New Zealand Law Society - Facilitating a more proportionate and cost effective discovery process

Facilitating a more proportionate and cost effective discovery process

This article is over 3 years old. More recent information on this subject may exist.

The exponential growth in the volumes and sources of electronic information adds complexities to the legal process that challenge traditional practices. As the document volumes increase (which is happening in every matter), then so too will the challenges and costs of managing the information.

Compounding these challenges are greater pressures from both the courts and clients to ensure the discovery process is proportionate and cost effective.

It has now become too expensive to “eye ball” every document, which may have been possible in a paper-based world. We simply need new practices to respond to these challenges to enable lawyers to get to the key documents quickly and cost effectively.

Options like predictive coding have emerged to enable a more proportionate and cost effective discovery process by prioritising what lawyers look at.

Challenging traditional practices

The eDiscovery landscape continues to evolve and discovery can no longer be managed proportionately or cost effectively solely relying on traditional practices. Managing electronic documents in the same way as paper will only lead to greater inefficiencies and increased costs. 

Much of the time and money spent on discovery exercises can be avoided.

Recent reports estimate that 70% of discovery costs are incurred at the document review stage. In a manual document review, many of the documents that lawyers come across are irrelevant for discovery purposes, which adds an unnecessary burden and cost to the review exercise.

Traditionally a document review would have involved a linear process of reviewing “document by document”. Aside from the increased cost, a linear process will not be as consistent or accurate as practices that fully leverage the use of technology. 

Solely human review is not always as effective as one may think. Large scale manual reviews involving numerous lawyers often produce inconsistent results as the review team may have varying degrees of knowledge about the subject matter. Many of the calls made by the initial reviewers frequently have to be corrected by senior lawyers later in the process – creating additional, yet avoidable, work and cost.

Problems with keyword searches

To reduce document volumes many lawyers have turned to using keyword searches to isolate what they consider is the most important information. 

The reality is that keyword searches are not effective in getting to the key documents, as they have the potential to miss things. Even after considerable time and effort testing and refining search terms, the results may produce documents that are not in the context intended. 

It is important to come up with new practices that reduce the volume of documents that require lawyers to review, and at the same time allow the legal team to get to the most important information at an earlier stage. It is becoming increasingly difficult to undertake discovery obligations without making more effective use of technology.

Predictive Coding – a proportionate approach

Predictive coding is developing as a proportionate option to improve the effectiveness of a document review. 

What is predictive coding?

Predictive coding is technology that analyses the decisions of a human review of a sample set of documents. The software then prioritises or ranks the remainder of documents based on the decisions made on the sample documents. The documents can then be reviewed by lawyers, usually identifying the key documents first. 

Predictive coding is also commonly known as Technology Assisted Review, Computer Assisted Review, Assisted Review or even Prioritisation Technology. No matter what the technology is called, they all essentially do the same thing – accelerating the document review by combining human expertise and technology.

How the predictive coding process works

The predictive coding process can differ slightly depending upon the choice of provider and software, although generally the process will be: 

a senior lawyer with detailed knowledge of the case will review a sample set of documents to “train” the system, identifying if documents are “relevant” or “not-relevant”;

the software then learns by the calls made by the lawyer using algorithms to recognise word meanings and associations between documents and their relevance. These characteristics are then applied across all of the remaining documents, to predict the likely relevance;

it is an iterative process that continues with a lawyer reviewing further documents to improve the accuracy. The software continues to learn with the more documents reviewed. This process continues until the legal team is satisfied with the confidence in the document categorisation being returned; the documents are then prioritised to enable the final review by the legal team.

Addressing lawyers’ reluctance about technology

For a long time there has been apprehension from the legal fraternity about the use of any technology that may take away from the importance of the role of the lawyer. There is always a typical reluctance for lawyers to move away from what they are accustomed to. 

Using a computer to select what lawyers are to look at has been happening for years through the use of keyword searches. Keywords were usually comfortable for lawyers, despite them frequently not appreciating their lack of effectiveness. 

Many may initially use predictive coding as a quality control tool to gain confidence in how it works against the calls made in a traditional review exercise.

The leading predictive coding products now allow the ability to audit the calls made. This transparency has helped provide greater confidence to lawyers about the use of predictive coding. Most studies justify the calls made and the improved accuracy over a traditional human review. Predictive coding will be more precise than a human review as humans get tired and humans make inconsistent mistakes. Computers do not ! 

The software is not a “silver bullet” and will not replace the role of the lawyer.

Reinforcing the importance of the role of the lawyer

Using predictive coding can maximise the expertise of the lawyer, especially the lawyer with expert knowledge about the matter who reviews the sample set of documents.

The choice of who is to conduct the initial review of the training set is crucial. It is important to use the expertise of a lawyer who is knowledgeable about the matter. If not then the results may be of a poor quality.

The software does not make the final calls on documents, but instead prioritises the information for lawyers to review. The legal team can then devote their energies into looking at the most important information, instead of trawling through large volumes of irrelevant material. By grouping similar documents together, the technology enables a more consistent and accurate document review. The irrelevant documents can quickly and easily be removed from the review set.

When to deploy predictive coding

The option of predictive coding software (or document prioritisation) is identified in the Discovery Checklist (in Schedule 9 of the rules) as one of the options that parties should consider on all matters.

Predictive coding will most commonly be used when there are large document volumes, although it can also be very effective on smaller cases especially when a firm does not have the internal resource to review information. 

The software can help provide smaller firms with a significant competitive advantage over firms that may persist with traditional manual practices. Firms may be able to run larger document cases, without having to acquire additional resource. 

Predictive coding is already being used in New Zealand law firms, especially by those that need to look at more innovative ways to compete with larger firms. 

Predictive coding alone will not reduce the cost of discovery nor will it be right for every case. It is still important to devise a suitable strategy to suit the requirements of the case. Predictive coding is just one of many tools available to help manage the challenges of the eDiscovery process. Like any comprehensive strategy, the approach should always involve sampling, testing and iteration of the process.

The choice of provider and software is crucial

It is important to not underestimate the value of using leading providers and software. It is imperative to align yourself with a provider who has considerable experience as well as proven predictive coding software. 

Devising an appropriate workflow and methodology is crucial, as it takes more time at the outset to set up your strategy. It is important that your strategy and software is able to adapt if new information emerges or the issues change.

Independent advice about the benefits and limitations of predictive coding providers and software can be invaluable to make an informed decision as to whether it is right for the requirements of your matter.


Today’s increasing data volumes and subsequent costs require us to look at changing traditional practices if discovery is to be managed proportionately and cost effectively. 

Predictive coding combines the expert knowledge of lawyers with technology to help facilitate a more proportionate approach to managing the discovery process.

By prioritising what lawyers look at, tools like predictive coding are options where technology can be applied to assist lawyers get to the key documents quicker, cheaper and more accurately than traditional practices.

Andrew King is a litigation support consultant at E-Discovery Consulting (, where he advises on strategies and tools to simplify the discovery process, including providing independent eDiscovery software advice. Andrew can be contacted on 027 247 2011 or

Lawyer Listing for Bots