Novetics

The Google Search Fallacy: Why Search-Result Co-Appearance Does Not Establish Competition, Exposure, or Consumer Confusion

Home/Research Insights/The Google Search Fallacy: Why Search-Result Co-Appearance Does Not Establish Competition, Exposure, or Consumer Confusion

Parties to trademark disputes sometimes treat the appearance of two brands in response to the same search query as a sign that something harmful is happening, for example, that the products compete, that consumers regularly encounter both in the market, or that consumers are likely to be confused. This paper argues that co-appearance is an ordinary output of search systems and establishes, at most, an opportunity for exposure. Organic rankings reflect many signals whose contributions cannot be observed from a results page, and paid placement and automated ad matching add further routes by which unrelated or competing items come to appear together. A results page reflects one outcome for one user at one moment, and research on attention shows that many listings receive little notice. Consumer research treats confusion as a mistaken belief about source, sponsorship, or affiliation, which is distinct from similarity, attention, or a click. This research insight extends the analysis to retail platforms such as Amazon and identifies the evidence needed before co-appearance can support conclusions about competition, exposure, or confusion. Co-appearance alone does not establish competition, meaningful exposure, or consumer confusion.

Introduction

Search engines and retail platforms routinely place unrelated, related, and competing items side by side. When two brands appear in response to the same query, the co-appearance is sometimes offered as evidence that the products compete, that consumers encounter both, or that consumers are likely to be confused about who stands behind them. A shared results page supports a much narrower conclusion. It shows only that an opportunity for exposure existed.

Establishing harm requires considerably more. The products must be purchasing alternatives for relevant buyers. Those buyers must actually see both listings. They must form a mistaken belief about source, sponsorship, or affiliation, and that belief must persist long enough to affect what they consider, choose, or think of a brand. Each of these links is a separate empirical question, and the fact of co-appearance answers none of them.

This research insight proceeds in five parts. The first explains why co-appearance does not show that two products are purchasing alternatives, particularly when the reasons for their placement cannot be determined. The second treats paid placement as a further mechanism that brings items together on a page and considers what consumer responses to it do and do not show. The third distinguishes potential exposure from actual attention. The fourth draws on consumer research on confusion to explain why seeing two brands together does not establish a belief in common source, sponsorship, or affiliation. The fifth applies the analysis to retail platforms, using Amazon as an example. The conclusion identifies the evidence needed at each stage.

Co-Appearance Does Not Establish That Products Are Purchasing Alternatives

Search engines organize information according to its estimated relevance to a query. They do not assess whether two products satisfy the same need or compete for the same purchase. Google describes the main considerations in its ranking systems as the meaning of the query, the relevance and quality of content, the usability of pages, and contextual factors such as the searcher’s location (Google, n.d.-a). A search for “jaguar” may return results about both automobiles and wildlife. Relevance to a query is a different property from substitutability in a market. Goldman (2005) observed that searchers pursue heterogeneous objectives. Search results alone therefore cannot establish that two products share buyers, uses, price points, or purchasing occasions. Nor do they reveal what the searcher intends to buy. Claims of competitive proximity require independent market evidence.

Several different mechanisms can produce the same apparent proximity. Google describes systems that interpret language, evaluate links between pages, and respond to the freshness of content, among other signals (Google, n.d.-b). Testimony in the federal antitrust litigation against Google (United States v. Google LLC, 2024) further revealed that Google’s search-ranking algorithms incorporate aggregated user-interaction data, including information about which search results users click on and how they subsequently interact with search results. These behavioral patterns, collected across large numbers of users and searches, help Google evaluate the relevance and usefulness of webpages when determining search rankings (Search Engine Land, 2023). 

Two results may appear together because they share a word, because their pages contain related content, because of the searcher’s geographic context, or because of other users’ past behavior. None of these mechanisms necessarily indicates that the brands compete.

Drawing conclusions is further complicated by uncertainty about the factors influencing search rankings and by changes to ranking algorithms over time. Burrell (2016) traced algorithmic opacity to three sources: deliberate corporate secrecy, technical complexity, and the inherent difficulty of interpreting the outputs of machine-learning systems. Ranking systems also change frequently. Google reports that it ran more than 700,000 experiments and launched 4,781 changes to Search in 2023 (Google, n.d.-d). A screenshot captures one set of results at one moment. It should not be construed as a stable classification of the brands as competitors.

Paid Placement Is Another Route to Co-Appearance

Organic ranking is not the only reason two items appear together. Sponsored listings appear when an advertiser targets a keyword and the search engine selects the ad for display. Google states that broad match, the most expansive of its keyword matching options, may show ads on searches that relate to the advertiser’s keyword, whereas phrase match and exact match are limited to searches that include or share the keyword’s meaning (Google, n.d.-c). As a result, an appearance on a search results page may occur in response to a query that the advertiser did not specifically select. Paid placement thus provides another mechanism for co-appearance. Like organic ranking, it reveals little about why a particular pairing occurred.

Co-Appearance Does Not Mean Consumers See Both Brands

Display is also not the same as attention. Pan et al. (2007) and Schultheiss et al. (2018) both found strong position effects in eye-tracking studies of Google users, with higher-ranked results receiving a disproportionate share of examination and selection. In a review of 41 eye-tracking studies, Lewandowski and Kammerer (2021) found that viewing behavior also depends on the search task, the design of the page, and characteristics of the individual user. A listing near the bottom of a page, or below the portion visible without scrolling, may never be noticed.

Rank can influence examination independently of preference. Ursu (2018) analyzed a field experiment in which Expedia randomly varied the order of hotel listings. Placement affected whether consumers examined a hotel, but once a hotel had been examined, its rank did not independently affect whether it was booked. Search placement can thus change a consumer’s opportunity to evaluate a brand without showing that the consumer would choose it over another.

The composition of modern results pages further complicates any inference about exposure. Advertisements, maps, images, videos, and AI-generated summaries all compete with conventional organic listings for attention. Search history and other personalization signals may also affect the results shown to a particular user. 

Seeing Both Brands Does Not Establish a Belief in Common Source, Sponsorship, or Affiliation

Confusion is a mistaken belief about source, sponsorship, or affiliation. It is related to, but distinct from, perceived similarity, association, attention, and preference. Warlop and Alba (2004) distinguished these concepts in their study of package imitation, cautioning that similarity and substitutability are not the same thing. Two listings can appear together, look somewhat alike, and draw the same consumer’s attention without that consumer believing they come from the same place. Users ordinarily understand a results page as an index of potentially relevant information. They do not necessarily interpret adjacent listings as sharing an owner. Search engines routinely display unrelated organizations next to one another, and treating that proximity as a statement of affiliation confuses algorithmic relevance with consumer interpretation (Goldman, 2005).

Courts have directed attention to the actual presentation and to the relevant consumer. In Network Automation, Inc. v. Advanced Systems Concepts, Inc. (2011), the Ninth Circuit held that the parties’ shared use of the Internet as a marketing channel shed little light on the likelihood of confusion. In Multi Time Machine, Inc. v. Amazon.com, Inc. (2015), the same court asked what a reasonably prudent online shopper would understand from the product names, manufacturers, and images displayed on the screen.  Two organic results may appear one after the other while prominently identifying different owners, displaying distinct logos, and describing unrelated products. The fact that Google or another search engine returned both offers little basis for inferring a shared source. Misleading labels or presentation could, by contrast, change how consumers interpret a particular page. The relevant inquiry concerns what the consumer actually sees and reasonably understands, and the mere presence of two names on the same screen does not answer it.

Application to Retail Platform Search: The Case of Amazon

The analysis above is framed around Google, but the same problems arise whenever a search system determines which items appear together. Retail marketplaces, travel sites, and app stores all return ranked results in response to typed queries. Amazon is a useful illustration because its search results have been the subject of litigation, regulatory action, and empirical research. Retail search differs from general web search in several respects, and those differences have implications for the analysis.

A retail platform returns products rather than web pages, and it is designed to show shoppers items they might buy instead of, or in addition to, the one they searched for. In Multi Time Machine, Inc. v. Amazon.com, Inc. (2015), for example, a search for a watch brand that Amazon did not carry returned competing watches from other manufacturers. Co-appearance on a retail platform can therefore come closer to a judgment about substitutability than co-appearance on Google. That assessment is made by the platform, however. It reflects the platform’s prediction of what a shopper might click or buy, not direct evidence that consumers regard the two products as alternatives, and the prediction is shaped by the platform’s own commercial interests.

Two features of retail search illustrate these interests. First, paid and negotiated placements are woven directly into the results. Amazon tells sellers that Sponsored Products ads appear in shopping results and on product detail pages, and that under automatic targeting it matches ads to shoppers’ queries using “multiple default strategies that we use on your behalf” (Amazon Ads, n.d.). One of those strategies shows an ad to shoppers viewing products that complement, rather than compete with, the advertised item. Second, the platform may favor its own products. Farronato et al. (2023), using search data collected from a panel of consumers, found that Amazon-branded products were ranked higher than observably similar products, with a prominence equal to 30% to 60% of that given to sponsored products. The Federal Trade Commission has also alleged that Amazon replaced relevant organic results with paid advertisements and increased the number of ads that worsen search quality (Federal Trade Commission v. Amazon.com, Inc., 2023). These are allegations rather than findings, but they illustrate that placement on a retail platform reflects advertising revenue and platform strategy as well as relevance.

The exposure limitations discussed earlier also apply. The products shown in response to a query can change with inventory, price, delivery options, and advertising budgets, so a single capture reflects one moment. The Federal Trade Commission’s complaint asserts that more than 70% of Amazon shoppers do not click past the first page of search results (Federal Trade Commission v. Amazon.com, Inc., 2023). Ursu’s (2018) finding that rank determines which products consumers examine came from Expedia, another platform that ranks purchasing options. A product that appears on a later page, or below a block of sponsored listings, may never be seen.

Amazon’s Brand Analytics reports, available to brand owners enrolled in its Brand Registry program, list the queries customers use to find a brand, together with query volume, impressions, clicks, cart additions, and purchases (Amazon, n.d.). Researchers have used product search data from Amazon to estimate consumer search and demand and to address questions of market structure and competition (Kim et al., 2010). These measures provide more direct evidence of consumer search and purchasing behavior, although establishing substitution between two products may require further analysis. A screenshot showing that both products appeared in the same set of results does not. The same reasoning extends to any platform that ranks results. On each of them, a co-appearance reflects the operator’s ranking system, which combines relevance, commercial objectives, and paid placement in ways an outside observer cannot see.

Conclusion

Appearing together in a search is an ordinary event. Organic ranking, paid placement, automated ad matching, personalization, and platform strategy can bring unrelated, related, and competing items onto the same page. A shared results page can be a useful starting point for investigation, but it shows only that two listings appeared together in one search environment at one moment.

Establishing harm requires evidence of additional events. The products must be alternatives for relevant buyers, those buyers must see both, they must form a mistaken belief about source, sponsorship, or affiliation, and that belief must affect consideration, choice, or brand evaluation. Consumer research and field evidence show that each of these links can fail, and that clicks and diversion are not the same as confusion. Exhibit 1 sets out what co-appearance establishes at each link and the evidence that would be needed to go further.

Exhibit 1

What Co-Appearance Establishes and the Evidence Needed to Go Further

Proposed inferenceWhat co-appearance establishesEvidence needed to support the inference
The products are purchasing alternativesBoth listings were judged relevant to the query in one renderingOverlapping uses, buyers, channels, and price points, together with evidence of cross-shopping, substitution, or switching
Relevant consumers saw both brandsBoth listings could have been encountered on that pageRepresentative queries, variation by device and location, ranking and impression data, attention and click data, and the composition of the searching audience
Consumers believed in a common source, sponsorship, or affiliationNo direct evidence of consumer beliefControlled consumer research that preserves the presentation context, separate measurement of ad recognition and source attribution, and credible evidence of actual confusion
The belief persisted and caused harmNothing about later consideration, choice, or brand evaluationEvidence that a mistaken belief changed what consumers considered or chose, and brand evaluation measures compared against matched controls

The underlying distinction is between what an algorithm displays and what consumers do, consider, and believe. Without evidence that bridges that gap, the appearance of two brands in the same search cannot, by itself, establish competition, exposure, or consumer confusion.

References

  1. Amazon. (n.d.). Amazon Brand Analytics. Retrieved November 19, 2026, from https://sell.amazon.com/tools/amazon-brand-analytics
  2. Amazon Ads. (n.d.). Targeting with Sponsored Products. Retrieved November 19, 2026, from https://advertising.amazon.com/library/guides/targeting-with-sponsored-products
  3. Burrell, J. (2016). How the machine “thinks”: Understanding opacity in machine learning algorithms. Big Data & Society, 3(1). https://doi.org/10.1177/2053951715622512
  4. Farronato, C., Fradkin, A., & MacKay, A. (2023). Self-preferencing at Amazon: Evidence from search rankings. AEA Papers and Proceedings, 113, 239–243. https://doi.org/10.1257/pandp.20231068
  5. Federal Trade Commission v. Amazon.com, Inc., No. 2:23-cv-01495 (W.D. Wash. filed Sept. 26, 2023).
  6. Golden, J., & Horton, J. J. (2021). The effects of search advertising on competitors: An experiment before a merger. Management Science, 67(1), 342–362. https://doi.org/10.1287/mnsc.2019.3534
  7. Goldman, E. (2005). Deregulating relevancy in Internet trademark law. Emory Law Journal, 54(1), 507.
  8. Google. (n.d.-a). How results are automatically generated. Retrieved November 19, 2026, from https://www.google.com/search/howsearchworks/how-search-works/ranking-results/
  9. Google. (n.d.-b). A guide to Google Search ranking systems. Google Search Central. Retrieved November 19, 2026, from https://developers.google.com/search/docs/appearance/ranking-systems-guide
  10. Google. (n.d.-c). Improve with AI-powered Search. Google Ads Help. Retrieved November 19, 2026, from https://support.google.com/google-ads/answer/12159290
  11. Google. (n.d.-d). Improving Search with rigorous testing. Retrieved November 19, 2026, from https://www.google.com/search/howsearchworks/how-search-works/rigorous-testing/
  12. Kim, J. B., Albuquerque, P., & Bronnenberg, B. J. (2010). Online demand under limited consumer search. Marketing Science, 29(6), 1001–1023. https://doi.org/10.1287/mksc.1100.0574
  13. Lewandowski, D., & Kammerer, Y. (2021). Factors influencing viewing behaviour on search engine results pages: A review of eye-tracking research. Behaviour & Information Technology, 40(14), 1485–1515. https://doi.org/10.1080/0144929X.2020.1761450
  14. Multi Time Machine, Inc. v. Amazon.com, Inc., 804 F.3d 930 (9th Cir. 2015).
  15. Network Automation, Inc. v. Advanced Systems Concepts, Inc., 638 F.3d 1137 (9th Cir. 2011).
  16. Pan, B., Hembrooke, H., Joachims, T., Lorigo, L., Gay, G., & Granka, L. (2007). In Google we trust: Users’ decisions on rank, position, and relevance. Journal of Computer-Mediated Communication, 12(3), 801–823.
  17. Schultheiss, S., Sünkler, S., & Lewandowski, D. (2018). We still trust in Google, but less than 10 years ago: An eye-tracking study. Information Research, 23(3), Article 799. https://informationr.net/ir/23-3/paper799.html
  18. Search Engine Land. (2023). How Google Search and ranking works, according to Google’s Pandu Nayak. https://searchengineland.com/how-google-search-ranking-works-pandu-nayak-435395
  19. United States v. Google LLC, 747 F. Supp. 3d 1 (D.D.C. 2024).
  20. Ursu, R. M. (2018). The power of rankings: Quantifying the effect of rankings on online consumer search and purchase decisions. Marketing Science, 37(4), 530–552. https://doi.org/10.1287/mksc.2017.1072
  21. Warlop, L., & Alba, J. W. (2004). Sincere flattery: Trade-dress imitation and consumer choice. Journal of Consumer Psychology, 14(1–2), 21–27.