The Main Event, Review Fundamentals Series Part 1

A multi-part series on the fundamentals eDiscovery practitioners need to know about document review planning and execution

So far, in our three Fundamentals Series, we have covered the fundamentals of collection, of processing, and of early case assessment.  We turn our attention now to the most time-consuming and expensive phase of an electronic discovery effort: review.

Numerous analyses have established that document review is far and away the most expensive discovery activity:

• One study estimated that 73% of the total costs to produce materials in discovery are attributable to document review
• Another study estimated that 58% of the total costs to produce materials in discovery are attributable to document review
• A market commentary estimated that approximately $5 are spent on review for every $1 spent on processing
• A survey found that “discovery costs for attorney review alone were roughly one-fourth of the total outside legal fees” [emphasis added]

The reason for these significant costs is the irreducible need for qualified people to spend time looking at a significant number of documents to make nuanced determinations about their relevance, their privilege, and much more.

Regardless of whether the required hours of review work are concentrated in a small case team or spread among a large contract review team, the total number of hours required has the potential to be in the hundreds or even thousands.  Even when using some form of technology-assisted review, the effort is still likely to be time-consuming and expensive relative to the rest of discovery.  Moreover, the quality and consistency of all those hours or work must be ensured so that all relevant information can be uncovered, so that complete responses can be provided, and so that inadvertent disclosure of privileged or confidential material can be avoided.

To help you to meet these challenges, we’re going to break review down into five subparts and discuss each in turn:

1. What gets reviewed
2. For what it gets reviewed
3. By whom it gets reviewed
4. Workflow design considerations
5. Quality control

We begin, in this Part, with what gets reviewed.

What Gets Reviewed

The first step in any review project is determining what materials are going to get reviewed, because the volume and composition of those materials will inform your subsequent decisions about who does the reviewing and how they go about it.  Without a clear picture of the what, you cannot make an effective plan for the who and the how.

In this context, we are not talking about the kinds of source identification that take place during preservation and collection.  Rather, we are talking about identifying what requires review from within the pool of collected, processed materials already loaded into an eDiscovery platform.  Traditionally, during processing, this pool of loaded materials has already:

• Had its system files, etc., removed
• Had its duplicates identified and removed
• Had any date restrictions applied

This will have left you with a pool of unique files, from the relevant time period, that could contain relevant information.  Ideally, you will then have engaged in some of the early case assessment activities we’ve previously discussed to gather information about the contents of this pool to help you decide what gets reviewed and how best to go about it.

From this pool then, you must decide whether everything gets reviewed, whether only the results of certain searches and filters get reviewed, whether only the results of a TAR process get reviewed, or whether some hybrid plan is employed:

• Everything
  • For smaller pools of materials (e., those with fewer than 10,000 documents) the simplest, fastest solution is often to just review everything. Reviewing everything is also the typical approach when the pool in question was derived not from materials you collected but from productions received from other parties.

• Search and Filter Results
  • Identifying your ultimate review set through the application of searches and filters is, by far, the most common approach. This typically requires reviewing both the results of the chosen searches and filters, as well as some of the remainder to verify its irrelevance.

• TAR Process Results
  • For particularly large pools of materials, or those for which effective searches can’t be crafted, or those for which speed is of the essence, a TAR process can be employed to identify the relevant materials within the pool. Those materials are then reviewed, along with some of the remainder to verify its irrelevance.

• Hybrid Plans
  • It is also common to employ a hybrid of these approaches specific to the exigencies of the case. For example, you might review all of the materials collected from the most critical custodian and then apply searches or a TAR process to the remaining materials.  Similarly, targeted searches might be used to quickly identify the most important materials for immediate review, and then a TAR process might be applied to everything else afterwards.

Whichever path you choose, you will also need to make determinations about the handling of families, threads, and near-duplicates to finalize your review set:

• Families
  • “Families” refers to the family groups of related documents, such as “parent” emails and “child” attachments. If you are reviewing everything, all family group members will already be included in your review set, but if you have applied searches or a TAR process, the results of those efforts will not be family group complete.  You will have a choice about whether to pull related family members in or to just review the actual results.  Most of the time, they are included – both for the context they provide and because production by family group is common.

• Threads
  • “Threads” refers to the threads of related emails going back and forth between participants, which often contain within themselves the text of the messages that preceded them. The single email at the end of the thread may contain the complete thread within itself.  Such emails are called “inclusive emails.”  Most review platforms will give you the option to identify inclusive emails and limit review to just those, excluding from the review set all of the individual preceding emails (and allowing review decisions to be propagated to them).

• Near-Duplicates
  • “Near-duplicates” refers to those documents that are extremely similar (or superficially identical) to other documents in your collection but that were not removed by deduplication during processing due to some small variation between them (e.g., a difference in source directory or another metadata property). Most review platforms will also give you the option to identify near-duplicates, either for grouped inclusion in the review set, or to exclude all but one instance (and then propagate review decisions across the group).

Scope and Process Negotiations

One of the most important factors in determining what gets included in your review set is the scope limitations and process decisions you negotiate with the other parties before, during, and after the meet and confer.  It is common to negotiate agreements to limit the scope to specific custodians, to specific enterprise sources, to specific date ranges, to specific file types, and more.  It is also common to negotiate over what searches should be run, what TAR process should be used, and other aspects of the review set identification process.  The more scope limitations you can negotiate the less time and money you will have to spend on review, and the more process elements you can negotiate up front, the fewer decisions you may need to defend later.

Upcoming in this Series

In the next Part, we will continue our discussion of review fundamentals with a look at what you’re reviewing for and who’s doing the reviewing.