What Really Matters in eDiscovery Workflow
Conceptually eDiscovery isn’t that hard to grasp. In practice, it’s a different story. In all but the smallest matters, it’s an intricate web of people, processes, and technology. No wonder some lawyers would still rather not deal with it if they can help it. Here’s the thing: law firms can’t help it, not if they want to stay competitive, and for corporations, it’s a fact of life.
The scope of eDiscovery is growing right along with today’s massive growth of data and data sources. Although most lawyers and eDiscovery practitioners use some form of technology, along with analytics, for document review– the most resource-intensive stage of the entire workflow—it doesn’t necessarily lead to an optimized workflow.
What’s the Problem?
The twin tasks in any review process are recall and precision. Recall is retrieving all the data that you need to review into the review platform. Precision is getting only the data you need to review into the platform. High recall and precision yield a defensible, highly relevant, and compact data set for review.
Unfortunately, too many lawyers excel at recall and ignore precision. High recall is easy to do. Just process all the data that could possibly be relevant into the review platform, then throw expensive resources at it until it’s done. You might have high recall, but without high precision your review set is unnecessarily large and very, very costly. And high recall also runs the risk of pulling potentially harmful documents into review, and worse into production.
Now let’s up the ante. Consider this exact same scenario—high recall at low precision–across multiple matters. Here’s what this expensive method is costing us: eDiscovery workflow expenses average 8% for collection, 19% for processing, and a whopping 73% for review. In financial terms, the total eDiscovery market revenue for 2016 was $7.5 billion. Of this, review expenses were a whopping $5.5 billion.
While traditional legal analytics can automate some manual review processes to reduce data sets early on, even more can be automated to achieve workflow nirvana.
Workflow Nirvana (Decision-Driven and Repeatable)
Your or your clients’ objectives are to gain new levels of insight into the data, procedures, and legal costs, to save significant costs on new cases, to apply quality control at every stage of the workflow, and to keep harmful documents out of production.
Best practices will improve your eDiscovery processes and outcomes; however, there’s one best practice which will optimize it. A decision-driven and repeatable eDiscovery workflow will combine the following: human expertise and machines to prioritize data, provide fewer false positives, increase efficiency and improve quality control and reporting. The result will be reduced costs where they affect clients the most—review—while improving speed, transparency and collaboration across all maters—not just a single matter. And because it’s repeatable, it doesn’t benefit just one single matter. It benefits them all. Yes, all matters.
Here’s how a decision-driven, repeatable workflow reduces review data expeditiously to simultaneously achieve high recall and high precision, while driving down eDiscovery costs across single and multiple matters.
Following data processing, specifically de-NISTing and deduplication, expect your workflow solution to apply basic and advanced filters to eliminate spam, junk, files and other obviously non-responsive data. Then, potentially privileged emails and non-responsive emails are flagged, reducing the data set further. Next, global filters and rule-based modifiers are applied.
Now expect the unexpected: predictive and automated analytics. Predictive analytics creates models from past data and applies them to predict future data events. Automated analytics acts based on its own results instead of halting a process to make recommendations. What this means is that expert decisions from prior matters—in fact, up to billions of decisions—can be applied to your matter to automate coding, increase quality control, and flag new data that needs to be reviewed. (This new type of analytics automates repetitive manual processes and enable effective workflows across multiple matters.)
Additional analytics are also applied using a combination of proprietary and third-party tools: email threading, flagging exact dupes, near-dupes, and documents with foreign characters, concept creation based on keywords, and further filters (such as dates, key concepts and concept searches). Optionally, predictive coding can be applied to capture new patterns and prioritize documents by how responsive they are likely to be. Finally, documents are batched by concept and ready for review.
This isn’t one-size-fits-all: as predictive analytics captures new patterns, automated analytics creates policy-driven review sets by similar parameters like privilege and responsiveness. And now that they’re freed from redundant tasks, reviewers spend their time on higher level strategies and creative human analysis.
Finally, at each phase of the process, you are able to view and report on important and detailed metrics across all of your matters—no matter which platform your matters are hosted on—for ongoing cost, matter and data management.
The Numbers Say It All
Let’s see how this quality-based and repeatable approach works in real life. Say, for example, a review set contained 200,000 post-extraction documents. An eDiscovery workflow product analyzed the document set and marked 26% as relevant. Testers calculated that they saved 2,560 review hours for a significant savings of $128,000 over average costs of manual review.
Savings were even greater when the review team applied multi-matter visibility. Using the same test review set as before, the savings nearly tripled. Client-based review retained just 15.6% of the same 200,000 documents, for a total savings of over $300,000 in average review costs.
Smart companies and law firms are using this best practice workflow right now. And once their first case it complete, each subsequent case can be used to continuously improve efficiency. By incorporating traditional legal analytics into a true big data analytics approach that reuses prior work product on new cases, lawyers gain unprecedented insight into their data, unlock substantial cost-savings on new cases, enable a new, deeper level of QC and keep harmful documents out of production. That’s eDiscovery workflow nirvana.