Precision Paradox: The Complex Reality of Targeting

Iman Ahmadi, PhD, Nadia Abou Nabout, PhD, and Bernd Skiera, PhD

Targeting specific audience segments has emerged as an important practice in today’s online advertising strategies. For advertisers, the primary appeal of online ads is their capacity to target users based on characteristics such as user demographics and online behavior. Recent research supports the notion that targeted ads are more effective, with product users more likely to click. However, as advertisers increasingly pursue targeting, one fundamental question looms—who to target? The answer seems to divide the industry into two camps: Those who believe in avoiding targeting at all costs and those advocating microtargeting, which involves targeting very narrow audience segments to maximize relevance and engagement. Our interviews with industry experts—whose jobs entail setting up targeted ad campaigns—underline this controversy as they reveal diverse approaches, from favoring narrow behavioral segments to targeting broader groups and letting platforms like Facebook optimize.

Too Many Options!

Advertisers also face the problem of having too many options. Major ad platforms such as Google and Facebook offer advertisers a huge range of audience segments for targeting (e.g., males, new dads, stroller enthusiasts). In 2023, Facebook offered more than 600 audience segments for targeting, including location-, demographic-, interest-, and behavior-based segments.¹ And it does not stop here! Advertisers can either combine these segments to broaden their audience pool (e.g., males OR new dads OR stroller enthusiasts) or target users at the intersection of multiple segments (males AND new dads AND stroller enthusiasts) for a more niche audience.

In an ideal world, each segment would undergo rigorous testing through randomized controlled trials before determining which segment to keep. However, given the large number of segments (and their combinations), such exhaustive testing is often impractical, as advertisers have time and budget constraints.

Profitability Challenge: Targeting without Testing?

We believe advertisers should target not only segments that are profitable but also those that produce greater profits than untargeted campaigns. Without this assurance, it is better for the advertiser to just not target at all because targeting impacts profit in multiple ways: It often results in extra data costs and reduces the number of reachable users. However, careful targeting ideally improves campaign performance metrics such as click-through rate (CTR) and conversion rate (CR). These opposing effects make it hard to predict an audience segment’s profitability before testing it in a real-life campaign.

We, therefore, suggest that advertisers should calculate the break-even performance (i.e., minimum lift in CTR, CR, and long-term margin per conversion) of all suitable audience segments before testing them in their campaigns.² This break-even performance ensures that targeting is at least as profitable as no targeting. It thereby equips advertisers with a tool to overcome the challenge of selecting audience segments at the campaign start, thus addressing the perennial "cold-start" problem.

The basic idea of calculating the break-even performance is as follows: Imagine a fitness club in NYC aiming to target Spotify listeners in the city who are interested in fitness. Out of a total of 1,000,000 users, targeting narrows this audience down to 50,000 fitness enthusiasts. We assume that advertisers have access to the necessary inputs for the break-even calculation, including performance data for untargeted audiences. Such data may come from past campaigns or be estimated based on expert knowledge and should include metrics like CTR, CR, and average margin per conversion. Regarding cost-per-thousand (CPM), most platforms like Spotify Ad Studio provide CPM information for both targeted and untargeted campaigns during setup. Advertisers also tend to have a strong understanding of data costs.

Without targeting, let us assume the advertiser sees a CTR of 1%, CR of 2%, and an average margin per conversion of $250. These values result in a performance value of $0.05 per user ((CTR×CR×margin)_untargeted = $0.05), leading to an estimated revenue of $50,000 for the untargeted campaign. The CPM_untargeted is $10, equivalent to 1 cent per user, resulting in a profit of $40,000 (or 4 cents per user).

Now, let us assume CPM_targeted is $14 (total targeted campaign cost of $700 = $14×50,000/1,000). Then, targeting NYC fitness enthusiasts needs to provide a revenue of 81.4 cents per user to match the untargeted campaign’s profit ((CTR×CR×margin)_targeted = ($40,000 + $700)/50,000 = $0.814). Assuming targeting lifts CTR, CR, and margin equally, each requires a 150% lift to make targeting at least as profitable as no targeting.

Is such a lift realistic? The literature on targeting indicates that achieving a lift in CTR of over 100% is, in fact, rare. Our model also reveals that the break-even performance must increase non-linearly as audience reach decreases. In other words, a targeted audience segment’s break-even performance must increase “exponentially” with a reduction in its reach when targeting narrower audiences. This large effect of reach on break-even performance indicates that very narrowly targeted audiences are unlikely to be profitable.

Thus, we recommend prioritizing the testing of audience segments where the required lift is within a realistic and achievable range. This approach streamlines advertisers’ choices, facilitating more informed decisions for setting up randomized control trials.

Is Narrow Targeting Overdone?

Our model also helps us to derive crucial insights. In a study that utilizes a unique dataset from Spotify Ad Studio for targeted audio campaigns, we found that the majority of audience segments offered on Spotify will require the CTR to double compared to an untargeted campaign. Such a lift is too high for most ad campaigns and exceeds what is practical for many targeted ad campaigns, as documented by the literature on targeted advertising. Notably, targeting narrow segments (≤ 5% reach) will most likely be unprofitable. The rationale? The performance boost needed to compensate for the loss in reach is too large (about 2.5 times the CTR of an untargeted campaign).

Bad news: Segment composition is not always 100% accurate.³ For instance, think of a “male” segment that inaccurately includes 40% females; this inaccuracy would inflate its true reach. Although our model allows us to account for these inaccuracies, poor data quality disproportionately affects narrow segments since the required break-even performance increases more for narrow segments compared to broader ones (assuming the same level of inaccurate data). Our empirical study on Apple’s App Tracking Transparency (ATT) framework supports this prediction. Introduced on April 26, 2021, ATT limits the use of third-party data under iOS 14.5 since it requires all apps to obtain the user’s consent for being tracked, leading to decreased data quality by allowing users to opt out of third-party tracking, massively affecting targeting practices on Facebook.^4,5 Analyzing data from 86 advertising campaigns on Facebook using a difference-in-difference approach, we found that narrow segments were more negatively affected than broader ones, leading to higher CPM and lower CTR for narrow segments. This finding likely extends to regulations restricting third-party data access because these regulations decrease data quality. Thus, it is important for advertisers to carefully assess the performance of narrow segments when they suspect data quality to be poor.

Final Thoughts

We find that targeting narrow segments can be ineffective in many cases, especially when there are concerns about data quality. The takeaway of our study is not to discourage targeting but rather to urge advertisers to scrutinize their segments meticulously. Our model aims to make advertisers aware of the break-even performance necessary for a targeted campaign to be at least as profitable as an untargeted one. If advertisers are confident in achieving the required performance lift, it is worth testing the segment in a randomized control trial.

. . . . . . . . . . . . . . . . . . .

Recommended Reading

Ahmadi, Iman, Nadia Abou Nabout, Bernd Skiera, Elham Maleki, and Johannes Fladenhofer (2024), “Overwhelming Targeting Options: Selecting Audience Segments for Online Advertising,” International Journal of Research in Marketing, 41(1), 24-40. https://doi.org/10.1016/j.ijresmar.2023.08.004

. . . . . . . . . . . . . . . . . . .

References

1. Meta (2023), “Core Audience Targeting,” Accessed February 13, 2023, https://www.facebook.com/business/news/Core-Audiences.

2. Ahmadi, Iman, Nadia Abou Nabout, Bernd Skiera, Elham Maleki, and Johannes Fladenhofer (2024), “Overwhelming Targeting Options: Selecting Audience Segments for Online Advertising,” International Journal of Research in Marketing, 41(1), 24-40. https://doi.org/10.1016/j.ijresmar.2023.08.004.

3. Neumann, Nico, Catherine Tucker, and Timothy Whitfield (2019), “Frontiers: How Effective Is Third-Party Consumer Profiling? Evidence from Field Studies,” Marketing Science 38(6), 918-26.https://doi.org/10.1287/mksc.2019.1188

4. Kraft, Lennart, Alexander Bleier, Bernd Skiera, and Tim Koschella (2024), “Granular Control and Privacy Decisions: Evidence from Apple’s App Tracking Transparency (ATT),” SSRN Journal. https://doi.org/10.2139/ssrn.4598472

5. Target Internet (2021), “How Apple iOS14 Changes Have Affected Facebook Third-Party Tracking,” Accessed May 22, 2024, https://www.targetinternet.com/resources/how-apple-ios14-changes-have-affected-facebook-third-party-tracking/

. . . . . . . . . . . . . . . . . . .

About Authors

Iman Ahmadi, PhD
Associate Professor of Marketing, Warwick Business School, University of Warwick (UK) 
Dr. Iman Ahmadi (PhD – Goethe University Frankfurt) is an expert in modeling and empirical research methods. His research interests include e-commerce, content-mining, competition analysis, pricing, and big markets. He regularly presents his work at international conferences and has published his work in the International Journal of Research in Marketing, Quantitative Marketing and Economics, Journal of Advertising, Journal of International Marketing, and Journal of Business Research.

Nadia Abou Nabout, PhD
Chaired Professor of Interactive Marketing & Social Media, WU Vienna (Austria) 
Dr. Nadia Abou Nabout (PhD – Goethe University Frankfurt) is an expert on digital marketing and programmatic advertising and is interested in topics such as brand safety, viewability, tracking and targeting, and advertising effectiveness. In her work, she aims to help companies make better marketing decisions and build upon extensive industry collaborations. Together with Bernd Skiera, Dr. Nadia Abou Nabout was one of three finalists in the “Gary L. Lilien 2011-12 ISMS-MSI Practice Prize Competition” and is a member of the editorial board of the International Journal of Research in Marketing and Journal of Retailing.

Bernd Skiera, PhD
Chaired Professor for Electronic Commerce, Goethe University Frankfurt (Germany) 
Dr. Bernd Skiera’s (PhD – University of Lueneburg) research interests include electronic commerce, online marketing, marketing analytics, consumer privacy, and value-based customer management. He is a fellow of the European Marketing Academy (EMAC) and received an ERC Advanced Research Grant from the European Research Council to examine the economic consequences of online tracking restrictions.

. . . . . . . . . . . . . . . . . . .

Download this Article

KCRR 2025 June  - Precision Paradox: The Complex Reality of Targeting (Ahmadi).pdf