Solving the Blank Page Problem in Ediscovery
A Conversation with Veteran Litigation Technologist Kostantino Athanasopoulos
When Kostantino "Dean" Athanasopoulos walks into a conference room these days, he doesn’t see the stacks of banker’s boxes that once defined legal discovery. As the eLitigation Technology Coordinator for the U.S. Attorney’s Office in Philadelphia—and a recipient of numerous prestigious awards—Athanasopoulos has spent years navigating the shift from paper to digital evidence.
But with that shift came a new challenge.
"Early in my career, we’d line up boxes and review them one by one. It was tedious, but straightforward," Athanasopoulos told me at a recent Everlaw event. "Contrast that with today, where teams are given access to vast amounts of digital information — terabytes of it — and often face an initial sense of utter bewilderment. They don’t know where to start. That’s the blank page problem."
Athanasopoulos recently shared his strategy for overcoming the blank page problem on eDiscovery Today. We recently sat down with him to walk through his approach to overcoming this modern discovery dilemma.
Why Linear Review Doesn’t Cut It Anymore
Athanasopoulos, who has been recognized for implementing innovative legal technology, sees this paralysis play out often. "Manual review of modern datasets isn’t just slow—it’s strategic suicide," he explained. "Deadlines tighten, costs balloon, and critical connections get missed because no one can see the forest for the trees."
Teams are given access to vast amounts of digital information — terabytes of it — and often face an initial sense of utter bewilderment. They don’t know where to start. That’s the blank page problem.
He broke down the three core symptoms:
Lack of clear direction: Reviewers spend significant time sifting through non-essential documents.
Collaboration breakdowns – Isolated reviewers duplicate work or miss patterns.
Narrative gaps – Without structure, turning evidence into arguments takes longer.
"But," he added, "the tools we have now change the game."
How Technology Transforms the Starting Point
Over his career Athanasopoulos has tested nearly every ediscovery innovation. Here’s his playbook for conquering the blank page:
Clustering: The Instant Map
"Clustering is like handing someone a flashlight in a dark warehouse," Athanasopoulos said. "Suddenly, you see document groups light up—a new member of a conspiracy here, emails to an unknown vendor there."
He also described cluster visualization as something that lets you look at your corpus like objects in a small box. “You can quickly shuffle, sort and pull out clusters that aren’t familiar to your case.”
Visual Analytics: Graphing Down the Data
Visual Analytics lets you see your data in graphs and charts to help break up the different categories in your datasets.
Athanasopoulos talked about an example: "The attorney knew the main evidence would be found in emails with invoice attachments and Excel files, so she used data visualizer to filter just those data types instead of looking at the entire corpus.”
His main tip: "Use email domain analysis charts to seek all the emails used and at what frequency. You may find a person of interest to prove your case."
Leveraging Machine Learning: Predictive Coding
"Predictive coding isn't an instantaneous fix, but rather a powerful learning tool in the ediscovery process," Athanasopoulos clarified. "Think of it as a form of guided learning for the system. By initially training your predictive coding model with a few hundred documents that legal experts have tagged for relevance, you empower it to identify crucial information that might otherwise be missed in a manual review."
He elaborated on the process: "It begins with a training phase where legal professionals manually review a subset of documents, categorizing them based on their relevance to specific issues in the case. This allows for the creation of multiple models tailored to different aspects of the litigation. The predictive coding software then analyzes these tagged documents, learning the patterns and characteristics that distinguish relevant from non-relevant information."
Athanasopoulos continued, "Once trained, the system enters the prediction phase, applying its learned understanding to the vast remainder of the document collection, effectively prioritizing and categorizing documents based on their predicted relevance. While not a magic bullet, after an initial period of human input – perhaps a couple of days of diligent tagging – predictive coding can significantly enhance the visibility of pertinent data, allowing teams to focus their review efforts far more effectively."
The process, Athanasopoulos noted, often involves a validation phase where experts review a sample of the system's predictions to ensure accuracy, followed by refinement rounds where the system is further tuned with additional training as needed. This iterative approach ensures the predictive coding model becomes increasingly adept at surfacing the most important evidence.
His pro tip: "After you develop a predictive coding model, save it and apply it to other similar cases. That way you don't have to start from scratch.”
Refining the Search: Leveraging Search Term Reports for Precision
"Many legal teams still approach search terms with a broad, almost indiscriminate method," Athanasopoulos observed. "However, the power of analytics lies in its ability to illuminate which search terms are truly yielding relevant results and, equally importantly, where further investigation might be fruitful."
Stop treating discovery like archaeology—dusting off one artifact at a time. Modern platforms let you construct the case as you go.
He then emphasized the value of Search Term Reports (STRs): "A crucial tool in this refinement process, these detailed reports offer invaluable insights into the efficacy of the search queries we employ within the ediscovery workflow. By meticulously analyzing the frequency with which specific terms appear – the 'hit counts' – and observing trends across different keywords, we can begin to discern the key themes emerging within the vast datasets."
Athanasopoulos explained how STRs directly address the "blank page" problem: "If our initial broad-stroke search terms return an overwhelming number of hits, or conversely, a surprisingly sparse result set, the Search Term Reports provide the data necessary for intelligent refinement of our keyword lists. This iterative process allows us to quickly pinpoint the document subsets that are most likely to contain pertinent evidence, shifting our focus away from the daunting task of aimlessly reviewing massive amounts of data."
By strategically utilizing Search Term Reports, Athanasopoulos underscored, legal teams can move beyond a "blunt instrument" approach to searching and instead employ precise, data-driven techniques that significantly enhance the efficiency and effectiveness of the ediscovery process, leading to fewer dead ends and a faster path to critical evidence.
Building the Case Like a Storyboard
Athanasopoulos’ most passionate advice? "Stop treating discovery like archaeology—dusting off one artifact at a time. Modern platforms let you construct the case as you go."
He described building a timeline where dragging documents onto key dates automatically populated an exhibit. "By deposition time, we had our narrative—and the receipts—ready to go."
The Human Element: Navigating Change Management in the Ediscovery Revolution
While these technological advancements offer powerful solutions to the "blank page" problem, their successful adoption hinges significantly on effective change management. Introducing new workflows and tools can be met with resistance, skepticism, and even overwhelm within legal teams accustomed to more traditional methods. Athanasopoulos, with his experience implementing innovative legal tech within the U.S. Attorney’s Office, understands this human element intimately.
The blank page isn’t about lacking information—it’s about lacking access to it. With the right tools, and the right approach to implementation, you’re not starting from zero.
"The technology is only as effective as the people using it," Athanasopoulos emphasized. "You can have the most sophisticated tools, but if your team isn't trained, doesn't trust it, or doesn't understand how it fits into their existing process, it's just expensive shelfware."
He highlighted several key considerations for navigating this change:
Communication and Transparency: The first step is clearly communicating the "why" behind the adoption of new ediscovery technologies. Legal teams need to understand the limitations of current processes, the benefits these tools offer (e.g., time savings, cost reduction, improved accuracy), and how they will ultimately enhance their ability to uncover critical evidence and build stronger cases. Transparency about the implementation timeline, training opportunities, and support systems is crucial in alleviating anxiety and fostering buy-in.
Gradual Implementation and Pilot Programs: A "big bang" approach to rolling out new technologies can be daunting. Athanasopoulos suggests a more phased approach, starting with pilot programs within smaller teams or on specific case types. This allows for early adopters to champion the tools, identify best practices, and provide valuable feedback that can inform a broader rollout. Success stories from these initial phases can be powerful in convincing more hesitant team members.
Comprehensive Training and Ongoing Support: Adequate training is paramount. This goes beyond basic software tutorials and should focus on how these tools integrate into existing legal workflows and contribute to strategic case building. Ongoing support, including readily available resources and dedicated points of contact for questions and troubleshooting, is essential for building confidence and proficiency. Athanasopoulos noted the importance of "mentorship," echoing his analogy for predictive coding, where experienced users can guide newer team members.
Addressing Concerns and Fostering Trust: Skepticism towards technology, particularly machine learning, is understandable within the legal profession. Addressing concerns openly, demonstrating the reliability and accuracy of the tools through data and case studies, and emphasizing the human oversight involved are crucial for building trust. Highlighting how these technologies augment human capabilities rather than replace them can also ease anxieties.
Empowering Early Adopters and Champions: Identifying and empowering early adopters within the legal team can be a powerful catalyst for change. These individuals can act as internal advocates, sharing their positive experiences and providing peer-to-peer support. Recognizing and celebrating their successes can further encourage broader adoption.
Iterative Feedback and Adaptation: Change management is not a static process. Regularly soliciting feedback from users, identifying areas of friction, and adapting training and workflows accordingly demonstrates a commitment to continuous improvement and ensures the technology truly meets the needs of the legal team.
From Overwhelmed to Strategic
As our conversation wrapped, Athanasopoulos reflected: "The blank page isn’t about lacking information—it’s about lacking access to it. With the right tools, and the right approach to implementation, you’re not starting from zero. You’re starting from insight."
His parting wisdom for legal teams? "Don’t just adopt technology. Adapt your process and empower your people. That’s how you turn data overload into courtroom advantage."
