Technology Assisted Review (TAR) in discovery: Where things stand and what’s ahead
Discovery has always been one of the more resource-intensive parts of litigation. Fortunately, tools like Technology Assisted Review (TAR) are helping lawyers tackle the burden more efficiently, particularly when facing large volumes of documents. While TAR isn’t new, its evolving capabilities and the court's apparent support of its use make it worth a closer look.
What is TAR?
TAR uses machine learning to assist with document review in discovery. Lawyers begin by tagging a sample set of documents as relevant or irrelevant, and the software uses that input to learn what to look for across the entire document set.
Older TAR systems relied on a one-time training phase. Once the machine was “taught,” it would apply that understanding to the rest of the documents. Newer TAR systems now offer continuous active learning, meaning the software keeps learning throughout the process. As more documents are reviewed and decisions are made by the legal team, the software refines its model, ideally resulting in a faster, more accurate review.
Where do New Zealand courts stand?
New Zealand courts haven’t yet set detailed guidelines around the use of TAR, but what has been said suggests that its use is acceptable, provided both parties agree.
The High Court Rules 2016 require parties to cooperate in managing discovery, including using technology effectively and efficiently. Applying this to TAR, the message is clear: its use should be agreed upon by the parties and its application must be properly documented.
Parties using TAR must also include an explanation in their discovery affidavit outlining how the technology was trained, and how it was used.
In White v James Hardie, one party used a continuous learning model and stopped the review once the TAR software reached a 90% confidence level that all relevant documents had been identified. While the Court didn’t set this as a formal benchmark, it may serve as a useful indicator of what’s considered acceptable.
In Minister of Education v James Hardie, the court noted that TAR allowed reviewers to focus on likely responsive documents, improving efficiency and avoiding unnecessary review of irrelevant material. This indicates that courts are open to its use, particularly when it can demonstrably save time and cost.
What about other jurisdictions?
TAR has received more detailed treatment in jurisdictions like the UK, Ireland and Australia with generally favourable views from the courts.
In Pyrrho Investments v MWB Property, a UK Court endorsed predictive coding (another term for TAR) for several reasons:
There was no evidence to suggest TAR is less accurate than manual review.
A full manual review would have been disproportionately expensive.
Using TAR promotes greater consistency than a large team of reviewers applying relevance criteria independently.
The Court also cited Irish Bank Resolution Corporation v Quinn, where the Irish High Court concluded that TAR is at least as accurate, and likely more accurate, than manual review, particularly in large-scale discovery. That decision referenced studies showing TAR achieved substantially better recall and precision rates than manual review.
In McConnell Dowell Constructors v Santam, the Supreme Court of Victoria (in Australia) faced a discovery exercise involving millions of documents. The court approved TAR’s use, noting its ability to significantly reduce the size of the review set. The decision also referred to Pyrrho and Irish Bank, adding weight to the view that TAR is not only acceptable but increasingly expected in complex, high-value litigation.
Both McConnell Dowell and Pyrrho touched on the principle of proportionality, reinforcing that the method and cost of document review should be appropriate to the value and complexity of the case.
Another consistent theme across jurisdictions is the benefit of agreement between the parties. Mutual consent helps ensure transparency and gives parties greater comfort that any failure to locate a particular document is more likely seen as an honest oversight rather than deliberate withholding.
Looking ahead in New Zealand
TAR is likely to become more common in New Zealand as the volume and complexity of document review continue to grow. While courts here haven’t yet developed detailed guidance, they appear to be open to its use, particularly when it leads to fairer, faster and more proportionate outcomes.
We anticipate that courts will:
Expect parties to agree on the use of TAR at the outset.
Require sufficient transparency around how the software is used.
Continue to weigh proportionality in assessing whether TAR is appropriate.
In practical terms, if TAR is used, it will be important to keep clear records of how it was trained, how it performed and the extent of legal oversight. This information should be included in the discovery affidavit to demonstrate compliance and build confidence in the process.
In short, TAR is not a shortcut; it’s a smarter way to manage large-scale discovery. When used correctly, it can reduce cost, improve consistency and let lawyers focus their time where it’s most valuable.
Contributed by:
Joey James and Harriet Lord
