Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 1 of 31
`
`LATHAM & WATKINS LLP
`Elizabeth L. Deeley (CA Bar No. 230798)
`elizabeth.deeley@lw.com
`Melanie M. Blunschi (CA Bar No. 234264)
`melanie.blunschi@lw.com
`Nicole C. Valco (CA Bar No. 258506)
`nicole.valco@lw.com
`505 Montgomery Street, Suite 2000
`San Francisco, CA 94111-6538
`Telephone: +1.415.391.0600
`Facsimile: +1.415.395.8095
`Andrew B. Clubok (pro hac vice)
`andrew.clubok@lw.com
`Susan E. Engel (pro hac vice)
`susan.engel@lw.com
`555 Eleventh Street, N.W., Suite 1000
`Washington, D.C. 20004-1304
`Telephone: +1.202.637.2200
`Facsimile: +1.202.637.2201
`Hilary H. Mattis (CA Bar No. 271498)
` hilary.mattis@lw.com
`140 Scott Drive
`Menlo Park, CA 94025-1008
`Telephone: +1.650.328.4600
`Facsimile: +1.650.463.2600
`Attorneys for Defendant Facebook, Inc.
`
`UNITED STATES DISTRICT COURT
`NORTHERN DISTRICT OF CALIFORNIA
`SAN FRANCISCO DIVISION
`
` DZ RESERVE and CAIN MAXWELL (d/b/a
`MAX MARTIALIS), individually and on
`behalf of all others similarly situated,
`
`Plaintiffs,
`
`v.
`FACEBOOK, INC.,
`
`Defendant.
`
` Case No. 3:18-cv-04978 JD
`
`FACEBOOK, INC.’S CORRECTED
`OPPOSITION TO PLAINTIFFS’ MOTION
`FOR CLASS CERTIFICATION
`
`Date: June 10, 2021
`Time: 10:00 a.m.
`Court: Courtroom 11, 19th Floor
`Hon. James Donato
`
`REDACTED VERSION REFILED PURSUANT TO DKT NO. 350 ORIGINALLY
`FILED AT DKT. 301-4
`FACEBOOK’S CORRECTED OPP. TO MOT.
`FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 2 of 31
`
`TABLE OF CONTENTS
`
`B.
`
`C.
`
`D.
`
`E.
`
`B.
`
`INTRODUCTION ............................................................................................................. 1
`BACKGROUND ............................................................................................................... 4
`A.
`Potential Reach Estimates Are A Targeting Tool, Not A Budget
`Tool ........................................................................................................................ 4
`During The Class Period, Facebook Updated The Way Potential
`Reach Was Estimated But Rejected Some Changes Because They
`Were Flawed .......................................................................................................... 6
`Every Ad Campaign Is A Different Product Sold At A Different
`Price ....................................................................................................................... 9
`Advertisers Rely On Different Information (Not Potential Reach)
`In Setting Their Budgets And Measuring Results ............................................... 11
`Plaintiffs’ Conduct and Contemporaneous Documents Contradict
`Their Claimed Reliance on Potential Reach ........................................................ 13
`LEGAL STANDARD ...................................................................................................... 14
`PLAINTIFFS HAVE NOT DEMONSTRATED TYPICALITY,
`ADEQUACY, OR UCL STANDING ............................................................................. 15
`PLAINTIFFS HAVE NOT DEMONSTRATED PREDOMINANCE
`BECAUSE THEY HAVE NO COMMON PROOF OF LIABILITY ............................. 17
`A.
`No Uniform Material Misstatement Or Omission Was Made To
`The Class .............................................................................................................. 18
`Plaintiffs’ Ad Auction Evidence Cannot Prove That Any Potential
`Reach Statement Caused Classwide Injury.......................................................... 22
`PLAINTIFFS HAVE NOT DEMONSTRATED PREDOMINANCE
`BECAUSE THEY HAVE NO CLASSWIDE DAMAGES OR
`RESTITUTION MODEL ................................................................................................ 24
`VII. A CLASS ACTION IS NOT SUPERIOR ....................................................................... 24
`VIII. PLAINTIFFS HAVE NOT MET THEIR BURDEN UNDER RULE
`23(B)(2)............................................................................................................................ 25
`CONCLUSION ................................................................................................................ 25
`
`I.
`II.
`
`III.
`IV.
`
`V.
`
`VI.
`
`IX.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`i
`
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 3 of 31
`
`
`TABLE OF AUTHORITIES
`
`CASES
`
`
`
`Page(s)
`
`All. Mortg. Co. v. Rothwell,
`10 Cal. 4th 1226 (1995) ........................................................................................................ 24
`
`Am. Express Co. v. Italian Colors Rest.,
`570 U.S. 228 (2013) .............................................................................................................. 14
`
`Avilez v. Pinkerton Gov’t Servs. Inc.,
`596 F. App’x 579 (9th Cir. 2015) ......................................................................................... 17
`
`Berger v. Home Depot USA, Inc.,
`741 F.3d 1061 (9th Cir. 2014) .............................................................................................. 18
`
`Brazil v. Dell Inc.,
`2010 WL 5387831 (N.D. Cal. Dec. 21, 2010) ...................................................................... 16
`
`Brickman v. Fitbit, Inc.,
`2017 WL 5569827 (N.D. Cal. Nov. 20, 2017) ..................................................................... 19
`
`Chowning v. Kohl’s Dep’t Stores, Inc.,
`733 F. App’x 404 (9th Cir. 2018) ......................................................................................... 24
`
`Comcast Corp. v. Behrend,
`569 U.S. 27 (2013) .................................................................................................... 15, 20, 24
`
`Ebner v. Fresh, Inc.,
`838 F.3d 958 (9th Cir. 2016) ................................................................................................ 22
`
`Ellis v. Costco Wholesale Corp.,
`657 F.3d 970 (9th Cir. 2011) ................................................................................................ 15
`
`Fox Test Prep v. Facebook, Inc.,
`588 F. App’x 733 (9th Cir. 2014) ......................................................................................... 17
`
`Harris v. Vector Mktg., Corp.,
`753 F. Supp. 2d 996 (N.D. Cal. 2010) .................................................................................. 16
`
`Hinesley v. Oakshade Town Ctr.,
`135 Cal. App. 4th 289 (2005) ............................................................................................... 22
`
`Hirsch v. USHealth Advisors, LLC,
`337 F.R.D. 118 (N.D. Tex. 2020) ......................................................................................... 16
`
`In re Facebook, Inc., PPC Advert. Litig.,
`282 F.R.D. 446 (N.D. Cal. 2012) .......................................................................................... 17
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`
`
`
`
`
`
`
`
`ii
`
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 4 of 31
`
`
`In re Hyundai & Kia Fuel Econ. Litig.,
`926 F.3d 539 (9th Cir. 2019) ................................................................................................ 25
`
`In re Tobacco II Cases,
`46 Cal. 4th 298 (2009) .................................................................................................... 14, 15
`
`In re Vioxx Class Cases,
`180 Cal. App. 4th 116 (2009) ............................................................................................... 18
`
`Jones v. ConAgra Foods, Inc.,
`2014 WL 2702726 (N.D. Cal. June 13, 2014) ...................................................................... 19
`
`Lanovaz v. Twinings N. Am., Inc.,
`726 F. App’x 590 (9th Cir. 2018) ......................................................................................... 25
`
`Lucas v. Berg, Inc.,
`212 F. Supp. 3d 950 (S.D. Cal. 2016) ................................................................................... 18
`
`Microsoft Corp. v. Baker,
`137 S. Ct. 1702 (2017) .......................................................................................................... 18
`
`Miller v. RP On-Site, LLC,
`2020 WL 6940936 (N.D. Cal. Nov. 25, 2020) ..................................................................... 16
`
`Mirkin v. Wasserman,
`5 Cal. 4th. 1082 (1993) ......................................................................................................... 17
`
`Olean Wholesale Grocery Coop., Inc. v. Bumble Bee Foods LLC,
`993 F.3d 774 (9th Cir. 2021) ......................................................................................... passim
`
`Reitman v. Champion Petfoods USA, Inc.,
`2019 WL 7169792 (C.D. Cal. Oct. 30, 2019) ....................................................................... 19
`
`Reitman v. Champion Petfoods USA, Inc.,
`830 F. App’x 880 (9th Cir. 2020) ......................................................................................... 19
`
`Roley v. Google LLC,
`2020 WL 8675968 (N.D. Cal. July 20, 2020) ....................................................................... 20
`
`Rushing v. Williams Sonoma, Inc.,
`2020 WL 6787135 (N.D. Cal. Oct. 8, 2020) ......................................................................... 17
`
`Sali v. Corona Reg’l Med. Ctr.,
`909 F.3d 996 (9th Cir. 2018) ................................................................................................ 15
`
`Shady Grove Orthopedic Assocs., P.A. v. Allstate Ins. Co.,
`559 U.S. 393 (2010) .............................................................................................................. 22
`
`iii
`
`Sonner v. Premier Nutrition Corp.,
`971 F.3d 834 (9th Cir. 2020) ................................................................................ 1, 22, 24, 25
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`
`
`
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`
`
`
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 5 of 31
`
`
`Spacone v. Sanford, L.P.,
`2018 WL 4139057 (C.D. Cal. Aug. 9, 2018) ........................................................................ 16
`
`Stearns v. Ticketmaster Corp.,
`655 F.3d 1013 (9th Cir. 2011) ........................................................................................ 16, 22
`
`Stormans, Inc. v. Selecky,
`586 F.3d 1109 (9th Cir. 2009) .............................................................................................. 25
`
`Sussex Fin. Enter., Inc. v. Bayerische Hypo-Und Vereinsbank AG,
`460 F. App’x 709 (9th Cir. 2011) ......................................................................................... 17
`
`Tucker v. Pac. Bell Mobile Servs.,
`208 Cal. App. 4th 201 (2012) ............................................................................................... 18
`
`Tyson Foods, Inc. v. Bouaphakeo,
`577 U.S. 442 (2016) ........................................................................................................ 17, 23
`
`Villiarimo v. Aloha Island Air, Inc.,
`281 F.3d 1054 (9th Cir. 2002) .............................................................................................. 15
`
`Walker v. Life Ins. Co. of the Sw.,
`953 F.3d 624 (9th Cir. 2020) .............................................................................. 14, 15, 17, 22
`
`Wal-Mart Stores, Inc. v. Dukes,
`564 U.S. 338 (2011) .............................................................................................................. 14
`
`Williams v. Gerber Prods. Co.,
`552 F.3d 934 (9th Cir. 2008) ................................................................................................ 17
`
`Zakaria v. Gerber Prod. Co.,
`755 F. App’x 623 (9th Cir. 2018) ......................................................................................... 23
`
`Zinser v. Accufix Research Inst., Inc.,
`253 F.3d 1180 (9th Cir. 2001) .......................................................................................... 3, 25
`
`RULES
`
`Fed. R. Civ. P. 23 ........................................................................................................ 2, 14, 21, 22
`
`Fed. R. Civ. P. 23(b)(2)........................................................................................................... 3, 25
`
`Fed. R. Civ. P. 23(b)(3)............................................................................................................... 24
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`iv
`
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 6 of 31
`
`
`I.
`
`INTRODUCTION
`Two advertisers—a firearm accessory reseller and a drop-shipping business—ask this
`Court to represent more than three million advertisers—including large corporations, medium-
`sized businesses, and individual proprietors that have varying objectives and purchasing
`experiences—based on their false assertions that Facebook intentionally and uniformly inflated
`Potential Reach estimates, supposedly causing advertisers across the board to pay more for
`Facebook ads than they otherwise would have. At the threshold, Plaintiffs ignore Facebook’s
`motion for judgment on the pleadings, which seeks dismissal of their UCL claim under Sonner v.
`Premier Nutrition Corp., 971 F.3d 834 (9th Cir. 2020), Dkt. 270. Beyond Sonner, their theory of
`liability and harm for both the UCL and fraud claims hinges on the proposition that every advertiser
`received a deceptively high Potential Reach estimate and that millions of advertisers relied on
`those estimates to increase their ad budgets, which supposedly drove up all ad prices. None of this
`is true. Plaintiffs’ narrative misconstrues the record and cannot disguise the fact that this case—
`with or without the UCL claim—is unsuitable for class treatment.
`Potential Reach is one of several tools Facebook provides advertisers to help them target
`their ad campaigns. Facebook calculates Potential Reach in real time when an advertiser selects
`targeting and placement criteria (from vast options) to provide a unique estimate of the number of
`people who received an ad on Facebook in the last 30 days and who match the selected criteria.
`Potential Reach is not designed to match census data (as Facebook discloses), but reflects real-
`time demographics, interests, and behaviors based on what people do and share on Facebook.
`Plaintiffs do not claim (and have no evidence) that Facebook made a computational error in its
`calculation. Rather, Plaintiffs believe Facebook should describe or compute Potential Reach
`differently, yet they fail to tell the Court that Facebook did make updates to improve its estimates,
`but rejected some proposed updates because they would have made Potential Reach less accurate.
`Most important for purposes of their motion for class certification (“Motion”), Plaintiffs
`fail to present evidence showing that advertisers as a class uniformly relied on “inflated” Potential
`Reach estimates to increase their budgets. In fact, Plaintiffs’ expert (Allenby) conducted a study
`showing that when presented with lower Potential Reach estimates, nearly half of the advertisers
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`
`
`
`
`1
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`
`
`
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 7 of 31
`
`
`2
`
`surveyed actually increased their budgets. Reibstein Ex. 3.1 Faced with these results, he invented
`a brand new, results-driven analysis to try to support Plaintiffs’ theory, but he still found that
`advertisers as a class do not rely on the description of Potential Reach as “people,” and 21% of
`advertisers did not increase their budgets based on allegedly inflated Potential Reach estimates.
`Reibstein ¶¶ 125-26 & Table 6. Those results are consistent with the named Plaintiffs’ own
`conduct—they did not rely on supposedly “inflated” Potential Reach to increase their budgets.
`Declarations from other advertisers confirm that they too do not rely on Potential Reach in their
`budget decisions, but instead use it (if at all) as a “directional guide” in selecting targeting criteria.
`E.g., Ex. B (Flanders) ¶ 8; Ex. D (McDowell) ¶ 10.2 That is no surprise; Potential Reach is not an
`estimate of advertisers’ results, which is what advertisers consider in setting budgets. E.g., Ex. A
`(Lowe) ¶ 8; Ex. C (Hampton) ¶ 10. In fact, Facebook provides entirely different (and
`unchallenged) estimates to predict advertisers’ results on the same webpage right below Potential
`Reach—these Estimated Daily Results estimate how many people the advertiser may actually
`reach or may click on an ad in a day based on the advertiser’s budget. See Ex. 78 at 11-12, 17-19;
`Ex. 93. Plaintiffs have not presented a single on-the-record statement from any other advertisers
`stating they relied on Potential Reach to set their budget. Without evidence that advertisers
`increased their budgets in reliance on “inflated” Potential Reach, Plaintiffs’ theory of classwide
`injury—which assumes based on their experts’ results-driven analysis that between 78% and 91%
`of advertisers increased budgets in response to Potential Reach—collapses.
`To paper over these fundamental problems, Plaintiffs’ experts use averages to assume away
`critical differences between advertisers. But to satisfy Rule 23, Plaintiffs must show by a
`preponderance of the evidence that their claims are capable of common proof, without using
`“average assumptions” that “mask individual differences among the class members,” and without
`sweeping into the class more than a de minimis number of uninjured class members. Olean
`
`1 Unless noted, all exhibits cited herein are attached to the Declarations of Michael Duffey and
`Melanie M. Blunschi (“Blunschi”) in Support of Facebook’s Opposition to Plaintiffs’ Motion .
`Facebook’s expert reports are cited herein by last name: Dr. Catherine Tucker (Dkt. 288-1), David
`Reibstein (Dkt. 289-1), Dr. Steven Tadelis, and Dr. Adam Porter (both filed concurrently). Cites
`to Reibstein Ex. 3 are to Reibstein’s concurrently-filed Supplemental Declaration. All internal
`quotation marks and citations are omitted and all emphases are added unless noted.
`2 Facebook reserves the right to submit additional testimony from other advertisers.
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`
`
`
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`
`
`
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 8 of 31
`
`
`Wholesale Grocery Coop., Inc. v. Bumble Bee Foods LLC, 993 F.3d 774, 791-92 (9th Cir. 2021).
`Plaintiffs cannot do so, and class certification should accordingly be denied.
`First, the remaining class representatives, Cain Maxwell and DZ Reserve, are atypical and
`inadequate, and lack UCL standing, because they did not rely on Potential Reach and lied about it.
`While telling the Court that DZ Reserve “would not have paid Facebook the money it did” had it
`known Potential Reach was allegedly inflated (Mot. at 12), DZ Reserve’s owner (and sole ad-
`buyer) continued to buy Facebook ads under false pretenses after joining this case. See infra
`Section II.E; Blunschi ¶¶ 69-77. And any claim that Maxwell relied on Potential Reach rings
`hollow given his admission that he actually relied on an online tutorial’s instructions to set his
`budgets—and that tutorial did not discuss Potential Reach. See infra Section II.E; Blunschi ¶ 89.
`Second, Plaintiffs cannot show predominance because there is no common proof of
`liability. Each ad an advertiser purchases is a different product with a different price based on
`different information and performance objectives. As set forth in Facebook’s Daubert motion
`(Dkt. 285), Plaintiff’s own expert (Allenby) finds that advertisers as a class do not rely on
`“inflated” Potential Reach to increase their budgets. That alone defeats class certification.
`Third, Plaintiffs also cannot show predominance because they have no viable classwide
`damages model. As explained in Facebook’s Daubert motion (Dkt. 286), the McFarlane model is
`based on a flawed “observation” about a single email and does not fit Plaintiffs’ theory of liability,
`and Plaintiffs offer no other measure of damages for their allegation that Potential Reach was
`misleadingly described as “people.” And as for Plaintiffs’ effort to show damages caused by the
`“inflated” Potential Reach estimates, that model relies on Allenby’s results-driven revised analysis,
`which is unreliable as set forth in Facebook’s second Daubert motion (Dkt. 285).
`Fourth, Plaintiffs cannot show superiority because class claims would require millions of
`mini-trials, and Plaintiffs do not attempt to satisfy their burden of showing “a suitable and realistic
`plan for trial.” Zinser v. Accufix Research Inst., Inc., 253 F.3d 1180, 1189 (9th Cir. 2001).
`Fifth, Plaintiffs’ Rule 23(b)(2) class fails because they are not entitled to injunctive relief,
`including for the reasons stated in Facebook’s motion for judgment on the pleadings, Dkt. 270.
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`
`
`
`
`
`
`
`
`3
`
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 9 of 31
`
`
`II.
`
`•
`
`BACKGROUND
`Plaintiffs devote the bulk of their Motion to a fabricated story that Facebook inflated
`Potential Reach to trick advertisers into spending more on Facebook.3 But they ignore key facts
`that doom class certification: (1) the proposed class of advertisers does not use Potential Reach as
`a tool for setting budgets; (2) Facebook updated Potential Reach throughout the Class Period but
`rejected the changes Plaintiffs now champion because they would have made Potential Reach less
`reliable; (3) advertisers bought different types of ads at different prices, and used different
`information to evaluate and set budgets for those ads; (4) the named Plaintiffs did not rely on
`Potential Reach; and (5) Plaintiffs’ experts assume away these issues through averages.
`A.
`Potential Reach Estimates Are A Targeting Tool, Not A Budget Tool
`Potential Reach is a planning tool that some advertisers use to help create a target audience
`for their ad campaigns—not for setting budgets. In declarations, Exs. A-J, Facebook advertisers
`explain that they use Facebook because it allows them to “target[] the right audiences with our
`ads,” Ex. H (Grantham) ¶ 3, or “target specific events, for specific groups of people,” Ex. B ¶ 3;
`see also Tucker ¶¶ 137-39 (advertisers received value from Facebook and continued to buy ads).
`Advertisers explained that Facebook advertising helps them get results:
`• “I choose to advertise on Facebook because it works. It is where my target audience is online
`and I get results. I check Google Analytics every other week and can tell that Facebook is
`currently responsible for approximately 37 percent of traffic driven to the Cork & Keg Tours
`website…In fact, I am planning to increase my advertising budget (likely to $500 per year)
`because of the strong results I have been getting with my Facebook ads and the expanded
`needs of my business.” Ex. I (Ventrice) ¶¶ 3, 13.
` “I do not know where my business would be without Facebook, especially since the start of
`the COVID-19 pandemic. … I could keep in contact and engage with my community and
`direct people to my website, and it really helped my sales.” Ex. C ¶ 5.
`• “What matters to me is actual results—people coming in to my taproom—and Facebook
`delivers good results.” Ex. B ¶¶ 3, 11.
`
`3 Plaintiffs base their UCL and fraudulent misrepresentation and concealment claims on three
`misstatements (collectively, “Potential Reach Statements”): (1) the “targeted” Potential Reach
`estimate provided for each ad set after the advertiser selected targeting and placement criteria
`(“Targeted Statement”); (2) the description of Potential Reach estimates as “people” instead of
`“accounts” (“People Statement”); and (3) a new statement challenged for the first time in the
`Motion, namely, the “default” Potential Reach estimate that appeared when a U.S. advertiser
`opened Ads Manager (“Default Statement”). Mot. at 3-4, 14. Contrary to Plaintiffs’ suggestion,
`the “Default Statement” has not been “uniform” during the Class Period—Plaintiffs reference a
`Default Statement of 240 million (id. at 3), but the so-called “default” was at times 220 million
`(Ex. 84) and 230 million (Dkt. 282-9 ¶ 18; Dkt. 282-6 at Fig. 1). And Plaintiffs admit that the age
`of people included in the “default” varied. Mot. at 3 n.2.
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`
`
`
`
`4
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`
`
`
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 10 of 31
`
`
`Advertisers can select from hundreds of thousands of targeting and placement criteria to
`define the audience they want, based on demographics (e.g., location, age, gender), interests (e.g.,
`bagels), and the platform to display their ads (e.g., Facebook, Instagram). Tadelis ¶¶ 30-32; Tucker
`¶ 15; Ex. 78 at 12; Ex. 89; Ex. 90. As an advertiser selects criteria, the Potential Reach estimate
`updates in real-time to provide an “estimate of the size of the audience that’s eligible to see your
`ad” based on the selected criteria. Ex. 85; Ex. 78 at 13; Ex. 80. Above the numerical Potential
`Reach estimate, a color-coded dial shows whether the target audience “is fairly broad,” “defined,”
`or “too specific.” Ex. 84; Ex. 93; Tucker ¶¶ 64-68. As one advertiser explains, “if I add
`preferences to my targeting criteria and the Potential Reach estimate does not change, then it
`indicates to me it is not a widely held preference.” Ex. B ¶ 8. Advertisers “might dial in the
`audience with additional targeting criteria if it is too broad,” Ex. D ¶ 10, and they recognize the
`value in Potential Reach as a “sanity check or directional guide” in crafting an audience, Ex. B ¶
`8; see also Tucker ¶¶ 63-68; Tadelis ¶¶ 35-38; Porter ¶ 64. But advertisers also understand that, as
`Facebook discloses, Potential Reach is not “an estimate of how many people will actually see your
`ad,” which depends on an advertiser’s budget (and other factors like the quality of the ad). Ex. 80
`at 3; Ex. 87; Ex. H ¶ 6; Ex. C ¶ 9. For that, Facebook provides estimates of actual results (Estimated
`Daily Results), including Estimated Daily Reach, which is right below—and is a fraction of—
`Potential Reach, and takes into account an advertiser’s budget. Ex. 94 (“The number of people
`your campaign actually ends up reaching depends on your budget and performance. Check your
`estimated daily results for that estimation.”); Ex. 88; Tucker ¶¶ 69-72; Tadelis ¶¶ 46-48.
`In declarations, actual advertisers and ad agencies explain that they use Potential Reach (if
`at all) to select their targeting criteria, not to set their budgets:
`• “Sometimes I use Potential Reach while planning an ad campaign to choose preferences. …
`I do not expect to reach anywhere close to [the Potential Reach estimate].” Ex. B ¶ 8.
`• “The only way I use Potential Reach estimates is refining and honing in on the target
`audience.” Ex. I ¶ 9.
` “I am never trying to reach everyone in the Potential Reach estimate. Sometimes I tinker
`with the Potential Reach meter to see how many people may possibly be included in my
`targeting criteria, but I hardly look at it.” Ex. F (Davis) ¶ 6.
`• “I may use the Potential Reach estimate as a guide to help determine whether to make my
`target audience narrower or broader, but Potential Reach has no impact on my budget
`decisions.” Ex. J (Pettys) ¶ 6.
`
`•
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`
`
`
`
`
`
`
`
`5
`
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 11 of 31
`
`
`•
`
`•
`
`6
`
`1 2 3 4 5 6 7 8 9
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`
`
`
`
`
`ATTO RNEYS AT LAW
`SAN F RANCIS CO
`
`
`
`
`
`
`
` Advertisers explain that they set and adjust their budgets based on a variety of information,
`including prior campaign results, real-time actual results that Facebook provides once they launch
`a campaign, and Estimated Daily Results (such as Reach and Link Clicks)—not Potential Reach:
`• “I do not rely on Potential Reach when setting the budget for my ads, and I do not increase
`or decrease my ad budgets based on Potential Reach estimates. I have a tight marketing
`budget and use the funds I have available to advertise.” Ex. A ¶ 8.
`• “The Estimated Daily Reach ranges that Facebook displays are more useful information for
`me and are what I sometimes consider instead of Potential Reach.” Ex. C ¶ 10.
` “We activate ads on Facebook and monitor them to see if they are performing, and then
`make adjustments as warranted by the performance—including while the campaign is
`running, or when placing the next campaign.” Ex. E (Davis) ¶ 10.
` “We adjust our ad campaigns while they are running based on measured ROI, and we
`generally have a good idea of what will work or not based on past performance.” Ex. H ¶ 5.
`Potential Reach is an estimate with constantly changing inputs. Ex. 92; Ex. 93; Ex. 115 at
`68:11-25; 74:9-75:10. Advertisers understand that Potential Reach is necessarily an estimate
`because it is a real-time calculation based on data, including self-reported user data, about users’
`interests, activities, and locations, among others. Ex. B ¶ 11 (“I understand that Potential Reach
`is only an estimate or guideline, and not an exact science.”). There is no “ground truth” for
`Potential Reach, Ex. E ¶ 10, but that does not matter to advertisers because Potential Reach is not
`what they use to measure results, Tucker ¶¶ 50-54, 56-62; Ex. 115 at 76:6-78:1. For example:
`• “If Potential Reach estimates were off from the actual number of people in my target
`audience by 10%, it would not affect my advertising decisions. I understand that Potential
`Reach is only an estimate or guideline, and not an exact science.” Ex. B ¶ 11.
`• “The Potential Reach estimates that Facebook provides are not important to Carvana’s
`advertising strategy, because we care if we are getting conversions, not reach. … Potential
`Reach is fluid on an hourly, daily, and weekly basis … .” Ex. H ¶¶ 5-6.
`• “Facebook’s interest targeting is based on user reported information as well as user behavior
`(such as engagement with certain pages). I will explain to my [agency] clients that, given
`the way interests are determined, the size of an audience represented by a given interest may
`be over or under-inclusive of the universe of people who really hold that interest.” Ex. E ¶ 8.
`B.
`During The Class Period, Facebook Updated The Way Potential Reach Was
`Estimated But Rejected Some Changes Because They Were Flawed
`Plaintiffs argue—based on misconstrued excerpts from documents—that Facebook knew
`Potential Reach was “inflated” but chose not to fix it to make more money. Not true. Facebook
`made numerous updates to Potential Reach during the seven-year Class Period, but it rejected
`changes that would have made Potential Reach less reliable. E.g., Dkt. 282-70 at 10, 15; Dk. 282-
`FACEBOOK’S CORRECTED OPP. TO
`MOT. FOR CLASS CERTIFICATION
`CASE NO. 3:18-CV-04978 JD
`
`
`

`

`Case 3:18-cv-04978-JD Document 363 Filed 12/21/21 Page 12 of 31
`
`
`7
`
`59; Tadelis ¶¶ 38-39, 40-45, 49-67; Porter ¶¶