If you speak and write a lot on the same subject, you develop a set of phrases which trip off the tongue semi-automatically. They can easily lose meaning as a result – the words “repeatable, defensible process” have long since lost any meaning because its mindless repetition (and boy, do they repeat it) just floats over one’s head like traffic noise. The message is important – vital even – but once is enough.
We all do it. I have a standard sentence which goes something like this: “You do not need to understand how the technology works, but you need to know what it does, what it costs, and what its benefits are”. Yes, I can see your eyes glazing as I write this, but it is a key message and there are finite ways of expressing it.
We need sometimes to unpick what lies behind these phrases and to expand a little on what we mean. Dean Gonsowski, Global Head of Information Governance at Recommind, is better than most at explaining the meaning behind the well-worn expressions. In an Inside Counsel article called Predictive Coding Cost Savings and Return on Investment Dean tackles the second line of defence raised by lawyers and clients who would like to close their eyes to the benefits which predictive coding technology can bring (the first, of course, is to whine about “black boxes” as an excuse for not even trying to understand what the technology achieves).
Dean’s article is about return on investment. Many companies, especially those ruled by accountants, consider a saving or a return to be invisible if you cannot quantify it precisely, even where common sense shows that an improvement in the bottom line must inevitably result from some action. Dean shows how it is done.
The article starts with the statistics of a typical eDiscovery / eDisclosure exercise, focusing on cost-per-document which, when multiplied by the number of documents, produces a (usually very large) cost. We can’t do much about the review cost per document (that is, the time taken for a competent lawyer to form a human evaluation of its relevance and importance) so we must tackle the number of documents to be reviewed. In Dean’s first example, keyword searches brought a starting volume of over 2 million documents down to 679,349 by getting rid of the easy targets; the use of predictive coding then reduced the reviewable documents from 679,349 to 175,018. Interested?
The article includes a table which compares the time and costs of linear review against the time and cost of the same exercises using predictive coding.
Dean Gonsowski then goes on to consider how the outcome of this calculation translates into a return on investment for those who adopt predictive coding for other cases over longer periods. If the savings look good on a case-by-case basis, then you can start drawing conclusions from the metrics emerging from multiple cases, and set that against the cost of buying and using predictive coding tools.
Most such explanations rapidly run into the sand either by drowning us in detail or by reliance on calculations too complex for the arts graduates which most lawyers are. Dean avoids that, and you do not need to be a genius with figures to get the message.