For this matter, the central feature of the 200,000-plus documents was that a very low percentage of them were relevant to the matter. Using linear review software, the review process could take months and cost hundreds of thousands of dollars, just to produce a very small percentage of the 200,000-plus documents.

Baker McKenzie realized that the data may be better suited for Technology Assisted Review (TAR).


In particular, the legal team wanted to try a particular form of TAR called Continuous Active Learning (CAL), in which the software continually refines its results based upon iterative feedback from reviewers.

As long-time users of FTI Technology’s Ringtail software, the firm turned to FTI Technology about using Ringtail’s CAL features on this matter. Together, Baker McKenzie and FTI Technology developed a cost-effective and defensible process:

  • Collecting a random sample of 1,000 documents from Ringtail’s off-the-shelf sampling feature.
  • Coding of those documents for responsiveness, privilege, confidentiality and issues by a Baker McKenzie Senior Associate.
  • Responsive documents were checked for coding accuracy by Baker McKenzie Senior Associates.
  • FTI Technology’s team leveraged a dashboard in Ringtail to track the percentage of responsive documents and calculate the range of likely responsive documents across the entire dataset (within a 95% confidence interval).
  • Ringtail’s CAL functionality then took the reviewers’ coding decisions and found 3,000 other documents in the data set that were scored most likely to be responsive.
  • Overnight, an offshore review team coded the 3,000 documents for responsiveness, privilege, confidentiality and issues.
  • Responsive materials were sent to Junior and Senior Associates for second-level review.
  • Ringtail would serve up the next 3,000 highest scored documents after evaluating the coding in the initial 1,000 document review and the previous 3,000 document set.
  • The offshore review team reviewed the new document set overnight and had their responsive calls doublechecked by Baker McKenzie Associates. The new responsive documents would help to further refine the CAL model and find another 3,000 potentially-responsive documents.
  • As this workflow continued through 30 sets of documents, the FTI Technology team monitored the level of responsive documents in each review. As the review progressed, increasingly smaller batches of the most responsive documents were drawn.
  • In addition, random sampling was conducted to ensure continued accuracy, refresh the model, confirm the level of responsive documents found and to track achieved recall level to allow decision makers to end the process.

Impact: The matter, likely the first in Australia using CAL, was a tremendous success.

Some of the benefits included:

Faster case strategy development

The Baker McKenzie team was able to develop case strategy faster because the most important documents were sent to senior lawyers quickly.

Significant cost savings:

Through the use of CAL to find the responsive documents quickly, Baker McKenzie used fewer review attorneys and estimates that they saved their client between AU $500,000 and $800,000 in e-discovery costs.

Defensible results:

The expert FTI Technology team documented every step in the process and could provide expert witness testimony on the defensibility of the process and results.