eDiscovery Processing: Quality Control

Quality control is a process that should occur throughout the entire processing phase. As discussed above, the processed electronic data is evidence. The amount of data being processed can be staggering.

As with any automated process, a set of criteria guides the processing. In some cases a technically correct result is achieved, but the result is not what was expected by the review team. For example, a file may be contained in the source collection that when opened reveals several thousand pages of binary characters. While producing that file may be a technically accurate result of processing, it is impractical for a review team to review thousands of pages of such characters.

By employing both automated and manual quality controls, the client can be notified of the existence of such files, without going through the expense of processing and reviewing said files. Thus, to ensure both a technically correct result as well as deliver the review teams expected result; it is a best practice to employ a combination of automated and manual controls.


Automation provides a great way to audit your electronic discovery process. Automated controls provide a consistently applied methodology to check for many important aspects of an electronic discovery project. In the above example, automation tools can be used to flag such files as containing a disproportionately high number on binary characters or size thresholds. Once the files are flagged, they can then be subject to visual inspection as the rest of the collection is processing, then integrated back into the collection. Other uses for automation in quality control include file count checks at each stage of the process to account for all data as well as checks that fielded information in conformance with field types.

Manual Controls

Manual controls are important to ensure that a review team receives an appropriate result. Visual inspection is an example of a manual control. In the example of the binary character file used above, only a person could determine definitively that there was no readable data contained in the converted version of that file.

Other manual controls include exception handling. Source data is unstructured data, and not all electronic files run through an automated process. There is some level of exception handling, such as password cracking or manipulation of the file to ensure proper rendering of the file as an image. Also, in data collections with a larger percentage of corrupted data, more visual inspection may be needed to ensure that the imaged data is a faithful representation of the original data.

As is true in all industries, the amount of manual work required for a particular collection can have an impact on the cost of the project. As is discussed in the section on Cost Drivers, the cost of collecting and processing electronic discovery can be quite great, and can be affected by a wide variety of factors. Determining the appropriate amount of manual work required for a given project will necessarily have to be weighed in conjunction with the many other factors affecting the project.

Source: EDRM: (edrm.net)