Taming Modern Data Challenges: Defining a “Conversation”

In our last post, we discussed how the rise of enterprise collaboration platforms is reshaping the eDiscovery landscape, the core challenges presented by these platforms, and the strategic approaches legal teams must take to adapt.

Historically, the concept of Electronically Stored Information (ESI) has long been synonymous with documents – emails, Word files, spreadsheets, and presentations. These traditional forms of ESI are static, discrete, and easily packaged into files that reflect a familiar structure for legal review.

But in today’s communication-driven workplace, that paradigm is rapidly shifting. The rise of messaging platforms like Slack, Microsoft Teams, WhatsApp, and SMS (for text) has ushered in a new era of conversational ESI – and with it, new challenges and opportunities for legal teams. In this post, we will discuss considerations associated with defining a “conversation”, how there is a current lack of standards for doing so, and how to address conversational-based ESI in discovery.

The Rise of Conversational Data

As long as there has been a discipline called “eDiscovery”, there has been conversational data. Historically, that has been most common in emails that reflect conversations between multiple parties. Outlook and other email platforms standardized email messaging to include the original message text by default when replying to or forwarding a message, resulting in a straightforward approach to capture email conversations. As a result, each subsequent email contained the prior conversation up to that point, so it became the usual practice to employ email threading to identify the email(s) within a thread that represents the entire conversation. Easy-peasy!

However, unlike email, modern chat and messaging tools result in ESI that is fluid, fragmented, and contextual. These platforms are designed for real-time, informal collaboration. Individuals communicate in bursts of short messages, either one to one or across multiple channels or group threads, creating a continuous stream of dialogue rather than discrete, standalone communications.

This conversational data reflects how business gets done today – decisions are often made “on the fly”, questions are asked and answered in rapid succession, and actions are coordinated across distributed teams. From a discovery standpoint, this data can be highly relevant, but it often defies the traditional assumptions and workflows built around document-centric discovery.

Key Differences Between Traditional ESI and Conversational ESI

Here are several ways conversational ESI stands apart from its more conventional counterparts:

• Structure vs. Flow: Traditional ESI (like Word docs or emails) is file-based, with clear start and end points. Chat data, by contrast, flows continuously, with no obvious “document” boundary.
• Temporal Proximity: Chat messages are often timestamped down to the second and can involve multiple participants in near real-time. This creates dense chronological data that’s essential for reconstructing timelines, but difficult to segment logically.
• Contextual Dependencies: A single message might make little sense in isolation. Its meaning may depend on prior messages, emojis, reactions, shared links, or even conversation threads occurring simultaneously in different channels.
• Metadata Complexity: Conversational platforms generate rich metadata – message IDs, user IDs, channel names, timestamps, reactions, edits, deletions – all of which are critical for understanding context and ensuring authenticity.
• Volume and Redundancy: High-frequency, low-substance messages (e.g., “OK”, “Got it”, or even a “thumbs-up” emoji) create noise in review. Conversely, meaningful decisions may occur in fleeting exchanges that might be easily overlooked if review is forced into a document paradigm, based on arbitrary time-based “chunks” like 24 hours per “document”.

All these factors must be considered in discovery of conversational-based ESI.

Multi-Channel Conversations

Complicating the discovery of conversational ESI even more is the fact that a “conversation” isn’t necessarily limited to just one platform. A single conversation can be represented across multiple platforms, including email, text and chat application. Consider this scenario:

• A member of the quality assurance (QA) department at a product manufacturing company sends an email to their boss with safety concerns about a new product about to be launched, which includes a link to a report in Excel stored in OneDrive with more information.
• When the boss doesn’t respond within a reasonable period of time, the QA team member sends a text to the boss saying: “Did you see my email with the report about the product? We may have some real problems here.”
• A short time later, the boss sends a text back, saying: “Yes, I saw it. Let’s discuss the issue among the team and get their reactions. Please post the document on the QA Slack group and spell out your concerns there and I’ll follow up with a meeting request to discuss.”
• The QA team member posts the report up on Slack as instructed and discussion ensues between QA team members about the safety concerns.
• The boss sends a meeting request which, with several members of the QA team working remotely, must be conducted on Teams. Because a couple of the team members are out on vacation, the meeting is recorded so that they watch the discussion once they return.

To capture that full conversation for discovery, a discovery team must collect ESI from Outlook 365, OneDrive, at least one mobile device, Slack and Teams. Then, they must be able to put the conversation together to understand what was discussed, which is considerably challenging!

Case Law Rulings Involving Conversational ESI

As evidence by the lack of case law rulings, standards on defining a “conversation” are elusive. Here are four rulings to consider when deciding how to address conversational based ESI:

• Sandoz, Inc. v. United Therapeutics Corp.: In this case, Plaintiff RareGen, LLC was ordered to produce additional contextual text messages, and not just the messages with search term hits, as the defendant and plaintiff Sandoz has been ordered to do in an earlier order. This ruling illustrates the importance of contextual messages in conversations.
• Lubrizol Corp. v. IBM Corp.: Here, the Court ordered the defendant to produce the entirety of any Slack conversation containing 20 or fewer total messages that had at least one responsive message; and the 10 messages preceding or following any responsive Slack message in a Slack channel containing more than 20 total messages. This ruling was one of the few that attempted (albeit arbitrarily) to define conversations to provide context to messages with hits.
• Pable v. Chicago Transit Authority: In this case, the case was dismissed due to the plaintiff’s failure to preserve Signal and text messages, and the plaintiff (and plaintiff’s counsel) were ordered to pay over $75K in sanctions. This case illustrates just how important preserving conversational data has become.
• We the Protesters, Inc. v. Sinyangwe: Here, the Court ordered the plaintiff to produce messages within 24-hour periods containing at least one responsive message that they had previously redacted as non-responsive or irrelevant. This ruling treats the single day conversation “chunks” as relevant and emphasizes the importance of clear agreements on handling of conversational data.

Based on these four cases, we are far from having a standard in defining a “conversation” for discovery purposes. However, the duty to preserve this data is clear and the expectation to produce contextual messages is becoming standard.

Evolving eDiscovery Strategies for Conversational ESI

The shift to conversational ESI requires a rethinking of collection, review, and production strategies in discovery:

• Collection: Legal and IT teams must work with APIs or third-party tools to extract messages while preserving context, threading, and metadata – a non-trivial task, especially across platforms with proprietary formats.
• Review: Instead of treating messages as standalone documents, review workflows must support the reconstruction of conversations in a way that reflects the user experience. Tools that can display message threads in conversational view, like how users saw them originally, are increasingly essential.
• Production: Producing chat data to opposing counsel raises practical questions: What defines a “document” in a chat? How do you paginate a scrollable thread? How do you handle sensitive emojis or reaction GIFs? Establishing clear ESI protocols and formatting agreements up front are vital to avoiding disputes downstream.
• Search and Relevance: Keyword searching in chat data is notoriously difficult. Abbreviations, typos, emojis, and slang all complicate traditional search techniques. AI-based tools that can interpret intent, sentiment, or topic may offer a path forward.

Conclusion

ESI is not just about documents anymore and conversational-based ESI from text and chat message platforms are a prime example of that. Taming these formats requires discovery teams to break out of the document-paradigm that is ill suited to support these newer data formats. It’s important to work with providers that are keeping abreast of case law precedents and evolving best practices in this rapidly growing area.

For more regarding Cimplifi forensics & collections capabilities, click here.