Whitepaper: Six Common Review and Production Issues and How to Fix Them
08-05-2014, Avansic - Ian Campbell, President and Co-Founder of iCONECT, and Gavin W. Manes, CEO of Avansic
In many instances, the “completion” (pretrial presentation) of an e-discovery project is the “production” of documents that will meet the requests made by the other side. The challenge in meeting production requests and specifications can be the variety of document types - from generic (Microsoft) Word and Excel documents to (Adobe) PDF. Some collections may also include CAD drawings, videos, or obscure file formats that are proprietary to an industry. Documents may contain privileged information, trade secrets, Personal Health Information, or Personally Identifiable Information (PII), which require redaction for appropriate protections.
Since productions themselves are meant to contain the relevant documents, a well-thought-out production process is key to both meeting the needs of the production request and the needs of the client. Through many years of experience in creating productions, we've been able to identify a number of issues that may arise, and offer suggestions on how to resolve them.

1. Redaction (Simple):
POTENTIAL ISSUE: There is currently not an accepted method available that allows the in situ redaction of native files.
RESOLUTION: The most accepted solution is not to produce the native, rather to convert the document that needs redaction into an unendorsed, production-quality image file (such as a TIFF) and apply the redactions to that version of the document. This simple process is followed by several other steps that must be performed and quality controlled. Since the text has been changed, the new version will need to be OCRed for the new, redacted text to be made available. Since the file is no longer in the native format, metadata associated with the file will need to be provided elsewhere, typically in a loadfile; however, depending on what was redacted, select metadata may also need to be redacted. For instance, if the subject of the email is redacted, the subject line from the loadfile must also be redacted.

2. Redaction (Document/Family):
POTENTIAL ISSUE: Redaction in native productions that contain email or other file containers (such as zip files) is especially sensitive. Consider an email with attachments; if an attachment contains a redaction, then the email cannot be produced natively because the receiving party could just open the email and extract the non-redacted version of the attachment.
RESOLUTION: In this case, at least the redacted attachment(s) and the email parent will need to be produced as image files (such as TIFFs). The same is true of embedded documents (for example, a spreadsheet within a word processing document). However, it is generally acceptable to produce the extracted text of the parent email if it's not redacted. Similarly, if the parent is redacted and the children are not, the children can still be produced natively.

3. Spreadsheet and Database Files:
POTENTIAL ISSUE: Spreadsheets are notoriously difficult to adapt into a letter-sized format. There may be too many rows or columns to easily read information and they may be blank, thereby creating thousands of pages of unnecessary documents. If TIFFed, reviewers may miss hidden columns, formulas, and other internal data (the same would be true if spreadsheets were printed hard copy). There is no common solution to how these spreadsheets are converted to images; one method is to expose all hidden columns, sheets, etc and another method is to print with last known print settings.
RESOLUTION: The best practice is to review spreadsheets in their native format, taking care to look for hidden rows, columns, sheets, and other nonvisible data.

4. Production Numbering:
POTENTIAL ISSUE: Numbering is a more complex operation that requires two things before beginning: all documents must be in order (generally, email attachment should occur after the parent email – not all tools automatically do this) and the type of numbering matters (i.e., by document or by page? continuation of previous numbering? native and images?). No matter how small, any changes to a production after everything has been numbered may require going back and re-numbering the entire production. For example, removing page 68 or document 52 (due to privilege or other concerns) would leave a gap that could be questioned by the receiving party.
RESOLUTION: To avoid numbering a production multiple times, decide upfront on a numbering schema and only apply that numbering once the production is final.

5. Review Team Communication:
POTENTIAL ISSUE: Confusion regarding process and results of review leading to incorrect coding, production of wrong documents, failure to redact, inappropriate endorsement, and more.
RESOLUTION: Creating and following a review guide, along with diligence in communicating among members of the review team and between clients and vendors, can eliminate many of the time and quality delays in review and production. If the members of the review team communicate with each other and their project manager, it is far more likely that documents will be coded consistently. The fewer the issue/coding tags and a limited number of users adding tags, the cleaner the review. Short tags with long, specific descriptions can help with this as well. Sorting by a large number of tags, such as exhibits for certain witnesses or issue-specific exhibits, can be helpful in the pretrial portion of litigation rather than strictly document review.

6. Review Quality Control:
POTENTIAL ISSUE: With many hands working with the documents throughout the review process there is always the potential for both human error and each party having their own interpretation of what is relevant and what isn't. RESOLUTION: Depending on work flow, the old “measure twice, cut once” rule applies. It's important to perform quality control checks on the reviewers by an entire second pass review using sampling or analytics. For instance, in a proposed production, locate the documents marked as confidential and perform analytics on those documents against the remainder of the set (non-confidential) to determine if there is information in the non-confidential set that should be marked confidential. Then, have reviewers reconsider the documents that analytics found.

Solutions can lie in the steps before production, particularly during review. Whether performed though keyword or conceptual searching or individual linear review, productions can be made easier by using a tool that allows you to organize documents as a precursor to production; a system that is flexible enough to allow foldering and easy sorting. Further, a good review tool can create a seamless handoff to the production stage, whether it be TIFF, native, or a hybrid/blended production. Leveraging the expertise of a vendor and software can greatly increase the quality of a production. And since the production of the documents is one of the final stages, and the documents that are included could contain key exhibits for the case, this all important step must be done with the utmost attention to detail.

Taken on their own, each of these issues can be handled as a review or production proceeds. However, if a project encounters several – or all – of the above, it may behoove the legal team to step back and examine the overall strategy in order to get back on track. It may mean halting review for a day or two but the benefit of proceeding in an orderly fashion for the remainder of the project outweighs the time and cost of re-doing and re-re-doing document batches and productions.