paint-brush
Google Fails to Undermine the Search Ad and Text Ad Marketby@legalpdf
177 reads

Google Fails to Undermine the Search Ad and Text Ad Market

by Legal PDF: Tech Court CasesAugust 10th, 2024
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

Although some advertisers seek to calculate ROI or ROAS from their digital advertising, and a subset of those consider ROI/ROAS as one factor when allocating spend between channels, this does not render all digital advertising interchangeable or part of the same market, nor does it mean plaintiffs’ advertising markets are unsound.
featured image - Google Fails to Undermine the Search Ad and Text Ad Market
Legal PDF: Tech Court Cases HackerNoon profile picture

United States of America v. Google LLC., Court Filing, retrieved on April 30, 2024, is part of HackerNoon’s Legal PDF Series. You can jump to any part of this filing here. This part is 15 of 37.

D. Google Makes No Attempt To Identify An Alternate Advertising Market And Fails To Undermine The Search Ad And Text Ad Markets

501. Dr. Israel concedes he did not define the boundaries of any alternate advertising market (Tr. 8836:18–21 (Israel (Def. Expert))) and Google’s various attempts to undermine the Search Ads and Text Ads markets cannot overcome the evidence showing those markets.


1. Advertisers’ Use Of ROI/ROAS Metrics And Efforts To Optimize Ad Spend Across Channels Do Not Show That Advertisers View Search Ads And Other Ad Types As Substitutes


502. Although some advertisers seek to calculate ROI or ROAS from their digital advertising, and a subset of those consider ROI/ROAS as one factor when allocating spend between channels, this does not render all digital advertising interchangeable or part of the same market, nor does it mean plaintiffs’ advertising markets are unsound. Tr. 5454:10–5455:15 (Jerath (Pls. Expert)) (to the extent advertisers “attempt[] to measure or use ROI/ROAS of different ad channels[, this] does not mean that these ad channels have really become interchangeable,” discussing UPXD103 at 26); id. 5451:6–5452:14.


503. As Dr. Israel concedes, ROI/ROAS is “not the only thing” advertisers consider and advertisers would not shift all ad spend based on ROI/ROAS; instead he asserts that “partial substitution” based on ROI “is what matters.” 8472:22–8473:20 (Israel (Def. Expert)).


504. As discussed above, advertisers typically maintain separate budgets for Search Ads and other channels and do not shift between them. Tr. 4863:11–4864:9 (Lim (JPMorgan)) (“[W]e know what we need to spend in paid search, and it kind of exists in its own right and is not typically influenced by budgets from other channels.”); Tr. 4856:16–4858:16 (Lim (JPMorgan)) (“Q. . . . And do you typically shift the spend between search text ads and digital display ads based on the -- if the relative cost of those ads would change? A. No. Q. And why is that? A. Paid search budgets are for paid search only. . . . [I]t is not transferable between a programmatic buy across webpages and paid search. They are distinct and different and separate. Q. And distinct in what way? A. They are not fungible. . . .”); Tr. 4862:7–14 (Lim (JPMorgan)) (JPMorgan does not shift spend between search and social channels based on relative cost of ads); Tr. 3960:12–3963:13 (Lowcock (IPG)) (even if Search Ads got more expensive “I’ve not had the experience where we would optimize away from search.”).


505. Some advertisers do seek to calculate ROI and ROAS and consider those calculations when budgeting. But even the subset of advertisers that use ROI/ROAS to aid crosschannel budgeting, recognize the differing purposes of different digital advertising channels. Tr. 5458:13–22 (Jerath (Pls. Expert)) (noting that ROI is “just one input” into advertiser decisions).


When assessing returns, advertisers do not view channels in isolation, but recognize that “more often than not, it’s a combination of everything that you’re doing that’s driving th[e] outcome.” Tr. 4893:8–4894:18 (Lim (JPMorgan)); Tr. 3980:4–3981:2 (Lowcock (IPG)) (“[Y]ou wouldn’t necessarily move out of a channel that—that might individually look like it’s not performing because you know it actually contributed to driving media performance elsewhere.”). The interplay between channels drives budget allocations.


506. Some advertisers relying on “last click” ROI/ROAS recognize the interplay between channels by setting different ROI/ROAS targets for different digital channels, noting upper funnel channels can operate at a lower ROI/ROAS because they drive customers to lower funnel, higher ROI/ROAS channels.


For example, Skechers—an advertiser whose documents both Plaintiffs and Google relied on[13] in support of their ROI/ROAS analysis—sets lower ROAS goals for advertising channels more suited to goals higher in the funnel, noting the importance of “giv[ing] mid funnel channels room to operate at a lower ROAS knowing they drive quality new customers who convert through other channels (brand search, direct, organic, email).” Tr. 5456:22–5458:11 (Jerath (Pls. Expert)) (analyzing UPXD103 at 27 (displaying UPX1017 at -395) and describing Skechers’ use of ROI as “very representative”).


507. Others use ROI to allocate budgets within the paid search channel but use different methods in other channels. Tr. 4863:3–4864:9 (Lim (JPMorgan)) (explaining budget for “paid search” built around ROI or net present value (NPV), while “[b]udgets for other channels are typically built around audience, the audience that you’re trying to reach and how often you want to reach them over what period of time.”).


Still others use so-called “MMM” or “Media Mix Modeling” utilities, which are sophisticated programs that seek to “estimate the effect of various marketing inputs on sales and use these conclusions to recommend more effective distributions of marketing dollars on a more frequent basis.” UPX0519 at .006; UPX0840 at -335 (MMMs “use statistical analysis to estimate the impact of marketing tactics, controlling for base drivers, on sales.”).


MMMs do not simply allocate across channels based on relative ROI, but consider a wide variety of factors, UPX0840 at -343 (illustrating other factors) and rely on extensive data over years involving a variety of inputs. UPX0028 at .041 (“MMMs generally require at least 1 year of data, preferably 2–3 years.”).


508. Even moving spend across ad channels does not necessarily mean ad channels are interchangeable; for example, advertisers could rebalance ad spend as part of a full-funnel advertising strategy.


Tr. 5454:18–5455:15 (Jerath (Pls. Expert)) (“[E]ven if spend is moved across [ad] channels, that does not necessarily mean that [ad] channels are interchangeable, they could still be complementary, but [an advertiser] can just rebalance to a full funnel kind of strategy” (discussing UPXD103 at 26)); id. 5455:16–5456:21 (explaining what he would expect an advertiser to do if it saw a decrease in ROI on Search Ads); id. 5459:7–5462:9 (responding to the Court’s question regarding movement of funds from search to another channel and providing example to illustrate that “money has been moved from search to display, but not because the channels are substitutes or interchangeable, rather, because the channels are complementary and mutually reinforcing”).


509. Advertisers set their budgets for channels for periodic cycles (i.e., quarterly or annually) and cannot quickly shift spend. Des. Tr. 150:8–151:3 (Jain (Google) Dep.) (“Budgets for advertisers are set on, I don’t know, some cycle or so forth. They’re not just instantly reactive”).


510. Practical difficulties prevent many advertisers from shifting spend based on ROI/ROAS, even if they wanted to. The concept of calculating and using ROI/ROAS to assess advertising return is complicated, which is one reason why Google’s internal documents state that “[i]n terms of spend allocation to competition, ROI is not what we should be worried about most.” UPX0519 at .002. Calculating ROI requires measuring the incremental gain from an advertisement, i.e., the amount of revenue generated by the advertisement. Tr. 8863:15–25 (Israel (Def. Expert)).


In the context of a single Search Ad, computing incrementality involves assessing what users would have done absent the Search Ad, i.e., determining, absent the ad, which users would have clicked on an organic link or otherwise navigated to the advertiser’s website. Tr. 8864:9–8866:6 (Israel (Def. Expert)) (“I pay some money on a text ad and I’d like to know if I didn’t have it here, what would they have done. I don’t ever know that perfectly.


You try to measure that by doing A–B experiments with the ad, without the ad and compare, but it’s an issue that needs to be studied.”). It is progressively more difficult for advertisers using multiple channels and multiple ads within each channels; cross-channel ROI comparisons are particularly difficult. UPX0519 at -001 (“Cross channel ROI comparisons appear to still be extremely difficult today, even for major players.”).


511. Thus, correctly calculating ROI/ROAS is a resource-heavy, operationally difficult, and expensive task that is not available to many advertisers. Tr. 5169:2–5170:6 (Booth (The Home Depot)) (outlining the “moderate to significant” internal and external resources, including both monetary and staffing, needed to conduct an incrementality test); PSX00676 at -240 (The Home Depot incremental test resulted in the loss of [redacted]% of revenues in the test markets.); Tr. 3981:7–20 (Lowcock (IPG)) (stating that ROI can be difficult (or impossible) for advertisers to measure); Des Tr. 118:7–119:10 (James (Amazon) Dep.) (“[V]iew-through effects that are difficult for us to measure.


Meaning that a user who is on Facebook might see a Facebook ad but may not be inclined to click on it, but it may actually trigger some form of purchase consideration or brand awareness with Amazon that causes them to make a purchase with Amazon at a later date, which makes it difficult for us to measure the effects of the advertising program . . . .”).


512. Google’s internal documents recognize that “cross channel ROI comparisons appear to still be extremely difficult today, even for major players.” UPX0519 at .001; UPX0506 at .012 (“Overwhelming majority of adv[ertisers] nowhere close to measuring ROI.”).


And, although one of Google’s employees testified (without support) advertisers’ ability to track ROI has improved in recent years, Tr. 1385:3–12 (Dischler (Google)), recent privacy initiatives limiting advertisers’ ability to track customers across channels has actually made it harder. Tr. 5453:4–5454:9 Jerath (Pls. Expert)) (“So to measure ROI, you need to be able to see what consumers do after you have shown them an ad. . . . And as consumer privacy concerns are increasing and consumer trackability is decreasing, it is more and more difficult to figure out that outcome.”); Tr. 8864:1–8 (Israel (Def. Expert)) (acknowledging ROI measurement is a “challenge,” “[d]ifficult to measure perfectly” and that “businesses tend not to perfectly know their ROI.”); Tr. 8864:9–17–8865:6, 8866:14–19 (Israel (Def. Expert)) (conceding one difficulty with advertising ROI is figuring out how much revenue and cost was due to different advertisements and that “which advertisement may have led to the purchase” is “another challenge” and is among “things that need to be studied.”).


513. Although Google presented testimony claiming it offered tools to smaller advertisers helping them calculate ROI, Google described only providing aggregated CPC information and clicks, which is insufficient to calculate incremental ROI. Tr. 7378:18–7379:15 (Raghavan (Google)) (“We can say we sent you so many clicks, here was the average price you paid, and now you go figure the rest, right.”). And Google’s documents recognize that “cross channel ROI comparisons appear to still be extremely difficult today, even for major players.” UPX0519 at .001.


2. Nike’s Boycott Of Social Ads On Facebook/Meta Did Not Impact Search Ads


514. During a period when Nike boycotted Facebook/Meta and purchased no social media ads on its platforms, Nike’s spend on Search Ads remained relatively unchanged compared to Nike’s spend on social ads—(1) non-Meta social spend rose from 1% to 10% of Nike’s advertising spend, dropping to only 7% after the boycott; (2) Search spend rose from 48% to 51%, remaining at 50% after the boycott; and (3) Display rose from 19% to 31%, dropping back down to 21% after the boycott. Tr. 8845:8–13, 8845:22–8846:8, Tr. 8847:13–16 (Israel (Def. Expert)) (discussing UPX2076 at -152, non-Meta social spend rose from 1% to 10% of Nike’s advertising spend, dropping to only 7% after the boycott); Tr. 8848:14–23, 8849:9–15 (Israel (Def. Expert)) (discussing UPX2076 at -152, Search Ads spend rose from 48% to 51%, remaining at 50% after the boycott); Tr. 8849:22–8850:9 (Israel (Def. Expert)) (discussing UPX2076 at -152, display rose from 19% to 31%, dropping back down to 21% after the boycott); Tr. 8850:3–6 (Israel (Def. Expert)) (rise in display ads spend during Nike’s Facebook boycott greater than rise in Search Ads spend).


515. During Nike’s Facebook boycott, constraints on available non-Meta, social ads inventory impeded Nike’s ability to increase spend on other platforms, causing shrinkage in Nike’s overall social ads spend, with no corresponding significant increase in Search Ads spending. Tr. 8846:17–24 (Israel (Def. Expert)) (“During the pause, social investment shrunk by more than half of just 8 percent of overall PM budget allocation due to limitations on scale opportunities within SnapChat, Pinterest, and Twitter.”); Tr. 8848:14–19, 8849:9–15 (Israel (Def. Expert)) (discussing UPX2076 at -152, Search Ad spend rose from 48% to 51%, remaining at 50% after the boycott).


516. [INTENTIONALLY LEFT BLANK]


3. Google Markets “Performance Max” As A Complement To Text Ads And Repeatedly Admitted It Cannot Replace Text Ads



517. Performance Max is a relatively new Google product which automatically constructs and manages advertising campaigns across all of Google’s advertising channels (i.e., search, display, Discovery, Gmail, and others). Tr. 1371:25–1372:24, 1478:21–25 (Dischler (Google)); Tr. 7375:1–17 (Raghavan (Google)) (explaining DXD-21 at .018); UPX2079 passim (Google Ads Help: “About Performance Max campaigns”); UPX0028 passim (Google document for launch of Performance Max); Tr. 1289:1–12, 1480:20–1481:23 (Dischler (Google)) (Performance Max automatically creates and places ads on all Google properties).



518. As Jerry Dishler, Google’s former head of ads, acknowledged, Performance Max is not a replacement for Text Ads or shopping ads. Tr. 1479:16–1480:19 (Dischler (Google)) (“Performance Max is not a replacement for text ads or shopping ads, correct? A. That’s correct”). As both Mr. Dischler and Dr. Israel conceded, Performance Max is technologically incapable of replacing Google’s Search Ads products, and Google therefore views it as complementary to its other Search Ads products. Tr. 1480:2–16 (Dischler (Google)); Tr. 8820:24–8821:19, 8825:15–8826:1 (Israel (Def. Expert)) (“[T]he artificial intelligence behind Performance Max is not developed to the point where it could take over all campaigns.”).


519. Google instructs advertisers and its sales personnel that Performance Max does not replace, but complements, Text Ads and other Search Ads, recommending advertisers purchase both Text Ads and Performance Max. Tr. 8820:24–8821:19 (Israel (Def. Expert)) (responding to UPX2079 at -494, “Performance Max . . . is designed to complement your keyword-based Search campaigns.”); UPX0028 at .026 (“We recommend adding Pmax as a complementary to search keywords campaign.”) (emphasis omitted); id. at .047 (“[I]t is highly recommended that ‘performance budgets’ do not cannibalize Google Ads Search budgets as Performance Max is not a substitute for Google Search campaigns.”) (emphasis omitted); UPX0931 at -075 (internal presentation instructing Google personnel to sell Performance Max “[a]s a complement to KW-targeted Search”).


520. Performance Max users cannot select keywords to target Search Ads. Tr. 8827:20–8828:24 (Israel (Def. Expert)) (“You can’t go into Performance Max and say, I want the following keywords”); UPX0028 at .023 (“Performance Max does not have keywords . . . .”).


521. Google intends to continue selling Text Ads and shopping ads in addition to Performance Max. Tr. 1480:12–19 (Dischler (Google)).


Continue Reading Here.


About HackerNoon Legal PDF Series: We bring you the most important technical and insightful public domain court case filings.


This court case retrieved on April 30, 2024, storage.courtlistener is part of the public domain. The court-created documents are works of the federal government, and under copyright law, are automatically placed in the public domain and may be shared without legal restriction.

[13] Tr. 8472:22–8473:4 (Israel (Def. Expert)). Google did not enter the document discussed by Dr. Israel into evidence.