Source-linked AI summary
Inefficiencies in Digital Advertising Markets
Brett R Gordon, Kinshuk Jerath, Zsolt Katona, Sridhar Narayanan, Jiwoong Shin, Kenneth C Wilbur
TL;DR
Digital advertising markets contain poorly understood inefficiencies that may hinder allocative efficiency, including unreliable incremental-effect measurement and intermediary frictions. The paper selectively reviews academic findings, practical developments, and promising future research on four digital advertising market inefficiencies. The review identifies difficulty measuring incremental ad effects and intermediary incentives and private information as recurring sources of allocative inefficiency.
Problem
Digital advertising markets contain poorly understood inefficiencies that may hinder allocative efficiency, including unreliable incremental-effect measurement and intermediary frictions.
Method
The paper selectively reviews academic findings, practical developments, and promising future research on four digital advertising market inefficiencies.
Results
The review identifies difficulty measuring incremental ad effects and intermediary incentives and private information as recurring sources of allocative inefficiency.
Takeaways & Limitations
Digital advertising inefficiencies arise in distinctive ways from measurement problems and the structure of channels containing numerous intermediaries.
Takeaways & Limitations
Fraud measurements warrant skepticism because intention is not directly observable in ad data, and detection involves tradeoffs between false positives and false negatives.
Abstract
from arXiv · showhide
Digital advertising markets are growing and attracting increased scrutiny. This paper explores four market inefficiencies that remain poorly understood: ad effect measurement, frictions between and within advertising channel members, ad blocking and ad fraud. These topics are not unique to digital advertising, but each manifests in new ways in markets for digital ads. We identify relevant findings in the academic literature, recent developments in practice, and promising topics for future research.
Digital Advertising Effect Measurement
Digital ad effect measurement seeks the incremental effect of advertising, but connecting exposures to outcomes and identifying causal effects remains difficult. The paper reviews experimental and observational approaches alongside challenges from strategic behavior, complex effects, and incomplete data.
- Measurement challenges: Most marketers either do not or cannot measure incremental ad effects reliably, despite abundant and granular digital data.Advertisers often cannot connect individual-level ad exposures to outcomes or link clicks and likes to sales and profits.
- Measurement challenges: Strategic advertiser, platform, and consumer behavior confounds the relationship between ad exposure and outcomes.Targeting responds to expected demand, platforms optimize delivery toward likely clickers, and consumers differ in attention; experiments may therefore recover intent-to-treat effects rather than treatment effects.
- Measurement challenges: Ad effects can be nonlinear and depend on exposure frequency, competitors, other media, and unobserved exposures, limiting measurement accuracy.Wear-in, wear-out, weariness, cross-media effects, and incomplete exposure measurement complicate a fully accurate view of campaign effects.
- Measurement approaches: Randomized experiments create exogenous variation that addresses strategic behavior, while observational methods remain feasible when experimentation is impractical.The paper distinguishes ex ante experimental designs from ex post observational analyses and notes that observational strategies require extra work to identify effects.
- Evidence from experiments: Experiments can overturn attribution-based conclusions: eBay’s branded-search ads produced a ROAS of negative 63% when paid traffic substituted toward organic search.The result depended on eBay’s highly ranked organic links and infrequent competitor keyword ads; other studies found weaker generalization across firms.
- Evidence from experiments: Large-scale advertising experiments often remain imprecise, with most ROAS confidence intervals wider than 100% and the smallest exceeding 50%.The review also reports that competent, expensive, large-scale attempts have failed to measure effects with reasonable precision.
Organizational Frictions and Inefficiencies
Digital advertising creates organizational inefficiencies when incentives and information differ within firms, across agencies, and throughout the advertising supply chain. These frictions can distort measurement, bidding, coordination, service quality, and spending.
- Organizational Frictions and Inefficiencies: Organizational inefficiencies arise both within firms and between marketing organizations, agencies, complementors, and other advertising-market participants.Within firms, officers, departments, and business units may pursue different incentives or objectives; across firms, agencies and supply-chain participants may face conflicting interests.
- Intra-firm Inefficiencies: Managers may present inflated lift metrics because retargeting exposes ads to consumers already more likely to purchase, creating correlation without causal identification.This can support self-interested reporting when firms lack credible causal metrics.
- Intra-firm Inefficiencies: Short-run campaign outcomes are easier to link to exposure than long-run brand effects, creating tension between measurable responses and longer-term firm objectives.The paper notes that reliable estimates of advertising effects on long-run outcomes such as brand attitudes are seldom possible.
- Intra-firm Inefficiencies: Internal functional groups can undermine advertising performance when procurement minimizes expenditure, marketing pursues different objectives, or affiliated brands bid against one another.Low bids may reduce service quality, while rivalrous bidding can occur among business units competing for the same keyword inventory.
- Inter-firm Inefficiencies: Intermediaries consume 42 cents of each programmatic advertising dollar in supply-chain data and transaction fees, nearly triple the traditional 15% agency commission.The 58-cent remainder purchases media inventory and audience exposure from publishers.
- Inter-firm Inefficiencies: Agency relationships also create moral-hazard and asymmetric-information risks because markups may be obscured and agencies may arbitrage low-cost inventory.Agencies counter that analytics can add value to inexpensive inventory, so the efficiency effect is not unambiguously negative.
- Inter-firm Inefficiencies: Advertisers often fail to coordinate specialist agencies even though advertising placements generate spillovers across traditional media, search, websites, sales, and social activity.Independent channel specialists compete for larger shares of the budget, limiting integrated allocation across media.
Ad Blocking
Ad blocking changes who can access, monetize, and control advertising-supported content, while creating conflicting incentives for platforms, publishers, advertisers, and consumers. Its overall effects remain unclear and motivate research on consumer behavior, market structure, and business models.
- Ad Blocking: Ad blocking threatens advertising-supported content because sought-after consumer segments are often especially likely to install blockers.Some blockers prevent ad requests, while more aggressive tools block after requests and can still waste advertiser spending.
- Ad Blocking: Platforms that sell ads face conflicting incentives because revenues depend both on consumer experience and advertising exposure.Browsers and operating systems can enable or restrict blocking, while mobile-platform rules make blocking in-app ads difficult.
- Ad Blocking: Publishers respond to blocking by detecting blockers, withholding content, engineering around them, requesting whitelisting, or using blocker-blocker-blockers.These responses create an ongoing contest over whether ads can be displayed and whether content remains accessible.
- Ad Blocking: 30% of the top 10,000 websites detect ad blockers, illustrating that blocking and countermeasures are widespread among major sites.The paper describes publisher investments in circumvention and requests for users to selectively disable blocking.
- Ad Blocking: The overall effect of ad blocking on advertisers, consumers, and publishers is unclear, with competing possibilities ranging from lower publisher quality to consumer benefits.The paper calls for evidence on usage, subscriptions, prices, ad placement, product-market outcomes, and advertiser responses.
- Ad Blocking: Ad blockers commonly monetize whitelisting by charging publishers, with large publishers typically paying 30% of otherwise blocked ad revenue.The German Supreme Court ruled that ad blocking and soliciting payment for whitelisting are legal.
Digital Advertising Fraud
Digital advertising fraud misrepresents inventory, audiences, delivery, or performance, often through machine-generated activity and deceptive supply-chain practices. The paper emphasizes its scale, measurement difficulty, and open questions about market design, detection, privacy, and regulation.
- Digital Advertising Fraud: Botnets make machine activity difficult to distinguish from human behavior by directing infected computers to mimic organic browsing and ad interactions.The 3ve botnet indictment described more than 1.7 million infected computers loading fabricated webpages and ads.
- Digital Advertising Fraud: 28% of global mobile media budgets were reported as wasted on fraud, and advertisers reported cutting spending because of inaccurate or false reporting.These figures indicate substantial reported economic exposure, although the paper cautions that fraud estimates require skepticism.
- Digital Advertising Fraud: Fraud can involve publishers, advertisers, intermediaries, or individuals misrepresenting inventory, manipulating auctions, claiming conversion credit, or fabricating influence and clicks.The paper identifies six basic motivations, including harming rivals, altering prices or commissions, and falsifying audience or performance measures.
- Digital Advertising Fraud: Fraud measurement is limited because intention is unobservable, so indirect detection trades off false positives against missed valid exposures.Detection firms may over-report fraud to attract business, while sellers may under-report it to reassure clients.
- Digital Advertising Fraud: Fraudulent video inventory exceeded truly available inventory by a factor of 30, while 72% of ads purchased falsely claimed to occur on Guardian.com.Fraudulent Financial Times inventory was offered across 25 ad exchanges, demonstrating that inventory misrepresentation can propagate through marketplaces.
Broader Discussion
The paper argues that digital-advertising inefficiencies interact and should be evaluated through consumer, marketer, and industry outcomes. It identifies major evidence gaps around causal measurement, incentives, regulation, and the trade-offs among market participants.
- Interactions between Inefficiencies: Agencies may misreport ad blocking or fraud when negative information could reduce client budgets and therefore their commissions.This links measurement and organizational incentives to the broader market inefficiencies discussed throughout the paper.
- Interactions between Inefficiencies: Ad blocking and ad fraud can reduce valid advertising exposures in treatment and control groups, worsening the statistical challenge of measuring ad effects.The paper treats this as a prominent interaction among otherwise distinct inefficiencies.
- Policy-Relevant Research Opportunities: Consumer welfare and producer profits are the two primary concerns for government policy because digital advertising connects consumers and marketers through intermediaries.Consumers supply attention and receive utility from content, ads, and products, while marketers seek profits from product sales.
- Consumer-level questions: More comprehensive evidence is needed because consumers, advertisers, and advertising channels are heterogeneous, especially when assessing advertising’s effects on search, competition, and consumer utility.The paper connects advertising avoidance and ad-blocker use with possible negative effects on consumer utility while noting that some advertising may improve product-market outcomes.
- Marketer-level questions: Little public evidence documents the financial returns to data-driven advertising decisions or tools such as automated content generation, programmatic buying, fraud detection, and brand-safety monitoring.The paper calls for research on advertiser behavior, agency structures, and contracting mechanisms across firms with different capabilities and ownership structures.
- Industry-level questions: Digital advertising regulation must trade off conflicting goals across players, including consumer privacy versus fraud detection and incumbent protection versus entry and innovation.The paper notes that regulation may be vulnerable to platform preparation and adversarial interference, potentially worsening the problems it seeks to resolve.
Conclusion
Digital advertising creates benefits but also four recurring issues that can hinder allocative efficiency: measurement uncertainty, intermediary frictions, ad blocking, and fraud. The survey reviews these inefficiencies to support further research and better-informed policy discussions without taking a regulatory position.
- Conclusion: Most marketers do not or cannot measure incremental ad effects, and this uncertainty may distort demand for advertising.The paper treats incremental-effect measurement as a central market-efficiency problem.
- Conclusion: Numerous intermediaries separate marketers from publishers, with each taking a cut while holding private information and incentives.These channel structures create frictions between market participants.
- Conclusion: Ad blocking can misappropriate advertising revenues and reduce incentives to provide media content.Consumers passively prevent advertisements from being displayed, shifting consequences beyond advertisers and publishers.
- Conclusion: Advertising fraud misrepresents advertising opportunities and directs exposures to machines to steal advertising budgets.The paper identifies fraud as another mechanism that undermines the allocation of advertising opportunities.
- Conclusion: The survey focuses on four issues because they are less well understood, while recognizing that antitrust, privacy, transparency, and related concerns also matter.The authors do not claim these four issues are necessarily more important than other policy considerations.
- Conclusion: The authors do not take a position on whether or how digital advertising markets should be regulated, but hope the survey informs research and policy making.They present the article as a contribution to scientific literature and policy information rather than a regulatory recommendation.