Migrated from eDJGroupInc.com. Author: Greg Buckles. Published: 2010-02-19 04:14:51Format, images and links may no longer function correctly. One of the topics bouncing around LTNY was the new SaaS pricing of $100/GB/year for processing, hosting and review by Houston based Sfile Technology Corporation. In this world of hidden pricing, NDA-wrapped RFPs and other hide-the-ball licensing games, Sfile is publishing their upfront pricing and undercutting the current market average of $300-500/GB by a huge margin. Aside from the usual belly-aching from other processing-hosting providers, I expected that most of my corporate and firm clients would be excited about the potential downward price pressure.
Instead, several expressed serious concern over last year’s plunging volume based pricing. That is right, concern that the market was pushing prices too low to actually sustain the quality expected in discovery. We have seen volume pricing slide from over $2,000/GB to somewhere in the $300-500/GB range in a very short time period. Many, including myself, railed against the artificially high prices back in 2000-2005 when our selection was limited to three to four national providers or local boutique shops. Discovery processing, hosting and production services have indeed become commoditized, but they have done so in the absence of certifications and standards bodies to protect the consumer against raw ‘pump and dump’ practices that ignore real QA/QC protocol and hide all of the ESI that could not be accessed, processed, indexed or fully rendered for proper attorney review.
Let us consider what should be involved in fully processing a single GB of typical custodial ESI. For the purposes of our scenario, we can assume that we actually have a representative mixture of communications, native files, local files and other enterprise data sources. Depending on the business vertical and the user’s job, the number of individualized (exploded) items per GB ranges from 2,500 – 10,000+. Given so many variables, we will just say that our sample has 5,000 items after everything is unpacked and readied for review. The provider has to first properly receive custody of the data, create a clean copy to work from and document all the steps for a new project in their environment. This is not magic, but doing it right requires well thought out protocol to prevent contamination of the source media and proper segregation of individual data sets from other clients and matters.
Once the initial inventory and processing is completed, the 5,000 items should be ready for de-duplication, search criteria execution and other culling steps. Each stage of data processing should include Quality Assurance reports/checks to track the total number of items and other integrity attributes. We expect a certain number of problem files, called ‘exceptions’ to occur within any diverse set of ESI. Most providers either downplay the potential importance of corrupted files, embedded objects, system files and other types of exception files, but these files are the key reason that a company would outsource ESI processing. Anyone can buy software to do basic processing. Even if you assume that the exceptions comprise less than one half of one percent of our sample, someone must still check these 250 files and clearly document what we do know about them and why we think that they could not be processed.
You could just look at the grouped exception report and take it for granted when it says that files are encrypted, corrupt, password protected or otherwise inaccessible. But how will you really know if you do not take the reasonable step of attempting to manually open at least a representative portion of them. In my days as a litigation support manager, I remember a provider claiming large quantities of Adobe PDF files were corrupt on a case, only to determine that one of our departments had a system that was writing to a little know PDF-A variant that the provider’s system could not open.
This is just performing quality control checks on the known exception files. It does not include any random sampling of the ‘good’ ESI to confirm that the system properly logged all the issues. Knowing what may be at stake in your matter, I am betting that you want someone to document reasonable manual checks so that you have confidence that hash values are unchanged, item text is fully indexed, and container files are properly handled. Providers talk up the value of de-duplication, but how many perform some level of statistical verification on YOUR data to ensure that items are not inadvertently suppressed. Many corporations are migrating from Exchange 2003 to 2007 or the new 2010 platform. Each of these handles email envelopes, item classes and BCC information slightly differently.
In our last step, the culled items are loaded to your review platform and potentially re-indexed (as many of the processing and review systems use different indexes). You can see where we need another QA/QC round. For these 5,000 items, even if we just do a one percent of the loaded items against the originals, you are still talking about 500 views and field property checks. We now have to store the original media and can removed the staged versions if everything has been accepted by the client. The real cost here is all technical and project management labor. No one makes a profit on the small jobs when you consider that experienced litigation support staff should make $40-80k.
So how can providers offer such low volume pricing? When you did into the details, you will often find serious limitations or partial pricing. In Sfile’s case, their standard processing rate does not include exception handling at all or any conversion of files. There are cases where the relevance criteria and collection methods yield only easy MS Office files with a very low exception rate. Some matters may resolve themselves in less than a year. For every matter profile, there should be a good defensible process that fits the customer budget. That is what we keep hoping for in the market. Until we reach that world of eDiscovery Nirvana, the old adage still applies, “You get what you pay for.”