The High Court has ordered disclosure to be carried out using predictive coding despite a party's objections to the use of the technology: Brown v BCA Trading Limited [2016] EWHC 1464 (Ch).

This is understood to be the first reported decision in which the English courts have approved the use of predictive coding following a contested application. A decision in February this year had endorsed the use of the technology for a large disclosure exercise, but in circumstances where the parties had already agreed on its use (Pyrrho Investments v MWB Property [2016] EWHC 256 (Ch) – see our blog post on that decision which also explains more about predictive coding).

The present decision illustrates that the comparative costs between predictive coding and keyword searching are likely to be a significant factor in determining whether it should be ordered. It also suggests that, where there is a dispute as to whether predictive coding is appropriate, the court may take particular account of the views of the party who holds the lion's share of the documents and therefore will have the greater burden on disclosure.

Judgment was given in May this year and the case was reported in the legal press but a transcript has only become available more recently (on Westlaw).


The underlying case is an "unfair prejudice" petition under section 994 of the Companies Act 2006. It was agreed that disclosure should be on a standard basis but there was a dispute as to whether the exercise should be conducted using predictive coding or a more traditional keyword approach. The majority of the relevant documents were in the hands of the respondents, who sought the order for predictive coding.


The court (Mr Registrar Jones in the Companies Court) made an order for predictive coding. The fact that most documents were in the hands of the respondents was not determinative, but it was relevant to take into account when considering their assertion that predictive coding would be the most reasonable and proportionate method of disclosure.

The registrar described it as "extremely significant" that the estimated costs for predictive coding were in the region of £132,000 compared to costs for a keyword search of at least £250,000 and possibly as much as £338,000 on a worst case scenario.

That was only persuasive to the extent that predictive coding would be effective and achieve the disclosure required. However, there was nothing to suggest that predictive coding would not be able to identify the documents that would otherwise be identified through, for example, keyword searches – and predictive coding would be considerably cheaper.

The registrar referred to Pyrrho and the factors considered by Master Matthews in that case to favour predictive coding. The registrar said that all of those factors effectively applied to the present case, save that one factor (that there was nothing in the CPR or practice directions to prohibit the use of such software) was neutral and that (unlike in Pyrrho) the parties had not agreed to the use of the software. The other factors from Pyrrho, which the registrar said applied here, include:

  • Experience in other jurisdictions has been that predictive coding software can be useful in appropriate cases.
  • There is no evidence to show that it leads to less accurate disclosure being given than manual review / keyword searches, and some evidence to the contrary.
  • There were no factors of any weight pointing against its use.

The registrar also made directions aimed at narrowing down the issues and the scope of disclosure before the predictive coding exercise took place. Essentially, the parties would have to identify relevant issues, the documents relating to them, their source and location, following which there would be discussions between the parties regarding the criteria to adopt and the general process of disclosure.