Singer et al v. Facebook, Inc.

Filing 388

ORDER RE MOTION TO CERTIFY CLASS AND DAUBERT MOTIONS. Signed by Judge James Donato on 3/29/2022. (jdlc3, COURT STAFF) (Filed on 3/29/2022)

1
2
3
4

UNITED STATES DISTRICT COURT

5

NORTHERN DISTRICT OF CALIFORNIA

6
DZ RESERVE, et al.,

7

Plaintiffs,

8

ORDER RE MOTION TO CERTIFY
CLASS AND DAUBERT MOTIONS

v.

9

META PLATFORMS, INC.,

10

Re: Dkt. Nos. 282, 285, 286

Defendant.

11

United States District Court
Northern District of California

Case No. 3:18-cv-04978-JD

12
13

In this action alleging fraud against Meta Platforms, Inc. (Meta), formerly known as

14

Facebook, named plaintiffs DZ Reserve and Cain Maxwell have asked to certify a class of United

15

States residents who paid Meta for placement of advertisements on social media platforms. Dkt.

16

No. 282. The gravamen of the lawsuit is that Meta inflated its potential advertising reach to

17

consumers, and charged artificially high premiums for ad placements. Meta opposes certification,

18

and filed two Daubert motions challenging the opinions and conclusions proffered by plaintiffs’

19

expert witnesses. Dkt. Nos. 285, 286.
Three claims alleged in the Third Amended Complaint (TAC) remain in play. Dkt. No.

20
21

332.1 The Court dismissed with prejudice plaintiffs’ claims for breach of the implied covenant of

22

good faith and fair dealing and a quasi-contract claim. Dkt. No. 255 at 2. The Court sustained

23

plaintiffs’ claims for fraudulent misrepresentation and fraudulent concealment, with the proviso

24

that plaintiffs could not pursue those claims for conduct before August 15, 2015. Id. at 1-2.

25

While the certification motion was pending, the Court granted a motion for judgment on the

26
27
28

1

The TAC was originally filed under seal as Dkt. No. 166. The Court denied the administrative
motion to seal the TAC without prejudice, see Dkt. No. 320, and the TAC was refiled as Dkt. No.
332.

1

pleadings and dismissed plaintiffs’ claim of restitution under the California Unfair Competition

2

Law (UCL). Dkt. No. 366. The UCL claim was sustained for injunctive relief only. Id. at 2.

3

Consequently, the claims subject to certification are fraudulent misrepresentation and fraudulent

4

concealment for damages, and the UCL for injunctive relief.
DISCUSSION

5
6
7

I.

BACKRGOUND
Before getting into the merits, a few words about Meta’s brief are in order. Meta fired a

blunderbuss of objections at certification. Virtually every page of its lengthy opposition brief

9

presented a new argument, often in just a paragraph or two of discussion. As a result, many of its

10

arguments were underdeveloped to the point where the Court had ample justification to disregard

11

United States District Court
Northern District of California

8

them. Even so, the Court undertook the burden of sorting through Meta’s brief to identify and

12

address what appear to be its main arguments. Meta aggravated this situation further by making

13

factual arguments much more suited to summary judgment proceedings than a class certification

14

motion. To be sure, as the ensuing certification standards make clear, the Court will review the

15

evidence as pertinent to the question of whether a class should certified. Meta’s arguments went

16

far beyond that inquiry.

17

The parties’ familiarity with the record is assumed. In pertinent part, the undisputed facts

18

are that Meta sells advertising to businesses and business owners like plaintiffs DZ Reserve and

19

Cain Maxwell. Dkt. No. 332 at ¶ 2 Meta’s Ads Manager platform is used by advertisers to

20

identify their advertising targets, including the demographic reach they desire. Id. at ¶ 3. After

21

advertisers select their targeting and placement criteria, the Ads Manager displays a “Potential

22

Reach” for the advertisement. See Dkt. No. 282-3. The Potential Reach is expressed as a number

23

of people that the ad may reach. Id. The default Potential Reach number, before any targeting

24

criteria are selected, is the Potential Reach for people in the United States aged 18 and up, which

25

was shown during the putative class period to be over 200 million people. Dkt. No. 281-9 at

26

¶¶ 55-60. As targeting criteria are selected, the Potential Reach is revised accordingly. Dkt. No.

27

282-3; 281-13 at 54:21-59:25. Meta describes the Potential Reach as an estimate of people in the

28

ad’s target audience. See Dkt. No. 296-17 at 3.
2

1

II.

CLASS CERFITICATION STANDARDS
Plaintiffs propose to certify this class under Federal Rules of Civil Procedure 23(a),

2
23(b)(2), and 23(b)(3):
3
4
5
6
7
8
9
10

United States District Court
Northern District of California

11
12

All United States residents (including natural persons and
incorporated entities) who, from August 15, 2014, to the present
(“Class Period”), paid for the placement of at least one advertisement
on Facebook’s platforms, including the Facebook and Instagram
platforms, which was purchased through Facebook’s Ads Manager or
Power Editor.
Excluded from the class are: (1) advertisements purchased pursuant
to agreements other than Facebook’s Terms of Service or Statement
of Rights and Responsibilities; (2) advertisements purchased using
only non-lookalike Custom Audiences as the targeting criteria; (3)
advertisements purchased using Reach and Frequency buying; (4)
advertisements purchased with the objectives of canvas app
engagement, canvas app installs, offer claims, event responses, page
likes, or external; and (5) advertisements for which Facebook
provided Potential Reach lower than 1000.
Dkt. No. 282 at 15.

13
The Court has written extensively on the standards for class certification, which informs
14
the discussion here. See, e.g., Sapan v. Yelp, Inc., No. 18-cv-3240-JD, 2021 WL 5302908 (N.D.
15
Cal. Nov. 15, 2021); Meek v. SkyWest, Inc., --- F. Supp. 3d ---, 2021 WL 4461180 (N.D. Cal. Sep.
16
17
18
19

29, 2021). A class action is “an exception to the usual rule that litigation is conducted by and on
behalf of the individual named parties only.” Comcast Corp. v. Behrend, 569 U.S. 27, 33 (2013)
(quotations omitted). The overall goal is “to select the metho[d] best suited to adjudication of the
controversy fairly and efficiently.” Amgen Inc. v. Connecticut Ret. Plans & Trust Funds, 568 U.S.

20
455, 460 (2013) (internal quotations omitted) (modification in original). Plaintiffs must show that
21
their proposed class satisfies all four requirements of Rule 23(a), and at least one of the
22
subsections of Rule 23(b). Comcast, 569 U.S. at 33 (2013); Zinser v. Accufix Research Inst., Inc.,
23
253 F.3d 1180, 1186 (9th Cir. 2001), amended by 273 F.3d 1266 (9th Cir. 2001). As the parties
24
seeking certification, plaintiffs bear the burden of showing that the requirements of Rule 23 are
25
met for their proposed class. Mazza v. Am. Honda Motor Co., 666 F.3d 581, 588 (9th Cir. 2012).
26
27

The Court’s class certification analysis “must be rigorous and may entail some overlap
with the merits of the plaintiff’s underlying claim,” but merits questions may be considered only

28
3

1

to the extent that they are “relevant to determining whether the Rule 23 prerequisites for class

2

certification are satisfied.” Amgen, 568 U.S. at 465-66 (internal quotations and citations omitted).

3

The class certification procedure is decidedly not an alternative form of summary judgment or an

4

occasion to hold a mini-trial on the merits. Alcantar v. Hobart Service, 800 F.3d 1047, 1053 (9th

5

Cir. 2015). The decision of whether to certify a class is entrusted to the sound discretion of the

6

district court. Zinser, 253 F.3d at 1186.

7

III.

8
9

RULE 23(B)(3) CLASS
The Rule 23(a) factors are the same for certification of the proposed class under Rule

23(b)(2) or (b)(3), and the conclusions reached here for the Rule 23(a) elements apply to both
types of classes. The main difference is the predominance element of Rule 23(b)(3), which Rule

11

United States District Court
Northern District of California

10

23(b)(2) does not require. The Court takes up the proposed Rule 23(b)(3) class first.

12

The Court granted Meta’s motion for judgment on the pleadings to dismiss plaintiffs’ UCL

13

claims for restitution, see Dkt. No. 366, so monetary relief is only available for plaintiffs’ common

14

law fraudulent concealment and fraudulent misrepresentation claims.

15

A.

16

Rule 23(a)(1) requires that a proposed class be “so numerous that joinder of all members is

Numerosity (23(a)(1))

17

impracticable.” Fed. R. Civ. P. 23(a)(1). Plaintiffs state, with evidentiary support, that “[d]uring

18

each year of the class period, more than 2 million United States advertisers purchased Facebook

19

ads.” Dkt. No. 282 at 15. Meta does not contest numerosity, and the Court finds this element is

20

satisfied.

21

B.

22

Rule 23(a) requires the named plaintiffs to demonstrate that their claims are typical of the

Typicality and Adequacy (23(a)(3)-(4))

23

putative class, and that they are capable of fairly and adequately protecting the interests of the

24

class. Fed. R. Civ. P. 23(a)(3)-(4). The named plaintiffs say typicality is satisfied because they

25

“bring the same legal claims as the rest of the putative [c]lass” and “rely on the same grounds for

26

liability as the rest of the class.” Dkt. No. 282 at 17. Plaintiffs also say that they are adequate

27

representatives because “[t]hey have no conflicts with the class,” have “participated actively in

28
4

1

this case,” and their counsel has no conflicts, has experience with class actions, and has

2

demonstrated a “willingness to vigorously prosecute this action.” Id.
Meta makes multiple objections to adequacy and typicality. The primary one is that the

3
4

proposed class is said to include a diverse population of advertisers ranging from “‘large

5

sophisticated corporations’ to ‘individuals and small businesses.’” Dkt. No. 294 at 16-17. In

6

Meta’s view, this means that the putative class members are necessarily in such disparate positions

7

vis-à-vis its advertising services that the named plaintiffs, as advertisers on the smaller end of the

8

spectrum, cannot fairly or adequately represent them. Id.
The objection is not well taken. To start, typicality is demonstrated when “the claims or

9

defenses of the representative parties are typical of the claims or defenses of the class.” Hanlon v.

11

United States District Court
Northern District of California

10

Chrysler Corp., 150 F.3d 1011, 1019 (9th Cir. 1998), overruled on other grounds by Wal-Mart

12

Stores, Inc. v. Dukes, 564 U.S. 338 (2011). “The test of typicality is whether other members have

13

the same or similar injury, whether the action is based on conduct which is not unique to the

14

named plaintiffs, and whether other class members have been injured by the same course of

15

conduct.” (internal quotation marks omitted). Wolin v. Jaguar Land Rover N. Am., LLC, 617 F.3d

16

1168, 1175 (9th Cir. 2010). “Under the rule’s permissive standards, representative claims are

17

‘typical’ if they are reasonably co-extensive with those of absent class members; they need not be

18

substantially identical.” Hanlon, 150 F.3d at 1019.
That is the situation here. Plaintiffs have adduced evidence indicating that, regardless of

19
20

size or buying power, Meta’s customers saw similar representations by Meta about its advertising

21

reach and programs. Advertisers were shown the same default Potential Reach of over 200

22

million people before they applied any targeting criteria. Dkt. No. 281-9 at ¶¶ 55-60. Plaintiffs’

23

expert, Dr. Charles Cowan, states that even with different targeting criteria for each advertiser,

24

inflated Potential Reach representations were made across Meta’s platform. Dkt. No. 281-11 at

25

¶ 33. All advertising customers were shown Potential Reach estimates that were inflated by a

26

similar percentage. Id. at ¶ 15.2

27
28

2

Dr. Cowan’s work is discussed in more detail later in the order.
5

1

It may be that class members differ in advertising budgets and scope of purchases, as Meta

2

suggests, but Meta has not shown that these differences defeat typicality or the named plaintiffs’

3

ability to adequately represent all class members. This is not a case where the record demonstrates

4

that the products, pricing, and programs accessed by class members were so dissimilar that

5

typicality and adequacy could not be established. See, e.g., In re Graphics Processing Units

6

Antitrust Litig., 253 F.R.D. 478, 489-90 (N.D. Cal. 2008) (denying certification of antitrust class

7

where evidence demonstrated putative class members purchased entirely different products at

8

different prices). In effect, Meta simply posits that typicality and adequacy cannot be established

9

because the class includes large and small ad purchasers. The problem with this approach is that it

10

is ipse dixit and not an evidence-based objection.

United States District Court
Northern District of California

11

Meta’s case citations do not lead to a different conclusion. It overreads In re Facebook,

12

Inc., PPC Advertising Litig., 282 F.R.D. 446 (N.D. Cal. 2012), aff’d sub nom. Fox Test Prep v.

13

Facebook, Inc., 588 F. App’x 733 (9th Cir. 2014), to stand for the proposition that a “‘diverse

14

group’ of advertisers” necessarily undercuts adequacy and typicality. Dkt. No. 293-4 at 16-17.

15

But that case in fact determined that typicality had been demonstrated. In re Facebook, Inc.,, 282

16

F.R.D. at 453-54. Adequacy was not found because the record failed to show that the named

17

plaintiffs had suffered a concrete injury from the challenged conduct. Id. at 454. That is not a

18

circumstance present here.

19

Meta also has not demonstrated an evidence-based reason to reject the adequacy of the

20

named plaintiffs generally. Adequacy of representation asks whether: “(1) the representative

21

plaintiffs and their counsel have any conflicts of interest with other class members, and (2) will the

22

representative plaintiffs and their counsel prosecute the action vigorously on behalf of the class?”

23

Staton v. Boeing Co., 327 F.3d 938, 957 (9th Cir. 2003). Meta did not make a serious effort at

24

answering either inquiry in the negative, and plaintiffs have demonstrated that no such concerns

25

are in play here. See Dkt. No. 282 at 16-17.

26

Meta’s effort to recast its typicality and adequacy challenges as questions of reliance and

27

UCL standing is equally unavailing. See Dkt. No. 294 at 15. To start, named plaintiffs

28

demonstrated reliance by proffering evidence that DZ Reserve was deterred from using Meta ads
6

1

after learning that the Potential Reach was an inaccurate metric. Dkt. No. 293-27 at 193:17-194:5.

2

Similarly, named plaintiff Maxwell relied on Potential Reach to set his budgets and would not

3

have spent money on Meta ads if he knew Potential Reach was inaccurate. See Dkt. No. 293-29 at

4

199:8-12; Dkt. No. 317-2 at 257:3-14. Meta says that the named plaintiffs would still have

5

purchased ads if they knew the Potential Reach was inaccurate. Dkt. No. 294 at 16. But plaintiffs

6

also indicated that they would have spent less on ads after learning the Potential Reach was

7

inaccurate, demonstrating that they were deceived into spending more money. See, e.g., Dkt. No.

8

317-3 at 105:21-106:5. This and similar evidence also establishes reliance for UCL standing

9

purposes. See Walker v. Life Insurance Co. of the Sw., 953 F.3d 624, 630 (9th Cir. 2020) (“To
bring a UCL claim, a plaintiff must establish he suffered ‘as a result of’ the defendant’s conduct.”)

11

United States District Court
Northern District of California

10

(quoting Cal. Bus. & Prof. Code § 17204); In re Tobacco II Cases, 46 Cal. 4th 298, 325 (Cal.

12

2009) (named plaintiffs, not absent ones, must provide evidence of actual reliance at the

13

certification stage).
Meta’s mention of an arbitration provision in contracts for advertising after May 2018,

14
15

Dkt. No. 294 at 17, also does not defeat the adequacy and typicality of the named plaintiffs. The

16

complaint in this case was filed in August 2018. Dkt. No. 1. Despite that, and knowing of the

17

arbitration clause and its possible application to plaintiffs, Meta never sought to compel

18

arbitration, and instead vigorously litigated this lawsuit in federal court as if arbitration were not

19

an option. A good argument can be made that Meta has waived arbitration on this record. See

20

Anderson v. Starbucks Corp., No. 20-cv-01178-JD, 2022 WL 797014 (N.D. Cal. March 16, 2022)

21

(and cases cited therein). In addition, the record shows that the named plaintiffs purchased ads

22

before and after May 2018, which indicates that they are adequate representatives for advertisers

23

who purchased ads both before and after May 28, 2018. See Dkt. No 328-2 at ¶ 21. If for some

24

presently unknown reason an adjustment to the class definition might be required on arbitration

25

grounds, the Court can alter or amend it at any time before entry of a final judgment. Fed. R. Civ.

26

P. 23(c)(1)(C); see also Powers v. Hamilton Cty. Pub. Def. Com'n, 501 F.3d 592, 619 (6th Cir.

27

2007).

28

Plaintiffs have satisfied the elements of adequacy and typicality.
7

1
2
3
4
5

C.

Commonality (23(a)(2)) and Predominance (23(b)(3)

The commonality requirement under Rule 23(a)(2) is satisfied when “there are questions of
law or fact common to the class.” Fed. R. Civ. P. 23(a)(2). Because “any competently crafted
class complaint literally raises common questions,” the Court’s task is to look for a common
contention “capable of classwide resolution -- which means that determination of its truth or
falsity will resolve an issue that is central to the validity of each one of the claims in one stroke.”

6
Alcantar, 800 F.3d at 1052 (internal quotations and citations omitted). What matters is the
7
8
9

“capacity of a class-wide proceeding to generate common answers apt to drive the resolution of
the litigation.” Dukes, 564 U.S. at 350 (internal quotations omitted) (emphasis in original). This
does not require total uniformity across a class. “The existence of shared legal issues with

10
divergent factual predicates is sufficient, as is a common core of salient facts coupled with

United States District Court
Northern District of California

11
12
13

disparate legal remedies within the class.” Hanlon, 150 F.3d at 1019. “[E]ven a single common
question will do.” Dukes, 564 U.S. at 359. The commonality standard imposed by Rule 23(a)(2)
is “rigorous.” Leyva v. Medline Indus. Inc., 716 F.3d 510, 512 (9th Cir. 2013).

14
Rule 23(b)(3) sets out the related but nonetheless distinct requirement that the common
15
questions of law or fact predominate over the individual ones. This inquiry focuses on whether
16
17

the “common questions present a significant aspect of the case and [if] they can be resolved for all
members of the class in a single adjudication.” Hanlon, 150 F.3d at 1022 (internal quotations and

18
citation omitted); see also Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442, 453 (2016). Each
19
element of a claim need not be susceptible to classwide proof, Amgen, 568 U.S. at 468-69, and the
20

“important questions apt to drive the resolution of the litigation are given more weight in the

21
predominance analysis over individualized questions which are of considerably less significance
22
23

to the claims of the class.” Torres v. Mercer Canyons Inc., 835 F.3d 1125, 1134 (9th Cir. 2016).
Rule 23(b)(3) permits certification when “one or more of the central issues in the action are

24
common to the class and can be said to predominate, . . . even though other important matters will
25
have to be tried separately, such as damages or some affirmative defenses peculiar to some
26

individual class members.” Tyson, 577 U.S. at 453 (internal quotations omitted).

27
28
8

1

“Rule 23(b)(3)’s predominance criterion is even more demanding than Rule 23(a),”

2

Comcast, 569 U.S. at 34, and the main concern under subsection (b)(3) is “the balance between

3

individual and common issues.” In re Hyundai and Kia Fuel Economy Litigation, 926 F.3d 539,

4

560 (9th Cir. 2019) (en banc) (internal quotations omitted). The Court finds it appropriate to

5

assess commonality and predominance in tandem, with a careful eye toward ensuring that the

6

specific requirements of each are fully satisfied. See, e.g., Just Film, Inc. v. Buono, 847 F.3d

7

1108, 1120-21 (9th Cir. 2017).

8

1.

Liability

Plaintiffs have demonstrated that the main liability issues are common to the class

10

members and are capable of resolution with common evidence. For the fraudulent concealment

11

United States District Court
Northern District of California

9

and fraudulent misrepresentation claims, plaintiffs must show: “(a) misrepresentation (false

12

representation, concealment, or nondisclosure); (b) knowledge of falsity (or ‘scienter’); (c) intent

13

to defraud, i.e. to induce reliance; (d) justifiable reliance; and (e) resulting damage.” Engalla v.

14

Permanente Med. Grp., Inc., 15 Cal. 4th 951, 974 (1997). For plaintiffs’ UCL claims (for which

15

only commonality must be shown as part of the 23(a) factors, given the unavailability of monetary

16

relief), plaintiffs must show that members of the public were likely to be deceived. Williams v.

17

Gerber Products Co., 552 F.3d 934, 938 (9th Cir. 2008) (claims under UCL and CLRA are

18

“governed by the ‘reasonable consumer’ test”; plaintiffs “must show that members of the public

19

are likely to be deceived”) (internal quotations and citations omitted).

20

Consequently, the main liability question is the same for all class members: did Meta’s

21

Potential Reach metric mislead advertisers? Meta does not disagree, and instead hurls a grab bag

22

of challenges to plaintiffs’ ability of proving an answer in their favor. Much of Meta’s argument

23

against commonality and predominance is simply that the evidence does not support plaintiffs’

24

case. That is not the pertinent inquiry at the certification stage. The question is whether it makes

25

sense under Rule 23 and as a matter of due process and efficiency to present the liability dispute to

26

a jury on behalf of a class. Whether plaintiffs can ultimately prove it up at trial is a different

27

matter altogether.

28
9

1

To the extent a merits inquiry is warranted, plaintiffs have adduced evidence showing that

2

all class members were exposed to a similar representation about the ability of Potential Reach to

3

reach “people,” namely unique individuals. See, e.g., Dkt. No. 282-3; Dkt. No. 281-9 at ¶¶ 55-60.

4

This is seen in the Ads Manager interface, which represented Potential Reach as a number of

5

people. See, e.g., Dkt. No. 281-8. The evidence further shows that Meta’s Potential Reach metric

6

was not actually an estimate of people reached, but an estimate of “accounts” reached. See Dkt.

7

No. 281-60 at ECF 10. Because the number of unique accounts and unique people were different,

8

this led to an inaccurate representation of how many people the advertisements could reach. See

9

Dkt. No. 281-11 at ¶ 15.

10

Meta does not dispute that the Potential Reach numbers were presented in terms of people.

United States District Court
Northern District of California

11

Instead, Meta says that the Potential Reach numbers were not uniformly inaccurate as a result of

12

different targeting criteria producing different Potential Reach numbers. Dkt. No. 293-4 at 18-20.

13

Even so, Potential Reach was always expressed as a number of “people,” and the discrepancy

14

between people and accounts made the number inaccurate, even if the numerical value of the

15

inaccuracy varied across advertisers. Consequently, plaintiffs have shown that the question of

16

whether Meta made misrepresentations to all class members can be shown through common

17

evidence.

18

Meta’s knowledge of the misleading statements, and intent to deceive, also lend

19

themselves to resolution by common evidence. See, e.g., Brickman v. Fitbit, Inc., No. 15-cv-

20

2077-JD, 2017 WL 5569827, at *6 (N.D. Cal. Nov. 20, 2017) (citing Small v. Fritz Cos., Inc., 30

21

Cal. 4th 167, 173-74 (2003)). Several documents show that Meta knew that its Potential Reach

22

estimate did not accurately reflect the number of people its advertisements could reach. See Dkt.

23

No. 281-25; Dkt. No. 281-27. Meta’s intent for advertisers to rely on its Potential Reach numbers

24

is also provable through common evidence. Meta knew that the potential reach number was the

25

most important number in its ads creation interface and that advertisers frequently relied on the

26

estimated audience to build their budgets and advertising strategies. Dkt. No. 281-8.

27
28

So too for materiality and reliance. In common law and UCL fraud cases, questions of
materiality and reliance do not necessarily undermine predominance and commonality. Brickman,
10

1

2017 WL 5569827, at *6-*7; Milan v. Clif Bar & Co., No. 18-cv-2354-JD, 2021 WL 4427427, at

2

*5 (N.D. Cal. Sep. 27, 2021). “[A] presumption, or at least an inference, of reliance arises

3

wherever there is a showing that a misrepresentation was material.” Tobacco II Cases, 46 Cal. 4th

4

at 327. A misrepresentation is material “if a reasonable man would attach importance to its

5

existence or nonexistence in determining his choice of action in the transaction in question.” Id.

6

(internal quotations omitted). The question of materiality “can be proved through evidence

7

common to the class.” Amgen, 568 U.S. at 467. Plaintiffs have established that materiality and

8

reliance can be shown in this case through common evidence. Potential Reach metrics were

9

shown to all advertisers in the Ads Manager. Dkt. No. 282-3; Dkt. No. 282-4. Meta has
acknowledged that Potential Reach is an important number for advertisers. Dkt. No. 281-8. A

11

United States District Court
Northern District of California

10

majority of advertisers rely on Potential Reach as a metric for their advertisements. Dkt. No. 281-

12

22.

13

Plaintiffs have also established that proof of injury is susceptible to common evidence.

14

Among other evidence, a report from Pivotal Research showed that Potential Reach numbers

15

exceeded census counts for various demographics, Dkt. No. 282-22 and several internal

16

documents indicated various causes of inflated Potential Reach levels, see, e.g., Dkt. No. 282-28;

17

282-7; 282-31; 282-32. Plaintiffs’ expert, Dr. Cowan, conducted a statistical analysis to determine

18

the percentage of inflation for both nationwide and targeted advertisements. See Dkt. No. 282-8.

19

He concluded that it was a statistical certainty that, for any advertisement with a Potential Reach

20

of at least 1,000 people or more, the estimate would be significantly inflated above the actual

21

number of people the advertisement could reach. Id.

22

Meta says that Dr. Cowan improperly assumed that the inflated estimates found in the

23

default national population (United States, aged 18-65) Potential Reach were equally applicable

24

across all targeted groups, and that each measure of inflation was distributed across targeted

25

groups. Dkt. No. 281-11 ¶ 82. Meta’s expert, Dr. Steven Tadelis, says that this is a flawed

26

assumption because Meta’s data sampling shows that sources of inflation are not distributed

27

evenly across all smaller demographics that an advertiser might choose. Dkt. No. 293-44 ¶ 125.

28

But Dr. Tadelis does not conclude that no inflation occurred at all, only that Dr. Cowan did not
11

1

measure the exact inflation resulting from any given targeting criteria because inflation for any

2

given sub population may be different from the inflation for the default national population. This

3

criticism does not foreclose classwide proof of injury.

4

2.

Damages and Daubert Motions re Dr. Allenby and Mr. McFarlane

5

While a damages methodology need not deliver mathematical precision, and may

6

accommodate some individual variability among class members, see In re Capacitors Antitrust

7

Litigation, No. 17-md-2801-JD, 2018 WL 5980139, at *9 (N.D. Cal. Nov. 14, 2018), it must be

8

capable of determining damages across the class in a reasonably accurate fashion. Comcast, 569

9

U.S. at 35 (plaintiffs bear burden of showing that “damages are susceptible of measurement across
the entire class for purposes of Rule 23(b)(3)”). The damages model “must measure only those

11

United States District Court
Northern District of California

10

damages attributable to” the plaintiffs’ theory of liability. Id. Put plainly, the damages model

12

must reasonably reflect the claims and evidence in the case.

13

Plaintiffs have proffered experts who analyzed the evidence to arrive at a price premium

14

that advertisers paid for inflated Potential Reach values. Dkt. No. 281-3 at 21. Dr. Cowan

15

measured the amount of inflation associated with Potential Reach as a result of the misleading

16

“people” metric. Id. Dr. Allenby used a “conjoint survey” to test the impact of inflated Potential

17

Reach on advertisers’ budgets. Id. Dr. Roughgarden, an auction expert, calculated a price

18

premium. Id. Dr. Levy, an economist, confirmed that Dr. Roughgarden’s price premium properly

19

considered supply and demand, and that damages could be calculated on a classwide basis. Id.

20

Plaintiffs also offer expert witness Mr. McFarlane, who opined about the price premium class

21

members paid compared to if no potential reach metric was provided at all. Id.

22

Meta offers little in its class certification brief to attack plaintiffs’ damages models. It

23

relies instead on two separately filed Daubert motions to exclude the opinions of Dr. Allenby and

24

Mr. McFarlane, and by extension, the portions of Dr. Levy and Dr. Roughgarden’s opinions that

25

rely on the reports of Dr. Allenby and Mr. McFarlane. Dkt. Nos. 284-4, 284-6.

26

Overall, Meta has not demonstrated a good reason to exclude Dr. Allenby’s work. Under

27

the familiar standards of Federal Rule of Evidence 702 and Daubert v. Merrell Dow

28

Pharmaceuticals, Inc., 509 U.S. 579 (1993), there is no “definitive checklist or test” used to
12

1

evaluate the reliability of proposed expert testimony. Daubert, 509 U.S. at 593-94. The question

2

for the Court at this stage is to decide whether Dr. Allenby will use a generally accepted method

3

for determining price premiums, or whether his approach is “junk science” akin to predicting

4

criminality by feeling the bumps on a person’s head. General Electric Co. v. Joiner, 522 U.S.

5

136, 153 n.6 (1997) (Stevens, J., concurring in part).

6

The “inquiry into the evidence’s ultimate admissibility should go to the weight that

7

evidence is given at the class certification stage.” Sali v. Corona Reg’l Med. Ctr., 9-9 F.3d 996,

8

1006 (9th Cir. 2018). The Court determines whether the expert evidence helps to establish

9

whether class certification is appropriate. See Ellis v. Costco Wholesale Corp., 657 F.3d 970, 982

10

(9th Cir. 2011).

United States District Court
Northern District of California

11

Dr. Allenby conducted a conjoint survey and analyzed the data using both a linear

12

regression model and a “logit model” (another type of statistical analysis) before determining that

13

the logit model did not best fit the data. Meta does not suggest that a conjoint survey is an

14

untested method, nor does it claim that it is improper to use a linear regression to analyze survey

15

data. Rather, Meta says that the specific regression that Dr. Allenby used was a novel type of

16

analysis that purposely excluded data from the analysis. Dkt. No. 284-4 at 10-12.

17

This Court has found conjoint analysis to be a reliable method of determining price

18

premiums. See, e.g., Milan v. Clif Bar & Co., No. 18-cv-2354-JD, 2021 WL 4467427, at * 7

19

(N.D. Cal. Sep. 27, 2021). Meta does not dispute the generally utility of conjoint analysis, and

20

focuses its critique on Dr. Allenby’s use of a linear regression model to analyze the data from the

21

conjoint survey. Dkt. No. 284-4 at 10. Plaintiffs have shown that Dr. Allenby chose a linear

22

regression model that is a standard method for analyzing this data. Dkt. No. 304-17 at 143:9-18;

23

304-20 at 57:23-58:7. Dr. Allenby’s choice of one particular data analysis method over another

24

goes to the weight of his opinion, not its admissibility. Fortune Dynamic, Inc. v. Victoria’s Secret

25

Stores Brand Mgmt., Inc., 618 F.3d 1025, 1036 (9th Cir. 2010). To the extent that Meta suggests

26

that Dr. Allenby improperly limited his data set, this too is a question of weight to be afforded to

27

the opinion, not its admissibility. In re Capacitors Antitrust Litig., No. 17-md-2801-JD, 2018 WL

28

5980139, at *6 (N.D. Cal. Nov. 14, 2018). Dr. Allenby states that he chose a subset of the data to
13

1

analyze based on the fact that his conjoint survey included allocations of advertising for both Meta

2

and Google ads, but only Meta ads are at issue in this case. Dkt. No. 304-17 at 288:10-289:8.

3

This is enough to be sound and useful for certification purposes. If evidence emerges at

4

trial that substantially impeaches Dr. Allenby’s methods and conclusions, the door may be opened

5

to consideration of decertification.

6

Meta’s objections to Mr. McFarlane’s report lead to a different outcome. Meta says that

7

Mr. McFarlane offered nothing more than his personal interpretation of documents and evidence.

8

Dkt. No. 284-6 at 7. Meta also says that Mr. McFarlane used a price premium figure that he did

9

not calculate, and merely applied it in an obvious fashion to the amount of money plaintiffs are

10

United States District Court
Northern District of California

11

said to have spent on advertising. Id. at 3.
These objections are well taken. Overall, Mr. McFarlane’s report does not offer any

12

specialized or scientific expertise, or anything beyond the typical knowledge and experience of a

13

jury. See Fed. R. Evid. 702; Daubert, 509 U.S. at 592. The documents Mr. McFarlane interprets

14

are reasonably intelligible to a jury without special assistance. Consequently, exclusion of Mr.

15

McFarlane’s opinions and report is required. Any portion of Dr. Roughgarden’s opinions that is

16

drawn on Mr. McFarlane’s work is also excluded, unless an independent basis for it is

17

demonstrated. The Court declines to undertake that analysis on the record as it currently stands.

18

Meta may pursue it in a motion in limine, as circumstances warrant.

19

Because plaintiffs have adequately shown that they can calculate damages on a classwide

20

basis using Dr. Allenby’s report and evidence from their other experts (excluding Mr. McFarlane),

21

they have shown an adequate damages model under Rule 23(b)(3).

22

D.

23

The final certification question is whether the ends of justice and efficiency are served by

24

certification. Rule 23(b)(3) requires a finding that proceeding as a class is superior to other ways

25

of adjudicating the controversy, which in this case would mean individual actions by each putative

26

class member. There can be no doubt here that a class is the superior method of handling the

27

claims of individual advertisers. The price premium at issue here for each advertiser is no more

28

than $32, Dkt. No. 281-3, and it is not likely for class members to recover large amounts

Superiority

14

1

individually if they prevailed. No reasonable person is likely to pursue these claims on his or her

2

own, especially given the cost and other resources required to litigate against a company like

3

Meta, which has already retained multiple experts and shown that it is committed to strongly

4

defending this case. This all “vividly points to the need for class treatment.” Just Film, 847 F.3d

5

at 1123.

6

IV.

7

RULE 23(B)(2) CLASS
Plaintiffs seek certification of a class under Rule 23(b)(2) for the UCL injunctive relief

8

claim. Such a class may be certified when “the party opposing the class has acted or refused to act

9

on grounds that apply generally to the class, so that final injunctive relief or corresponding
declaratory relief is appropriate respecting the class as a whole.” Fed. R. Civ. P. 23(b)(2). “Class

11

United States District Court
Northern District of California

10

certification under Rule 23(b)(2) is appropriate only where the primary relief sought is declaratory

12

or injunctive.” Zinser, 253 F.3d at 1195. The primary use of Rule 23(b)(2) classes has been the

13

certification of civil rights class actions, but courts have certified many different kinds of classes

14

under Rule 23(b)(2). See Parsons v. Ryan, 754 F.3d 657, 686 (9th Cir. 2014). The Rule 23(a)

15

requirements of numerosity, commonality, typicality, and adequacy must also be shown for a Rule

16

23(b)(2) class. Zinser, 253 F.3d at 186. As discussed, plaintiffs have met their burden for proving

17

the Rule 23(a) requirements.

18

For Rule 23(b)(2), the Court is not required “to examine the viability or bases of class

19

members’ claims for declaratory and injunctive relief, but only to look at whether class members

20

seek uniform relief from a practice applicable to all of them.” Rodriguez v. Hayes, 591 F.3d 1105,

21

1125 (9th Cir. 2010). “It is sufficient to meet the requirements of Rule 23(b)(2) that class

22

members complain of a pattern or practice that is generally applicable to the class as a whole.” Id.

23

(quoting Walters v. Reno, 145 F.3d 1032, 1047 (9th Cir. 1998)) (internal quotations omitted).

24

The California Supreme Court has held that “[i]njunctions are the ‘primary form of relief

25

available under the UCL to protect consumers from unfair business practices.” Kwikset Corp. v.

26

Superior Court, 51 Cal.4th 310, 337 (2011); see also Tobacco Cases II, 46 Cal. 4th at 319. For

27

the proposed Rule 23(b)(2) class, plaintiffs seek injunctive relief in the form of an order directing

28
15

1

Meta to “either (a) correct the [Potential Reach] metric by removing known sources of inflation, or

2

(b) remove the [Potential Reach] metric altogether.” Dkt. No. 281-3 at 18.

3

Plaintiffs have standing to seek an injunction. As our circuit has determined, “a previously

4

deceived consumer may have standing to seek an injunction against false advertising or labeling,

5

even though the consumer now knows or suspects that the advertising was false at the time of the

6

original purchase,” because “[k]nowledge that the advertisement or label was false in the past does

7

not equate to knowledge that it will remain false in the future.” Davidson v. Kimberly-Clark

8

Corp., 889 F.3d 956, 969 (9th Cir. 2018). Plaintiffs have proffered deposition testimony to the

9

effect that they would consider purchasing ads from Meta again if Meta corrected or removed the
misleading Potential Reach metric. Dkt. No. 282-65 at 242:18-23; Dkt. No. 282-64 at 105:24-

11

United States District Court
Northern District of California

10

106:5. This establishes plaintiffs’ standing to pursue injunctive relief in this case.

12

Meta’s arguments to the contrary are unavailing. To start, Meta repeats the same

13

arguments that it already made in its motion for judgment on the pleadings, Dkt. No. 270, that

14

plaintiffs have failed to show they lack an adequate remedy at law. The Court has already

15

determined that plaintiffs have shown an inadequate remedy at law for their injunctive relief claim

16

under the UCL. Dkt. No. 366 at 2.

17

Meta also says that plaintiffs did not show they face a threat of actual future harm because

18

at least one inflation source has already been remediated and Meta updated disclosures about

19

multiple accounts. Dkt. No. 293-4 at 25. This is a merits question that is not properly decided at

20

the class certification stage.

21

Meta’s passing comment that the injunction plaintiffs seek is “overbroad and unworkable,”

22

Dkt. No. 293-4 at 25, is no basis for denying certification. The remark was not developed in a

23

meaningful way, and concerns about the scope of an inunction are premature at this stage. See

24

B.K. ex rel. Tinsley v. Snyder, 922 F.3d 957, 972 (9th Cir. 2019). There is considerably more to be

25

done in this case, namely trial, before the specific terms of an injunction might warrant debate.

26

Consequently, a Rule 23(b)(2) class is appropriate for plaintiffs’ UCL claims.

27
28
16

1

CONCLUSION

2

The Court certifies the proposed class under Rule 23(b)(3) for the common law fraud

3

claims, and under Rule 23(b)(2) for the UCL injunction claim. Plaintiffs DZ Reserve, Inc. and

4

Cain Maxwell are appointed class representatives, and their counsel at Cohen Milstein Sellers &

5

Toll PLLC and the Law Offices of Charles Reichmann are appointed class counsel.

6
7
8
9
10

United States District Court
Northern District of California

11

Meta’s motion to exclude the report and testimony of Dr. Allenby is denied. Meta’s
motion to exclude the report and testimony of Mr. McFarlane is granted.
Plaintiffs are directed to file by April 29, 2022, a proposed plan for dissemination of notice
to the classes. Plaintiffs will meet and confer with Meta at least 10 days in advance of filing the
plan so that the proposal can be submitted on a joint basis, to the fullest extent possible.
A status conference is set for May 26, 2022, at 10:00 a.m. in Courtroom 11, 19th Floor,

12

San Francisco. The parties are directed to file a joint statement by May 19, 2022, with proposed

13

dates for the final pretrial conference and trial.

14
15
16
17

The parties are referred to Magistrate Judge Hixson for a settlement conference to be held
as his schedule permits.
IT IS SO ORDERED.
Dated: March 29, 2022

18
19
JAMES DONATO
United States District Judge

20
21
22
23
24
25
26
27
28
17