When you think about eDiscovery, the first thing that comes to mind is probably document review. Review is the phase of the EDRM in eDiscovery where you actually discover potentially relevant documents for a matter. The goal of document review is to find relevant data that uncovers information and evidence about the facts of the matter.
In this guide, we’ll cover the main challenges and best practices of document review, as well as a step-by-step review process that will help you get to the signal faster in your next matter.
Document review is the act of sorting through all data related to a case to identify responsive documents to produce and privileged documents to withhold from opposing counsel.
During the document review process, the attorney in charge usually tags documents according to different criteria such as their relevance to the matter, their responsiveness to a discovery request, or their privilege. However, there’s no one-size-fits-all process for the document review. Review is an iterative learning process, meaning you will need to repeat and refine your work over and over again.
Due to the exploding volumes of evidence available for eDiscovery, your team should limit the scope of what needs to be reviewed whenever possible using targeted collection, filtering, and culling techniques prior to a manual review.
These data culling processes together with the review itself are generally performed with the help of eDiscovery software, which includes features such as automatic deduplication or smart filters to reduce the number of documents that require eyes-on review, as well as tools to make the doc review process easier, such as automatic tags, redactions, and advanced searching tools.
The main purpose of document review is to uncover factual issues in a case and develop legal strategies to pursue based on information that is found in a collection of documents.
Once legal document review is completed, you need to produce any discoverable information to opposing parties, which includes any non-privileged information that’s relevant to any claims of the matter and that it’s proportional to the needs of the case.
Although document review is most prevalent in litigation, it can get particularly intense in mergers and acquisitions, subpoena and third-party requests, and internal investigations, where large volumes of data — mostly ESI — are generated.
In today’s digital world, billions of emails, text messages, presentations, database files, Slack messages, and the like are sent and received every day, creating eDiscovery dilemmas for lawyers and judges. Any one of them could be the pivotal document that either builds a case or destroys one.
If there is one case that demonstrates the awesome power of eDiscovery, it is United States v. Microsoft, the infamous antitrust trial that began in the late 1990s. A key contention was that Microsoft was conspiring against Sun Microsystems. Thanks to the review process, an internal email from Bill Gates surfaced. In it, he asked: “Do we have a clear plan on what we want Apple to do to undermine Sun?” It was a modern-day smoking gun.
A more recent case demonstrates the danger inherent in the review process. The Samsung v. Apple litigation pitted two tech giants against each other in an epic court battle between the world’s dominant smartphone makers. But the case also showed that even the biggest companies can stumble in the face of eDiscovery review.
During the document review, Samsung’s outside counsel improperly redacted a sensitive Apple contract it had obtained during the discovery process, a contract that was under a protective order, limiting its use to the litigation alone. The associate then uploaded the contract to Samsung’s internal intranet, where it “went viral.” When Apple found out, sanctions soon followed.
Another example comes from Oracle v. Google, which pitted two more tech companies against each other in a high-stakes copyright dispute. In that case, Google’s famous search abilities didn’t seem to transfer to its document review process. During its privilege review, the company screened out an email that was labeled “Attorney Work Product.” But Google’s review team did not catch nine drafts of the same email, which were then produced to Oracle. Worse yet, that email became a key piece of evidence against the company.
And, apart from the risks of accidentally producing the wrong document to opposing parties, there are the costs. It’s estimated that over 70% of all the costs involved in eDiscovery are generated by the review process alone. And the dollar value behind that percentage doesn’t stop growing as data volumes explode.
Not surprisingly, a recent in-housing survey conducted by Logikcull found that review costs are the most important factor to consider when evaluating eDiscovery expenses. They are also the main source of friction with outside counsel, as 41% of respondents reported review costs to be the number one source of pushback with OC.
As you can see, document review is not a matter to be taken lightly. A well-executed doc review process can win a matter for your client—while saving them thousands of dollars. But even a well-intentioned review team can make mistakes that can sink a case or expose litigants to a negative inference or even incur sanctions.
The document review process is the most complex and expensive part of eDiscovery. Because of this complexity, review is usually handled by a team of reviewers—usually a combination of attorneys and paralegals–who generally rely on modern eDiscovery technology to make their job easier.
In order to begin, many teams create a document review guide to define a systematic process and help reviewers make informed decisions. The guide should explain the facts in the case, the search terms being used, and the codes to be used to tag documents as responsive, non-responsive, privileged, and other determinations.
Your review guide should consider:
As discussed in our Introduction to eDiscovery, Federal Rule of Civil Procedure 26(f) requires parties to meet and confer to discuss the nature of the claims, compare views, consider settling options, set the preserving requirements for discoverable information, and to develop a consolidated discovery plan addressing discovery of electronically stored information and the form or forms in which it should be produced. Many states have similar requirements—and whether required or not, the meet and confer process can be essential to properly scope and plan discovery.
According to Rule 26(f), eDiscovery topics covered must include preservation, the form of production, and privilege and work-product protection claims. Other important issues to be discussed include the scope of the review, the issues in dispute, the search protocols, and the mechanics of the production of materials between parties after review. Parties should also facilitate communication channels between team leaders in order to resolve issues quickly.
The ultimate goal of these meet and confer rules is to save the parties—and the courts— time and money.
One of the most important things to consider before starting your document review is whether your review set has been sufficiently narrowed down so that your review is as low-lift and low-cost as possible. There are different techniques to achieve that goal, but two sure-shot methods are to make a sound data collection — preferably by prioritizing custodians, rather than collecting all at once—and culling your data through deduplication, filtering, search terms, and other techniques before moving on to an eyes-on review.
Once the scope of your review is clear, determine what form your information will come in. Will the data set in question be all ESI or a mixture of ESI and paper? If you have paper documents, converting them to an electronic format by scanning them can greatly simplify the document review process, keeping the entire review in one collection.
Consider how the ESI will be searched. Parties often redact ESI in the way they once redacted paper documents, by blacking out text. To make that possible, ESI is converted to an image file, such as TIFF, in a process that makes it no longer searchable. Files you can’t search aren’t much help, so after redaction, searchability needs to be restored by using “optical character recognition.” OCR technology can extract the remaining text from the TIFF image file, restoring searchability to the non-redacted sections.
This TIFF-OCR method is most appropriate for documents that are largely text-based. It is much less fitting for complex, dynamic documents, such as spreadsheets or databases. Consider the types of files likely to be reviewed before agreeing to a conversion method.
Similarly, parties should address Bates numbering early in the process. The Bates system is still the standard tool for tracking reviewed documents, but native files are not easily Bates numbered. However, Bates numbers can serve as filenames for native files. If you expect to handle a large number of native files, negotiate the Bates numbering conventions to be used during your meet-and-confer efforts.
It is possible to begin privilege review before human eyes even see the first document. Modern document review software can search and tag documents that have privileged names, email addresses, law firm domains, or legal terms before reviewers examine those documents. This way, reviewers are aware that a document is potentially privileged and can double-check it to confirm it is in fact a privileged communication or attorney work product.
To-do list before a document review:
Perhaps the most important consideration in the review process is the search terms to be employed. As mentioned earlier, this process will be iterative. You will continually refine your search as the review process begins. After negotiating an initial list of search terms, you will begin reviewing the documents returned using these terms.
If too many irrelevant documents are being returned you are experiencing “noise,” also commonly referred to as “false positive” search results. If that is the case, search criteria may be refined, optimized, and tweaked to get the desired results. You will have to narrow search terms if too many irrelevant documents are found and broaden the search if too few relevant documents are found. This may also involve testing different search technologies.
Use Misspellings: This may be counter-intuitive, but even in the age of spell check, people get words wrong. Include common misspellings in your list of search terms.
Look for code words and jargon: Consider terms that only the players in a matter might know. Use language real people use in their emails, not lawyerly terminology.
Learn Boolean logic: Boolean searches use connectors such as “AND”, “OR”, and “NOT” to provide more refined searches. Proximity searches can identify when specific words are used in the same sentence or paragraph.
Play Your Wild Cards: Symbols can replace one or more characters (i.e., e-disc*very)
Stemming: Reduces a word to its root form, as in “eating,” “eat,” and “eater.”
Stop Words Stop Pointless Searches: Stop words tell the search engine to ignore common words that will not help find important information, like the, and, or with.
Quality control and due diligence should be performed at every stage of the review process to ensure consistent and accurate document designation. One important consideration is to allow the review team to tag documents for “further review” when unsure of coding. The review platform should also be able to apply quality control restrictions such as automatically identifying duplicate documents, or grouping families of documents such as threaded emails and attachments for easy identification.
In order to QC your work, include a second level review of all designated production documents (or a sample of them) by senior attorneys and randomly review the team’s coding to check for inconsistencies. Provide feedback to the reviewers as they work in order to improve quality and spot issues as they arise.
As we briefly highlighted above in the Oracle v. Google example, a failure to protect privileged documents can ruin a case. When privileged documents are produced to opposing counsel, privilege could be waived. In some cases, it is possible to demand the documents be returned and stricken from the record, but, as attorneys often say, you can’t unring a bell. That is, opposing counsel knows what the document says and will certainly use that knowledge to the best of their ability.
The work product doctrine protects records such as documents that were prepared by the client, the attorney, or an agent for either the client or the attorney. Attorney-client privilege pertains to communications by and between the attorney (or his/her agent) and a client that were made for the purpose of giving or receiving legal advice.
In order to review for privilege, search the review data set using a list of search terms that will likely identify privileged documents. Some possible search terms include:
Be sure to provide all reviewers with a list of people and topics that may involve privilege. Parties should have a discovery agreement in place that allows inadvertently produced privileged documents to be returned. A Rule 502(d) order may also be sought, in order to protect against waiving privilege through inadvertent disclosure. Otherwise, the court may deem the privilege to be waived
Remember that what you see on screen as an email or a document is actually a partial representation of all of the data and information in an electronic document. Meeting the obligation to produce data means choosing suitable forms of production which deliver the content without destroying the hidden and associated information in a file, such as its metadata.
Requesting parties often demand native production of reviewed documents, but native production is complicated and often involves a form of production that closely approximates the contents and usability of the source. When negotiating the production of reviewed materials to outside parties, determine if native production is the best approach or if there is a more appropriate form of production.
Important questions to consider include:
The document review process is one of the most critical phases of an eDiscovery project. In order to properly manage this process, make sure to:
With the growing volumes of evidence and the expense of manual review, many legal teams are considering computer assisted review to help weed through digital records faster and with less human intervention. Predictive coding is the common name for this process which is emerging as an accepted practice in some extremely large cases.