Close
Enter your
text here

Analysis

eDiscovery Best Practices: Message Thread Review Saves Costs and Improves Consistency
Insanity is doing the same thing over and over again and expecting a different result.  But, in ESI review, it can be even worse when you get a different result.

One of the biggest challenges when reviewing ESI is identifying duplicates so that your reviewers aren’t reviewing the same files again and again.  Not only does that drive up costs unnecessarily, but it could lead to problems if the same file is categorized differently by different reviewers (for example, inadvertent production of a duplicate of a privileged file if it is not correctly categorized).

Of course, there are a number of ways to identify duplicates.  Exact duplicates (that contain the exact same content in the same file format) can be identified through hash values, which are a digital fingerprint of the content of the file.  MD5 and SHA-1 are the most popular hashing algorithms, which can identify exact duplicates of a file, so that they can be removed from the review population.  Since many of the same emails are emailed to multiple parties and the same files are stored on different drives, deduplication through hashing can save considerable review costs.

Sometimes, files are not exact duplicates but contain the same (or almost the same) information.  One example is a Word document published to an Adobe PDF file – the content is the same, but the file format is different, so the hash value will be different.  Near-deduplication can be used to identify files where most or all of the content matches so they can be verified as duplicates and eliminated from review.

Then, there is message thread analysis.  Of course, most email messages are part of a larger discussion, which could be just between two parties, or include a number of parties in the discussion.  To review each email in the discussion thread would result in much of the same information being reviewed over and over again.  Instead, message thread analysis pulls those messages together and enables them to be reviewed as an entire discussion.  That includes any side conversations within the discussion that may or may not be related to the original topic (e.g., a side discussion about lunch plans or did you see American Idol last night).

FirstPass®, powered by Venio FPR™, is one example of an application that provides a mechanism for message thread analysis of Outlook emails that pulls the entire thread into one conversation for review as one big “tree”.  The “tree” representation gives you the ability to see all of the conversations within the discussion and focus your review on the last emails in each conversation to see what is said without having to review each email.  Side conversations are “branches” of the tree and FirstPass enables you to tag individual messages, specific branches or the entire tree as responsive, non-responsive, privileged or some other designation.  Also, because of the way that Outlook tracks emails in the thread, FirstPass identifies messages that are missing from the collection with a red X, enabling you to investigate and determine if additional collection is needed and avoiding potential spoliation claims.

With message thread analysis, you can minimize review of duplicative information within emails, saving time and cost and ensuring consistency in the review.

So, what do you think?  Does your review tool support message thread analysis?   Please share any comments you might have or if you’d like to know more about a particular topic.

eDiscovery Best Practices: Competency Ethics – It’s Not Just About the Law Anymore
A few months ago at LegalTech New York, I conducted a thought leader interview with Tom O’Connor of Gulf Coast Legal Technology Center, who didn’t exactly mince words when talking about the trend for attorneys to “finally tak[e] technology seriously”.  As he noted, “lawyers are finally trying to take some time to try to get up to speed – whining and screaming pitifully all the way about how it’s not fair, and the sanctions are too high and there’s too much data.  Get a life, get a grip.  Use the tools that are out there that have been given to you for years.”

Strong words, indeed.  The American Bar Association (ABA) Model Rules of Professional Conduct (Model Rules) require that an attorney possess and demonstrate a certain requisite level of knowledge in order to be considered competent to handle a given matter.  Specifically, Model Rule 1.1 states that, “[a] lawyer shall provide competent representation to a client. Competent representation requires the legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation.”

Preparation not only means understanding a specific area of the law (for example, antitrust or patent law, both highly specialized.).  It also means having the technical knowledge and skills necessary to serve the client in the area of discovery.

The ethical responsibilities of counsel these days includes competently directing and managing the identification, preservation, collection, processing, analysis, review and production of electronically stored information (ESI) required to be produced pursuant to lawful discovery requests.  If counsel does not have that level of competency in a particular area, he or she is obligated to either acquire the knowledge or skill necessary to support those needs, or include someone else who does have the requisite skills as part of the representation.

Not too long ago, I met with an attorney and discussed how they handled preservation obligations with their clients.  The attorney indicated that he expected his clients to self-manage their own preservation and collection.  When I asked him why he didn’t try to get more involved to make sure it was being handled properly, he said, “I don’t want to alarm them.  They might decide they need a bigger firm.”

Recent case law is full of cases where counsel didn’t fully understand their eDiscovery obligations, and got themselves and their clients “burned” in the process.  If your organization gets involved in litigation, make sure to include eDiscovery competence among the factors you consider when determining counsel qualifications to represent you.

So, what do you think?  Is your counsel eDiscovery savvy?  If not, do they use a provider that is?  Please share any comments you might have or if you’d like to know more about a particular topic.

eDiscovery Best Practices: 4 Steps to Effective eDiscovery With Software Analytics
I read an interesting article from Texas Lawyer via Law.com entitled “4 Steps to Effective E-Discovery With Software Analytics” that has some interesting takes on project management principles related to eDiscovery and I’ve interjected some of my thoughts into the analysis below.  A copy of the full article is located here.  The steps are as follows:

1. With the vendor, negotiate clear terms that serve the project’s key objectives.  The article notes the important of tying each collection and review milestone (e.g., collecting and imaging data; filtering data by file type; removing duplicates; processing data for review in a specific review platform; processing data to allow for optical character recognition (OCR) searching; and converting data into a tag image file format (TIFF) for final production to opposing counsel) to contract terms with the vendor.

The specific milestones will vary – for example, conversion to TIFF may not be necessary if the parties agree to a native production – so it’s important to know the size and complexity of the project, and choose only an experienced eDiscovery vendor who can handle the variations.

2. Collect and process data.  Forensically sound data collection and culling of obviously unresponsive files (such as system files) to drastically decrease the overall review costs are key services that a vendor provides in this area.  As we’ve noted many times on this blog, effective culling can save considerable review costs – each gigabyte (GB) culled can save $16-$18K in attorney review costs.

The article notes that a hidden cost is the OCR process of translating extracted text into a searchable form and that it’s an optimal negotiation point with the vendor.  This may have been true when most collections were paper based, but as most collections today are electronic based, the percentage of documents requiring OCR is considerably less than it used to be.  However, it is important to be prepared that there are some native files which will be “image only”, such as TIFFs and scanned PDFs – those will require OCR to be effectively searched.

3. Select a data and document review platform.  Factors such as ease of use, robustness, and reliability of analytic tools, support staff accessibility to fix software bugs quickly, monthly user and hosting fees, and software training and support fees should be considered when selecting a document review platform.

The article notes that a hidden cost is selecting a platform with which the firm’s litigation support staff has no experience as follow-up consultation with the vendor could be costly.  This can be true, though a good vendor training program and an intuitive interface can minimize or even eliminate this component.

The article also notes that to take advantage of the vendor’s more modern technology “[a] viable option is to use a vendor’s review platform that fits the needs of the current data set and then transfer the data to the in-house system”.  I’m not sure why the need exists to transfer the data back – there are a number of vendors that provide a cost-effective solution appropriate for the duration of the case.

4. Designate clear areas of responsibility.  By doing so, you minimize or eliminate inefficiencies in the project and the article mentions the RACI matrix to determine who is responsible (individuals responsible for performing each task, such as review or litigation support), accountable (the attorney in charge of discovery), consulted (the lead attorney on the case), and informed (the client).

Managing these areas of responsibility effectively is probably the biggest key to project success and the article does a nice job of providing a handy reference model (the RACI matrix) for defining responsibility within the project.

So, what do you think?  Do you have any specific thoughts about this article?   Please share any comments you might have or if you’d like to know more about a particular topic.

eDiscovery Trends: Sedona Conference Database Principles
A few months ago, eDiscovery Daily posted about discovery of databases and how few legal teams understand database discovery and know how to handle it.  We provided a little pop quiz to test your knowledge of databases, with the answers here.

Last month, The Sedona Conference® Working Group on Electronic Document Retention & Production (WG1) published the Public Comment Version of The Sedona Conference® Database Principles – Addressing the Preservation & Production of Databases &Database Information in Civil Litigation to provide guidance and recommendations to both requesting and producing parties to simplify discovery of databases and information derived from databases.  You can download the publication here.

As noted in the Executive Overview of the publication, some of the issues that make database discovery so challenging include:

  • More enterprise-level information is being stored in searchable data repositories, rather than in discrete electronic files,
  • The diverse and complicated ways in which database information can be stored has made it difficult to develop universal “best-practice” approaches to requesting and producing information stored in databases,
  • Retention guidelines that make sense for archival databases (databases that add new information without deleting past records) rapidly break down when applied to transactional databases where much of the system’s data may be retained for a limited time – as short as thirty days or even thirty seconds.

The commentary is broken into three primary sections:

  • Section I: Introduction to databases and database theory,
  • Section II: Application of The Sedona Principles, designed for all forms of ESI, to discovery of databases,
  • Section III: Proposal of six new Principles that pertain specifically to databases with commentary to support the Working Group’s recommendations.  The principles are stated as follows:
    • Absent a specific showing of need or relevance, a requesting party is entitled only to database fields that contain relevant information, not the entire database in which the information resides or the underlying database application or database engine.
    • Due to differences in the way that information is stored or programmed into a database, not all information in a database may be equally accessible, and a party’s request for such information must be analyzed for relevance and proportionality.
    • Requesting and responding parties should use empirical information, such as that generated from test queries and pilot projects, to ascertain the burden to produce information stored in databases and to reach consensus on the scope of discovery.
    • A responding party must use reasonable measures to validate ESI collected from database systems to ensure completeness and accuracy of the data acquisition.
    • Verifying information that has been correctly exported from a larger database or repository is a separate analysis from establishing the accuracy, authenticity, or admissibility of the substantive information contained within the data.
    • The way in which a requesting party intends to use database information is an important factor in determining an appropriate format of production.

To submit a public comment, you can download a public comment form here, complete it and fax (yes, fax) it to The Sedona Conference® at 928-284-4240.  You can also email a general comment to them at tsc@sedona.net.

eDiscovery Daily will be delving into this document in more detail in future posts.  Stay tuned!

So, what do you think?  Do you have a need for guidelines for database discovery?   Please share any comments you might have or if you’d like to know more about a particular topic.

eDiscovery Best Practices: Your ESI Collection May Be Larger Than You Think
Here’s a sample scenario: You identify custodians relevant to the case and collect files from each.  Roughly 100 gigabytes (GB) of Microsoft Outlook email PST files and loose “efiles” is collected in total from the custodians.  You identify a vendor to process the files to load into a review tool, so that you can perform first pass review and, eventually, linear review and produce the files to opposing counsel.  After processing, the vendor sends you a bill – and they’ve charged you to process over 200 GB!!  What happened?!?

Did the vendor accidentally “double-bill” you?  That would be great – but no.  There’s a much more logical explanation and, unfortunately, you may wind up paying a lot more to process these files that you expected.

Many of the files in most ESI collections are stored in what are known as “archive” or “container” files.  For example, as noted above, Outlook emails are typically saved for each custodian in a personal storage (.PST) file format, which is an expanding container file. For most custodians, all of their email (and the corresponding attachments, if present) resides in a few PST files.  The scanned size for the PST file is the size of the file on disk.

Did you ever see one of those vacuum bags that you store clothes in and then suck all the air out so that the clothes won’t take as much space?  The PST file is like one of those vacuum bags – it typically stores the emails and attachments in a compressed format to save space.  When the emails and attachments are processed into a review tool, they are expanded into their normal size.  This expanded size can be 1.5 to 2 times larger than the scanned size (or more).  And, that’s what many vendors will bill on – the expanded size.

There are other types of archive container files that compress the contents – .zip and .rar files are two examples of compressed container files.  These files are often used to not only to compress files for storage on hard drives, but they are also used to compact or group a set of files when transmitting them, usually in – you guessed it – email.  With email comprising a majority of most ESI collections and the popularity of other archive container files for compressing file collections, the expanded size of your collection may be considerably larger than it appears when stored on disk.  It’s important to be prepared for that and know your options when processing that data, so you can effectively anticipate those processing costs.

So, what do you think?  Have you ever been surprised by processing costs of your ESI?   Please share any comments you might have or if you’d like to know more about a particular topic.

eDiscovery Best Practices: Testing Your Search Using Sampling

Friday, we talked about how to determine an appropriate sample size to test your search results as well as the items NOT retrieved by the search, using a site that provides a sample size calculator.  Yesterday, we talked about how to make sure the sample size is randomly selected.

Today, we’ll walk through an example of how you can test and refine a search using sampling.

TEST #1: Let’s say in an oil company we’re looking for documents related to oil rights.  To try to be as inclusive as possible, we will search for “oil” AND “rights”.  Here is the result:

  • Files retrieved with “oil” AND “rights”: 200,000
  • Files NOT retrieved with “oil” AND “rights”: 1,000,000

Using the site to determine an appropriate sample size that we identified before, we determine a sample size of 662 for the retrieved files and 664 for the non-retrieved files to achieve a 99% confidence level with a margin of error of 5%.  We then use this site to generate random numbers and then proceed to review each item in the retrieved and NOT retrieved items sets to determine responsiveness to the case.  Here are the results:

  • Retrieved Items: 662 reviewed, 24 responsive, 3.6% responsive rate.
  • NOT Retrieved Items: 664 reviewed, 661 non-responsive, 99.5% non-responsive rate.

Nearly every item in the NOT retrieved category was non-responsive, which is good.  But, only 3.6% of the retrieved items were responsive, which means our search was WAY over-inclusive.  At that rate, 192,800 out of 200,000 files retrieved will be NOT responsive and will be a waste of time and resource to review.  Why?  Because, as we determined during the review, almost every published and copyrighted document in our oil company has the phrase “All Rights Reserved” in the document and will be retrieved.

TEST #2: Let’s try again.  This time, we’ll conduct a phrase search for “oil rights” (which requires those words as an exact phrase).  Here is the result:

  • Files retrieved with “oil rights”: 1,500
  • Files NOT retrieved with “oil rights”: 1,198,500

This time, we determine a sample size of 461 for the retrieved files and (again) 664 for the NOT retrieved files to achieve a 99% confidence level with a margin of error of 5%.  Even though, we still have a sample size of 664 for the NOT retrieved files, we generate a new list of random numbers to review those items, as well as the 461 randomly selected retrieved items.  Here are the results:

  • Retrieved Items: 461 reviewed, 435 responsive, 94.4% responsive rate.
  • NOT Retrieved Items: 664 reviewed, 523 non-responsive, 78.8% non-responsive rate.

Nearly every item in the retrieved category was responsive, which is good.  But, only 78.8% of the NOT retrieved items were responsive, which means over 20% of the NOT retrieved items were actually responsive to the case (we also failed to retrieve 8 of the items identified as responsive in the first iteration).  So, now what?

TEST #3: If you saw this previous post, you know that proximity searching is a good alternative for finding hits that are close to each other without requiring the exact phrase.  So, this time, we’ll conduct a proximity search for “oil within 5 words of rights”.  Here is the result:

  • Files retrieved with “oil within 5 words of rights”: 5,700
  • Files NOT retrieved with “oil within 5 words of rights”: 1,194,300

This time, we determine a sample size of 595 for the retrieved files and (once again) 664 for the NOT retrieved files, generating a new list of random numbers for both sets of items.  Here are the results:

  • Retrieved Items: 595 reviewed, 542 responsive, 91.1% responsive rate.
  • NOT Retrieved Items: 664 reviewed, 655 non-responsive, 98.6% non-responsive rate.

Over 90% of the items in the retrieved category were responsive AND nearly every item in the NOT retrieved category was non-responsive, which is GREAT.  Also, all but one of the items previously identified as responsive was retrieved.  So, this is a search that appears to maximize recall and precision.

Had we proceeded with the original search, we would have reviewed 200,000 files – 192,800 of which would have been NOT responsive to the case.  By testing and refining, we only had to review 8,815 files –  3,710 sample files reviewed plus the remaining retrieved items from the third search (5,700 – 595 = 5,105) – most of which ARE responsive to the case.  We saved tens of thousands in review costs while still retrieving most of the responsive files, using a defensible approach.

Keep in mind that this is a simple example — we’re not taking into account misspellings and other variations we may want to include in our criteria.

So, what do you think?  Do you use sampling to test your search results?   Please share any comments you might have or if you’d like to know more about a particular topic.

eDiscovery Best Practices: A “Random” Idea on Search Sampling
Friday, we talked about how to determine an appropriate sample size to test your search results as well as the items NOT retrieved by the search, using a site that provides a sample size calculator.  Today, we’ll talk about how to make sure the sample size is randomly selected.

A randomly selected sample gives each file an equal chance of being reviewed and eliminates the chance of bias being introduced into the sample which might skew the results.  Merely selecting the first or last x number of items (or any other group) in the set may not reflect the population as a whole – for example, all of those items could come from a single custodian.  To ensure a fair, defensible sample, it needs to be selected randomly.

So, how do you select the numbers randomly?  Once again, the Internet helps us out here.

One site, Random.org, has a random integer generator which will randomly generate whole numbers.  You simply need to supply the number of random integers that you need to be generated, the starting number and ending number of the range within which the randomly generated numbers should fall.  The site will then generate a list of numbers that you can copy and paste into a text file or even a spreadsheet.  The site also provides an Advanced mode, which provides options for the numbers (e.g., decimal, hexadecimal), output format and how the randomization is ‘seeded’ (to generate the numbers).

In the example from Friday, you would provide 660 as the number of random integers to be generated, with a starting number of 1 and an ending number of 100,000 to get a list of random numbers for testing your search that yielded 100,000 files with hits (664, 1 and 1,000,000 respectively to get a list of numbers to test the non-hits).  You could paste the numbers into a spreadsheet, sort them and then retrieve the files by position in the result set based on the random numbers retrieved and review each of them to determine whether they reflect the intent of the search.  You’ll then have a good sense of how effective your search was, based on the random sample.  And, probably more importantly, using that random sample to test your search results will be a highly defensible method to verify your approach in court.

Tomorrow, we’ll walk through a sample iteration to show how the sampling will ultimately help us refine our search.

So, what do you think?  Do you use sampling to test your search results?   Please share any comments you might have or if you’d like to know more about a particular topic.

eDiscovery Best Practices: Determining Appropriate Sample Size to Test Your Search
We’ve talked about searching best practices quite a bit on this blog.  One part of searching best practices (as part of the “STARR” approach I described in an earlier post) is to test your search results (both the result set and the files not retrieved) to determine whether the search you performed is effective at maximizing both precision and recall to the extent possible, so that you retrieve as many responsive files as possible without having to review too many non-responsive files.  One question I often get is: how many files do you need to review to test the search?

If you remember from statistics class in high school or college, statistical sampling is choosing a percentage of the results population at random for inspection to gather information about the population as a whole.  This saves considerable time, effort and cost over reviewing every item in the results population and enables you to obtain a “confidence level” that the characteristics of the population reflect your sample.  Statistical sampling is a method used for everything from exit polls to predict elections to marketing surveys to poll customers on brand popularity and is a generally accepted method of drawing conclusions for an overall results population.  You can sample a small portion of a large set to obtain a 95% or 99% confidence level in your findings (with a margin of error, of course).

So, does that mean you have to find your old statistics book and dust off your calculator or (gasp!) slide rule?  Thankfully, no.

There are several sites that provide sample size calculators to help you determine an appropriate sample size, including this one.  You’ll simply need to identify a desired confidence level (typically 95% to 99%), an acceptable margin of error (typically 5% or less) and the population size.

So, if you perform a search that retrieves 100,000 files and you want a sample size that provides a 99% confidence level with a margin of error of 5%, you’ll need to review 660 of the retrieved files to achieve that level of confidence in your sample (only 383 files if a 95% confidence level will do).  If 1,000,000 files were not retrieved, you would only need to review 664 of the not retrieved files to achieve that same level of confidence (99%, with a 5% margin of error) in your sample.  As you can see, the sample size doesn’t need to increase much when the population gets really large and you can review a relatively small subset to understand your collection and defend your search methodology to the court.

On Monday, we will talk about how to randomly select the files to review for your sample.  Same bat time, same bat channel!

So, what do you think?  Do you use sampling to test your search results?   Please share any comments you might have or if you’d like to know more about a particular topic.

Working Successfully with eDiscovery and Litigation Support Service Providers: Other Evaluation Criteria
In the last posts in this blog series, we talked about evaluating service provider pricing, quality, scalability and flexibility.  There are a few other things you may wish to look at as well, that may be especially significant for large, long-term projects or relationships.  Those things are:
  1. Litigation Experience:  Select a service provider that has litigation experience versus general business experience.   A non-litigation service provider that does scanning — for example — may be able to technically meet your requirements.  They are probably not, however, accustomed to the inflexible schedules and changing priorities that are commonplace in litigation work.
  2. Corporate Profile and Tenure:  For a large project, be sure to select a service provider that’s been around for a while and has a proven track record.  You want to be confident that the service provider that starts your project will be around to finish your project.
  3. Security and Confidentiality:  You want to ensure that your documents, data, and information are secure and kept confidential.  This means that you require a secure physical facility, secure systems, and appropriate confidentiality guidelines and agreements.
  4. SaaS Service Providers: For them, you need to evaluate the technology functionality and ensure that it includes the features you require, that those features are easy to access and to use, and that access, system reliability, system speed, and system security meet your requirements.
  5. Facility Location and Accessibility:  For many projects and many types of services, it won’t be necessary to spend time on the project site.   For other projects, that might not be the case.  For example, if a service provide is staffing a large document review project at its facility, the litigation team may need to spend time at the facility overseeing work and doing quality control reviews.  In such a case, the geographic location and the facility’s access to airports and hotels may be a consideration.

A lot goes into selecting the right service provider for a project, and it’s worth the time and effort to do a careful, thorough evaluation.  In the next posts in this series, we’ll discuss the vendor evaluation and selection process.

What has been your experience with evaluating and selecting service providers?  What evaluation criteria have you found to be most important?  Please share any comments you might have and let us know if you’d like to know more about an eDiscovery topic.

eDiscovery Trends: Forbes on the Rise of Predictive Coding
First the New York Times with an article about eDiscovery, now Forbes.  Who’s next, The Wall Street Journal?  😉

Forbes published a blog post entitled E-Discovery And the Rise of Predictive Coding a few days ago.  Written by Ben Kerschberg, Founder of Consero Group LLC, it gets into some legal issues and considerations regarding predictive coding that are interesting.  For some background on predictive coding, check out our December blog posts, here and here.

First, the author provides a very brief history of document review, starting with bankers boxes and WordPerfect and “[a]fter an interim phase best characterized by simple keyword searches and optical character recognition”, it evolved to predictive coding.  OK, that’s like saying that Gone with the Wind started with various suitors courting Scarlett O’Hara and after an interim phase best characterized by the Civil War, marriage and heartache, Rhett says to Scarlett, “Frankly, my dear, I don’t give a damn.”  A bit oversimplification of how review has evolved.

Nonetheless, the article gets into a couple of important legal issues raised by predictive coding.  They are:

  • Satisfying Reasonable Search Requirements: Whether counsel can utilize the benefits of predictive coding and still meet legal obligations to conduct a reasonable search for responsive documents under the federal rules.  The question is, what constitutes a reasonable search under Federal Rule 26(g)(1)(A), which requires that the responding attorney attest by signature that “with respect to a disclosure, it is complete and correct as of the time it is made”?
  • Protecting Privilege: Whether counsel can protect attorney-client privilege for their client when a privileged document is inadvertently disclosed.  Fed. Rule of. Evidence 502 provides that a court may order that a privilege or protection is not waived by disclosure if the disclosure was inadvertent and the holder of the privilege took reasonable steps to prevent disclosure.  Again, what’s reasonable?

The author concludes that the use of predictive coding is reasonable, because it a) makes document review more efficient by providing only those documents to the reviewer that have been selected by the algorithm; b) makes it more likely that responsive documents will be produced, saving time and resources; and c) refines relevant subsets for review, which can then be validated statistically.

So, what do you think?  Does predictive coding enable attorneys to satisfy these legal issues?   Is it reasonable?  Please share any comments you might have or if you’d like to know more about a particular topic.