Nichols v. Noom Inc.

Background

This putative class action concerns allegations that Noom Inc. used an automatic-renewal feature to trick consumers into signing up for its weight-loss program and made cancellation difficult. The discovery dispute involved GroupsMagic, Noom’s repository of coach chats. Each user-coach chat has a unique access code, and the repository contains 9.8 million access codes. Plaintiffs sought the complete chat-log histories for 2,500 randomly selected users so they could estimate how often users complained about automatic enrollment and cancellation difficulties and project those findings onto the larger population.

Noom proposed limiting the sampling population to users of its Healthy Weight program, applying search terms before sampling, and producing only communications within 48 hours of a search-term hit. Noom argued that plaintiffs’ proposed production could include large amounts of irrelevant and highly personal information, including health-related and identifying information, and would impose a substantial review and redaction burden.

Court’s analysis

The court agreed that the sample should be drawn only from Healthy Weight users because those users formed the relevant population for the proposed class. Noom was required to provide the total number of unique access codes in that population. The court did not impose a start-date restriction if Noom could segregate the Healthy Weight users.

The court rejected Noom’s proposal to apply search terms before generating the random sample. Because coach chats could contain relevant information about cancellation, plaintiffs could use the chats to count complaints about automatic renewal or difficulty canceling. The court stated that Noom could later challenge the legal validity or admissibility of plaintiffs’ statistical analysis, but could not restrict plaintiffs’ use of the produced information on that basis.

The court also required the initial random sample of 2,500 chats to be generated before further filtering. After that sample was selected, Noom could use search terms and its proposed content-analysis process—independent reviews followed by reconciliation of disagreements—to identify the users whose chats contained relevant complaints. Noom did not have to produce the chats of users in the non-complaining subset because those chats did not contain relevant communications for purposes of the discovery request.

The court rejected Noom’s proposed 48-hour production window around search-term hits. Noom had not explained why 48 hours was appropriate, and relevant discussions could occur over weeks or involve the broader user experience. The court therefore declined to impose that limitation or another similar limitation on the complaining subset.

The court also noted that Noom could rely on the existing protective order instead of redacting all personally identifying and medical-related information from the limited production. Noom could additionally designate the chats as attorneys’-eyes only. The court reminded both parties that users’ identifying information could not be publicly filed without redaction.

Disposition

The court concluded that plaintiffs’ motion, ECF No. 241, was granted in part and denied in part. This was a discovery ruling about the method and scope of sampling and producing GroupsMagic coach chats; it did not decide the underlying claims about Noom’s enrollment or cancellation practices.