It is no secret or surprise to anyone in the legal profession that document review takes up the bulk of the time, effort, and costs surrounding litigation. As the cost associated with litigation continues to rise across the United States, the need for document review to become more efficient has been at the forefront of many legal minds for well over a decade.
The case for tool-assisted solutions is obvious, and one that hardly needs any more explaining at this stage: in 2012, the RAND Corporation published a research brief that concluded that document review by itself absorbed about 73% of the cost associated with litigation, while also signaling that bringing labor cost down wasn’t a realistic option: “The rates paid to [project attorneys] may well have bottomed out, with further reductions of any significant size unlikely. Some companies have turned to LPOs with access to lawyers in other countries (…) but this practice may raise concerns about information security, oversight, maintaining attorney-client privilege, and logistics” (page 2).
In addition to labor costs being an unlikely source of cost reduction, RAND also notes that human reviewer speed has simply reached its apex. Even lawyers with relevant expertise and knowledge top out at roughly 100 documents per hour, a pace they reached decades ago. In 2018, LawGeex, an AI-assisted contract review tool, pitted 20 corporate lawyers against their solution. The results were telling, with the AI reaching a higher accuracy level than the lawyers (94% to an 85% average). More importantly was the time save, however: the lawyers did the job, on average, in 92 minutes (times ranging from 51 to 156 minutes). The AI did the same job at a better accuracy level in just 26 seconds.
With the above in mind, the continued rise of eDiscovery tools and technology-assisted review solutions is hardly surprising. Review is typically done by some combination of manual and tool-assisted labor. A 2011 study accurately sums up the balance that must be struck between human and artificial intelligence: “[Document review] requires both the ability to perform routine repetitive tasks in an accurate and timely manner as well as the ability to apply human judgment, reasoning and making fine distinctions about complex matters. (…) The recommended approach to achieve greater accuracy and efficiency is to allocate tasks between humans and computers that play to their respective strengths rather than to their respective weaknesses” (Page 14).
At its core, document review is a culling operation, preparing for production the evidence that a party involved in litigation must disclose under the Federal Rules of Civil Procedure. The initial search for documents typically yields a large amount of potentially responsive (or relevant) documents, the task for document review is to go through these potential sources of evidence and separate what should and should not be disclosed.
During the review process the original batch of search results is reviewed and split into two:
This part is simple enough: irrelevant documents are withheld, while relevant documents move on to the next stage, which is to determine if privilege or confidentiality rules apply.
If a responsive document turns out to contain privileged or confidential information (and that privilege or confidentiality has not been waived or otherwise addressed), that document need not (and should not) be disclosed. If privilege or confidentiality-related rules do not apply, the document must be produced. The next step is to determine whether or not the privileged or confidential information can be redacted or not.
If the privileged or confidential information can be redacted, then the information that needs to be withheld should be, while the rest of the document is left intact. Following this operation, the document must be marked as ‘redacted’ and disclosed. In cases where redaction isn’t feasible, then the document should be marked as ‘privileged’ and be withheld. Simple enough, but proper redaction is a key element to ensuring privilege remains intact, confidentiality protected and potential damage to the cause avoided.
In essence, that is what document review is: a culling operation where legal professionals and AI-based tools go through a massive amount of information to see what is and isn’t relevant, and which parts of the relevant bits must be disclosed.
In many ways, the challenges of document review all boil down to a single factor: getting it wrong. In a vacuum, reviewing a vast set of documents and making the appropriate decisions with regards to relevance and privilege is challenging already. Litigation isn’t a vacuum however, it’s a pressure cooker. The window between the start of litigation and initial disclosures has narrowed significantly (from 120 to 90 days), coupled with a continuing growth in data generated (or documents to review) means that document review is in a constant state of having to do more work in less time.
As much as it may seem the obvious conclusion, the holy grail for document review isn’t to make it cheaper. Rather, efficiency is. In light of this, rather than looking at how to reduce the costs of legal review, let’s look at how you can create a more efficient legal review process.
There truly is no way to properly emphasize how important it is for legal teams to have playbooks and protocols ready to go. Once the ball gets rolling, there is simply too much to do in not enough time to fuss about planning ahead properly. However, when it comes to document review, where the details matter greatly and the process is too complicated to reliably generate the desired results without a solid playbook informing your decisions.
A playbook should contain a standardized schedule that starts whenever litigation can reasonably be expected. Reasonable expectation of litigation is the point when legal teams must start preserving evidence through legal holds according to FRCP 37(e). The playbook should also provide an overview of the responsibilities of every team member involved, as well as a standardized list of tags to use when coding documents following review.
As mentioned above, the time allowance on initial disclosure and discovery requests is very limited and deadlines very tight. At the same time, those deadlines are pretty much set in stone, which means they are predictable. Typically, even if the individuals vary (wildly) on a case-by-case basis, the steps involved don’t: this is why playbooks are a useable tool in the first place. Using templates and automation to play into the predictable nature of the process can be a big time saver and help ensure consistency.
For the review process itself, having a consistent process of collecting data helps speed collection. It also reduces the amount of confusion that may occur around the collection process, since everybody involved in it will know exactly what goes out since communications are standardized.
By reviewing the review team, we’re not talking about second- or third-level reviews (yet): review teams should be built for productivity. The task of performing document review requires a skill set that mixes project management, legal expertise, subject matter expertise, and technological skill.
In addition to those measurable skills, there are also intangibles such as the reviewer being able to deal well with stress, being a good communicator, and being detail-oriented. All of these are important for a member of the review team to have.
When the chips are down, the people in a review team are the ones who operate the technology and execute the process detailed in the playbooks and plans. Making sure those people are the right ones is vital.
The document review as discussed here is considered a first-level document review, with the second and third level taking place after initial disclosures to get a sense of what can be used for or against in court (second level) and to ensure that the evidence produced can be found again later if needed (third level).
In that first level, the balance between work done by in-house or external counsel varies between organizations (and even from case-to-case at times). From a cost reduction point of view, the more work that can be done in-house, the better. At the same time, it is paramount that the quality of the work does not suffer: if the in-house team is unable to handle a more detailed and fine-toothed culling operation, then it makes no sense to ask them to do so.
Equipping the in-house review team with the right people, proper processes and adequate tools should help them to take a growing role in the process. By periodically evaluating the performance of the team and the disparities between the dataset sent to external counsel and the final production, you can readjust the balance between the in-house and external counsel responsibilities when it comes to document review.
Tools are unavoidable when it comes to document review nowadays, and it’s been that way for decades now. As we noted earlier, the amount of data is growing at a terrifying pace, human review speed has reached its apex, and the time allowed to perform review is being reduced rather than expanded. Invariably, tools are needed to make up the difference.
Getting a tool that provides an end-to-end solution for document review can make a difference when it comes to achieving the desired result in a case. Modern-day solutions offer a wide range of applications of Artificial Intelligence and dynamic reporting. Automated classification and tagging speeds up parts of the review process tremendously, enabling legal teams to be more effective and efficient with the limited time they are given.
The final point here is to emphasize that no matter how perfect a solution may be, it can only ever be as good as the people that use it and the process it is part of. Assuming the first two parts are in place, the search for the right solution should very much consider the user-friendliness of a tool. Having a tool is one thing, having a solution another: for document review, only a tool that the review team can easily use to achieve results can actually be called a solution.
In addition to user-friendliness, user support is invaluable. Although every eDiscovery vendor builds their solution knowing the stakes for its users, even the strongest links in the chain will break down at times. Having the proper support available from a vendor to address such issues is crucial. When deciding on which eDiscovery solution vendors to consider, make use of resources such as Gartner Peer Insights or G2.
By virtue of being such a large time sink for most legal departments, document review is nearly always at or near the top of the list of potential avenues for improvement of legal department performance. In most cases, the solution is a combination of procedural improvements, tool selection, and reviewer skill.
Striking the right balance between those three factors is no easy feat: no two litigation matters are the same, and the rules within which to operate, and the technology available, are subject to change. Nevertheless, having the proper tools in the hands of the right people in the context of a robust process will lead to better legal outcomes than would otherwise be the case.