Big Data

Pros and Cons: Predictive Coding – Should I Invest?

Pros and Cons: Predictive Coding – Should I Invest?

Years of eDiscovery and painstaking document review processes has now led to the buzzy new technology known as predictive coding. It’s yet another tool in technology assisted review that can sort like documents from algorithmic search criteria. It’s the next big thing law firms are using to reduce attorney time, but does it have a real purpose in the IT world? Here are some factors to consider when investing.

Pros and Cons of Predictive Coding


Great at making binary decisions

Predictive coding works best when you have a large volume of data and simply want to know whether a document is responsive or not responsive to a given set of criteria. An algorithm can eliminate irrelevant data in a snap, and for organizations that have a clear goal in mind, this can expedite the process enormously in providing a small sample size of documents, anywhere from five to 20 percent.

Big cost saver if used effectively

On paper, it seems obvious that predictive coding would go faster than traditional search terms, but is it fast enough that it’s worth investing in? eDiscovery counsel Shannon Capone Kirk claimed that on one occasion, the use of predictive coding reduced attorney time by seven weeks for up to $300,000 in savings.

Statistical results can be a great help

There is such a thing as Transparent Predictive Coding that provides users with information about the search it performed. Through Prediction Insight, users get a numerical probability along with relevant metadata that can allow you to decide whether a document really is important.

It can learn over time

Predictive Coding isn’t science fiction, but Directed Training technology makes it such that the software actually learns as you refine the search results, going as far as to provide suggestions on new documents and searches.


Bad at making nuanced decisions

Predictive coding has not yet perfected the ability to refine results based on factors other than responsiveness. So when it comes to finding essential work product or attorney-client communication information, it falls short.

Not a substitute for other, simpler search tools

Traditional document review tools aren’t dead and gone just yet. They remain cheaper and still effective ways to collect data. Technology that can analyze conversational patterns in email threads or find email attachments could better serve your company’s needs than predictive coding.

It’s only as smart as its teacher

Predictive coding depends on the amount of data being analyzed, the tool being used (of which there are several) and the user’s end goal. If a goal and the search terms aren’t strong, then the results won’t be either.

Legal sanctions still a risk

Your obligations as a defendant involve providing a reasonable inquiry that all documents have been accurately, thoroughly and truthfully reviewed. The problem is, no law is yet on the books that sounds off one way or another on the effectiveness of a given piece of predictive coding software in providing a reasonable search. It’s all dependent on the case. So if your search is ultimately contested and it’s found that the process was not sufficiently supervised, your company could be at risk of additional costs from legal sanctions.

False Positives won’t save you any time or money

Predictive coding isn’t fool-proof, and any shortcomings are based on either a bad product or a bad user. If this is the case, you’ll get a lot of false positives, or data tagged as responsive when it’s really just irrelevant. Having to sift through all of the errors won’t save you time or money, so be sure your search is accurate.

Find out more on predictive coding from a legal expert, Shannon Capone Kirk, on our Viewpoint: Is Predictive Coding a Panacea?

Explore More


CIO Talk Network

CIO Talk Network, Editor, CIO Talk Network

CIO Talk Network is a trusted resource for Business and IT Leaders offering targeted, real-life, in-depth, unbiased, and thought-provoking content and conversations on topics critical to their success. CTN features thought leadership direct... More   View all posts
Add Comment
Click here to post a comment


Automate Everything - The New Cupid for BPOs 1 MPU 300X250
CIO Talk Network


Not Member Yet?


  • Name

  • Contact Info

  • About Yourself

  • Minimum length of 8 characters
  • Upload
  • Location

  • Professional Background

  • Other Social Profiles

  • Areas of Interest