*1 Plaintiffs and defendant Google LLC (“Google”) ask the Court to resolve a dispute concerning plaintiffs' requests for documents about Google's selection of data for distribution to third-party participants in RTB auctions. Dkt. No. 336. The Court held a hearing on this dispute on November 1, 2022. Dkt. No. 349.

For the reasons explained below, the Court will require Google to produce documents and information responsive to the following requests: RFPs 91-94, 97, and 101 (to the extent it has not already done so); RFP 99 (as modified); RFP 100 (as agreed); and RFP 102 (by written answer only). Google need not produce documents responsive to RFPs 95 and 96.

I. BACKGROUND

The parties' dispute concerns plaintiffs' RFPs 91-97 and 99-102. Dkt. No. 336 at 1; Dkt. No. 336-1. Plaintiffs explain that Google uses an “automated data selection process” to select data for distribution to RTB participants in an RTB auction.[1] Plaintiffs ask for “documents sufficient to show” (1) the automated data selection process (RFPs 91, 93), (2) each step taken in the automated data selection process (RFP 92), (3) the data inputs into the automated data selection process and their sequence (RFP 94), (4) the data sources from which the automated data selection process selects data (RFP 95), (5) the architecture of the software program that comprises the automated data selection process (RFP 96), and (6) the criteria used by the automated data selection process to select data (RFP 97). In addition, plaintiffs ask for “all documents and communications” concerning any restrictions used to exclude data from the set of data distributed to RTB participants (RFP 99), “documents sufficient to show” how user opt-out signals and user consent signals impact the automated data selection process (RFP 100), and “all developer documents” provided to Google RTB participants that explain the structure of a bid request and/or how to interpret the data provided in a bid request (RFP 101). Finally, plaintiffs ask for “documents sufficient to identify” the Google employees primarily responsible for designing the automated data selection process (RFP 102).

Google's sole objection to most of these requests is that they are duplicative of document requests plaintiffs have already served and for which Google has already searched for and produced responsive documents. Google argues that it should not be required to revisit these requests or undertake any further efforts to search for responsive documents. For other requests, Google objects that the Court has already determined that the discovery plaintiffs seek is not relevant, and argues that it should not be required to relitigate these issues. See Dkt. No. 336 at 3.

II. DISCUSSSION

*2 For purposes of this order, the Court groups the requests according to Google's objections.

A. RFPs 91-94, 97, and 101

Google argues that RFPs 91-94, 97, and 101 are entirely duplicative of the documents and information Google agreed to produce in response to plaintiffs' RFP 22. As documented in a prior discovery order, Google agreed to produce documents sufficient to show the information described in the following subparts of RFP 22:

See Dkt. No. 184 at 6 (noting that Google agreed to produce documents sufficient to show the information described in all subparts of RFP 22, except for subparts (i), (j), and (q)(i)); see also Dkt. No. 141 at 3, 6; Dkt. No. 141-1. In addition, the Court ordered Google to produce documents sufficient to show the account holder information shared with third parties through the RTB process if Google maintains a record of such information. See Dkt. No. 184 at 6-7. In that prior order, the Court concluded that Google need not produce documents showing how Google collects or stores account holder information. See id. at 7.

*3 The parties appear to agree that RFPs 91-94, 97, and 101 are encompassed within the scope of production Google was required to make with respect to previously served RFP 22, even though these pending requests are more focused. They disagree about whether Google actually has produced documents sufficient to show the information specifically called for in RFPs 91-94, 97, and 101. Google says that it has already produced responsive non-privileged documents sufficient to show the categories of information encompassed by these RFPs. Dkt. No. 336 at 5. Plaintiffs say that Google has not previously produced documents responsive to these specific RFPs, and in particular, that Google has not produced documents showing in sufficient detail how the automated data selection process operates. Id. at 1.

At the hearing, the Court pressed Google regarding whether its prior production of documents relied solely on the application of search terms to ESI, or whether it had undertaken to identify responsive documents by investigating custodial and non-custodial sources of information, such as asking the Google employees with relevant knowledge to identify responsive documents for production. Google represented to the Court that its search for responsive documents had indeed included both kinds of searches—i.e. ESI searches and targeted, subject matter-specific searches. See Dkt. No. 349.

As Google apparently does not object to the scope of RFPs 91-94, 97, and 101, but objects only that the requests are duplicative, the Court orders Google to produce all non-privileged documents responsive to these requests by November 18, 2022. The Court emphasizes that this is not the kind of production that requires application of search terms to ESI; rather, it requires a deliberate and focused investigation of the documentation Google maintains about the operation of its automated data selection process. If Google has already undertaken this investigation and produced the responsive documents, it will have nothing more to do to comply with order.

During the hearing, Google suggested that it does not necessarily have documents that show all of the details of the automated data selection process that plaintiffs say they require. In that case, plaintiffs may of course use other means to obtain the discovery they need, including deposing any witnesses whose testimony may be necessary to provide a more complete understanding of the process or to identify relevant sources of information about the process. If this deposition testimony is important for class certification briefing, the Court expects the parties to cooperate in promptly scheduling such depositions.

B. RFP 95

RFP 95 asks for documents sufficient to show “the various data sources within Google” from which data is selected for distribution by the automated data selection process. Plaintiffs say that “[i]nformation about the specific data sources accessed is an integral and necessary piece of the RTB puzzle; without it, it is not possible to understand how Google RTB selects the Account Holder information it distributes to third parties.” Dkt. No. 336 at 2. Google responds that plaintiffs merely seek discovery of Google's collection and storage of account holder information, subjects this Court has previously concluded are not relevant to plaintiffs' claims. Id. at 5.

It is not clear where the dispute lies. The Court has already ordered that plaintiffs may obtain discovery (e.g., via RFP 22 and RFPs 91-94, 97, and 101) regarding how the automated data selection process operates, including what account holder data is distributed to third-party RTB participants and how that data is selected. The Court also has already ordered that plaintiffs may not obtain discovery of how Google generally collects or stores account holder information. Plaintiffs do not explain what additional information they believe RFP 95 encompasses beyond what the Court has already ordered, or how that additional information is relevant to any claim or defense in the case. To the extent they seek discovery about the data Google collects and stores, they do not explain why the Court should revisit its prior order on this point.

*4 Accordingly, the Court denies plaintiffs' request for an order compelling Google to produce documents responsive to RFP 95.

C. RFP 96

RFP 96 asks for documents sufficient to show “the architecture of the software program(s)” that comprise the automated data selection process. Google says that this is highly sensitive information and that production of such detailed technical information is unnecessary for plaintiffs to understand how data is distributed through the RTB auction. Dkt. No. 336 at 5. Plaintiffs argue that Google has not shown that the architecture of the software underlying the data selection process is sensitive or trade secret, but even if it is, the protective order affords adequate protection. Id. at 2.

The Court is skeptical that discovery of the architecture-level details of Google's software is relevant and proportional to the needs of the case, particularly in view of Google's representation at the hearing that it has no objection to producing (and did not withhold from its prior production) internal design documents that reveal how the automated data selection process operates. Dkt. No. 349. Absent a more specific showing of need for information about the architecture of Google's software, the Court agrees that production of design documents, including schematics, showing how the automated data selection process operates should be sufficient.

Accordingly, the Court denies plaintiffs' request for an order compelling Google to produce documents responsive to RFP 96.

D. RFP 99

RFP 98 and RFP 99 both concern Google's use of filters or restrictions to exclude data from distribution in an RTB auction. Google has agreed to produce “documents sufficient to show the filters or restrictions that Google actually uses to exclude user-specific data from bid request to third parties via RTB, if any” in response to RFP 98, which is not in dispute. Dkt. No. 336-1 at 15-16. As revised, RFP 99 asks for “all documents and communications concerning any restrictions used to exclude data from the set of data distributed to Google RTB participants in a bid request.” Dkt. No. 336 at 2-3 & n.1. RFP 99 is not limited to “documents sufficient to show” but would require Google to search for all responsive documents, including emails, concerning restrictions used to exclude data. Google objects that such a request is overbroad and unduly burdensome; it proposes instead to produce documents sufficient to show “the filters or restrictions that Google actually uses to exclude user-specific data from bid requests to third parties via RTB”—i.e. the same scope of production it has committed to make in response to RFP 98. Id. at 5.

At the hearing, plaintiffs explained that they have two principal concerns. First, they want Google to produce documents showing not merely what the filters and restrictions are called, but what they do and how they work. Second, they want to understand whether and how the filters or restrictions have changed over the time period relevant to the putative class. These are reasonable concerns. However, the Court agrees with Google that the scope of RFP 99 is substantially broader than this.

*5 The Court resolves this dispute as follows: Google must produce documents sufficient to show: the filters or restrictions that Google uses to exclude user-specific data from bid requests to third parties via RTB; how they work; and how they have changed, if at all, over the time period relevant to the putative class. Google must produce responsive documents by November 18, 2022.

E. RFP 100

RFP 100 asks for documents sufficient to show “how user opt-out signals and/or user consent signals impact” the automated data selection process. As discussed at the hearing, the parties appear to agree on this request. See Dkt. No. 336 at 6. Accordingly, the Court orders Google to produce documents sufficient to show how the data shared with third party bid request recipients via RTB is impacted by user opt-out signals and/or user consent signals, including the following seven specific signals: Do Not Track (DNT), Global Privacy Control (GPC), the US Privacy String (defined by the IAB for CCPA, etc. compliance), NIA opt-out cookies (Network Advertising Initiative), IAB Trust and Consent Framework (TCF) signals (used in Europe for GDPR compliance), Android's advertising ID opt-out, and Google's personalized advertising opt-out. Google must produce responsive documents by November 18, 2022.

F. RFP 102

RFP 102 asks for documents sufficient to identify the Google employees who are or were primarily responsible for designing the automated data selection process. Google objects that this request is duplicative of plaintiffs' RFP 19 and Interrogatory No. 4, and refuses to produce responsive documents on that basis. Dkt. No. 336 at 6. As a compromise, plaintiffs propose that Google provide the information in the form of a written answer, rather than documents. Id. at 3. Google does not agree. Id. at 6.

While plaintiffs should have perhaps sought this discovery initially by means of an interrogatory instead of a document request, the dispute is easily resolved by adopting the solution plaintiffs propose. Google shall provide a written response to RFP 102 that identifies the employees who are or were primarily responsible for designing the automated data selection process. If, as Google says, this information may be found in its earlier response to Interrogatory No. 4 or in the documents responsive to RFP 19, it should be relatively easy for Google to provide the information plaintiffs request. Google must provide this written response by November 18, 2022.

III. CONCLUSION

In sum, by November 18, 2022, Google must produce documents and information responsive to RFPs 91-94, 97, and 101 (to the extent it has not already done so); RFP 99 (as modified); RFP 100 (as agreed); and RFP 102 (by written answer only). Google need not produce documents responsive to RFPs 95 and 96.

IT IS SO ORDERED.

In many of their requests, plaintiffs refer to this automated data selection process as the “Decision-Making Process,” which plaintiffs define as “any system, including, but not limited to, any software, program, application, algorithm, or artificial intelligence (or any collection of the same) that operates to programmatically select data that is distributed in a Google RTB auction.” See Dkt. No. 366 at 1.