.
There are a number of practical steps that legal and litigation support professionals can take to achieve a useful and cost-efficient production. Consistency is the most important element of a production; those that come from a variety of sources are bound to introduce inconsistencies and unnecessary collisions which lead to manual time needed to normalize. In short, decide at the beginning who will create the productions and stick with that group, set production formats and time frames at the very beginning, and use production templates. Either have a vendor create the productions or your litigation support team, but not both. Having templates ready for mass exports on the fly will help with last-minute attorney requests for depositions and more.
The importance of agreeing on production formats and time frames is critical. Following the agreed-upon format will help smooth the eDiscovery process and remember to keep a copy for yourself. Be sure to ask for metadata that allows for de-duplication like conversation index and a description of the hash methodology. Also ensure you have updated custodian lists for de-duplicated data for each production. Most productions have family intact (parent-child relationships) so if yours does not, make sure everyone knows ahead of time.
Establish a production protocol and agreement from the outset. Include clearly defined goals in the protocol such as “we want to collect (these kinds of) devices” and “(this kind) of data.” Allow for necessary changes in technology such as a new collection method – this is particularly important in the fast-paced world of social media. Ensure that you don't restrict parties from being able to view, share, or transfer data to other parties in order to comply with HIPAA, protective orders, and more. The last step that should not be overlooked is asking someone with eDiscovery production experience to review the production.
In these protocols we must at a minimum: have parent child relationships for documents, know the custodian(s) of documents, get all derived metadata from the native files, understand the deduplication methods used, and receive any current/updated deduplicated metadata such as paths and custodians.
Be careful of protocols that have undescribed deduplication or near-duplication methods, only produce parts of email threads (often called threading), or just don't seem normal for the source data type. For instance, producing pictures from a smartphone in PDFs with no additional metadata is not the way those devices typically save them.
Quality control of outbound productions can be broken in two categories: before production or during the creation of a production. Before production, check images for issues such as documents that don't image well, and check redactions carefully, especially if they are in areas that would normally be produced in a .dat file. For example, email header fields.
During the creation of productions, be careful in breaking document families apart. Don't remove documents before sending to opposing as this creates gaps in numbering and may require re-numbering. Again, be very careful with redactions as information may still reside in the .dat and .txt files that was intended to be removed. Finally, consider the time to create the production. Most modern eDiscovery tools will have automated quality control and exception handling for productions, but endorsing many pages in order will still take computer time.
There are four key factors to consider for inbound productions: how is it helpful, language, how much should be reviewed, and quality control. The first is the most global – did you get what you asked for, did you ask for the right things, how does this information inform on the case, and which documents are key. There may be specific terminology and fields for production and it may include foreign languages, which is important for searching, and analytics. Deduplication and artificial intelligence can be used in figuring out how much should actually be reviewed.
Quality control of inbound productions should be done before loading to a review tool. The same processes can be used as outbound quality control, being careful of text, metadata, and missing natives. Matching fields and data is important, including deduplication data, parent-child relationships, and custodian information (including deduplicated data paths). Identifying where there is missing text as well as exceptions and redactions can forestall issues in review and search. In essence, it is wise to perform the same quality control on inbound production you would on your own outbound productions.
Recent eDiscovery tools may have options to enhance inbound image and native productions. For images, modern tools convert these to searchable documents; when a user searches for a term it is highlighted in the Bates-stamped image loaded to the review tool. These modern tools may fill in missing metadata for produced native documents. For example, if an Excel is produced with minimal metadata in the .dat file, a modern tool will reprocess that excel and may discover hidden rows, hidden text, embedded documents and formulas. This enhancement example is a good reason to request natives AND images for all documents.
There are several broad recommendations regarding productions that can help save time and effort. Enhancing inbound productions as much as possible is the first. The second is ensuring they adhere to agreed-upon standards and changes as soon as possible if necessary; asking for overlays or corrected productions as the need arises. Having a plan to handle redactions and clawback agreements is very important as those occur in many cases. Finally, expect a production as good as what you produce; if you adhere to the production standard, that is a solid foundation upon which to build if other parties do not.