Written By T. (Ned) Choungprayoon, Ph.D. Candidate at Stockholm School of Economics
During my transition out of an academic career trajectory (not as a newsletter writer), I interviewed with many marketing analytics companies; mobile game application companies, advertising agencies, consulting firms—and so forth. Recently, Google just changed its stance, by canceling its plan to eliminate third-party cookies in Chrome (which track people’s search behavior, for example). Instead, Google now aims to give users more control by allowing them to accept or reject cookies. Regulators are not very happy with this decision and have shown more concern about data protection and privacy. Since most marketing analytics companies rely on cookies, and the elimination is no longer imminent, they want to know one thing: “If users accept cookies, how long should cookies be allowed to track them and what is the effect of that?” This timely and relevant question has been discussed widely within the industry.
As an academic researcher in marketing, I was prepared for this question by reading articles from top marketing journals. One that interested me is a recent article in IJRM about the effect of cookie restrictions by Klaus Miller and Bernd Skiera. This article is well-written and easy to understand for both academics and industry audiences, plus it's open access. It briefly overviews cookies (i.e., online tracking technologies) and explains the economic consequences of restricting them. The authors utilized observational data from a programmatic marketplace, which facilitates the automated buying and selling of online display, mobile, and video advertising inventory. Their analysis focuses on a key aspect of cookie management: the lifetime restriction of cookies, which refers to the maximum duration a cookie can remain on a user’s device before it is automatically deleted, conditional on the user giving consent in the first place.
In one of my job interviews, I impressed the interviewers by explaining the potential effects, highlighting what I learned from the article (and telling them that they could read this paper without a paywall). That’s why I chose to interview Klaus Miller for my final article before transitioning into the industry.
“We chose IJRM because it's known to publish innovative and timely topics. And I think our paper is very relevant now, especially in the broader context of the economics of privacy.”
- Klaus Miller -
Power of Cookies
Restricting tracking technologies’ lifetime, such as shortening the lifespan of cookies, could impact how advertisers target users and measure the effectiveness of their campaigns. Cookies enable publishers to offer more personalized and relevant ad experiences to their audience, which attracts higher bids from advertisers and increases the value of their ad inventory. This subsequently affects the revenues publishers generate. (Un)Fortunately, the effect depends on the type of cookies, if their values increase or decrease over time, and the restriction time imposed. Some cookies are either flat in value or are deleted by users before reaching the cookies’ lifetime restriction (For example, you might clear your cookies at least once a year). This means that the actual economic loss might be less severe than anticipated (depending on the restriction time). In some cases, shortening the lifetime of cookies that are decreasing in value over time can be advantageous. This limitation allows for the introduction of new cookies, effectively resetting the tracking data. This reset enables advertisers to refresh their data, better capturing current user behavior and preferences.
“We just want to learn what it would mean for the online advertising market if we have shorter cookie lifetimes or longer cookie lifetimes. It is a trade-off for the regulator or for the society to solve. More privacy and shorter lifetimes have more negative effects on the industry. Or do we want longer lifetimes with less privacy for the user, but more economic prosperity for the firms.”
- Klaus Miller -
Power of Simulations
Working in finance, I am personally a fan of simulations to illustrate scenarios. In the paper, simulations play a crucial role in estimating the impacts of cookie restrictions in the advertising market. The simulations allow researchers to model and analyze complex scenarios when real-world experiments are infeasible to conduct or when data with exogenous variation is hard to obtain. By assuming that advertisers' and users’ behaviors remain constant (because why not?), Klaus and Bernd could isolate the impact of cookie restrictions on advertising revenue, providing an idealized picture of potential outcomes on different types of cookies (e.g. cookies with constant value vs. ones with increasing values).
“Ideally, we would run an experiment where different lifetime restrictions are imposed, that would be the gold standard; but we couldn't do this and a quasi-experiment was not available. So what we're doing is we're essentially simulating the market and making an assumption that the advertisers would continue to advertise as they were advertising in the past and the user behavior doesn't change, which of course are maybe reasonable assumptions. And of course, you can debate about whether this is true or not, but we have some good arguments in the paper.”
- Klaus Miller -
Power of Numerical Example
Reading through Klaus and Bernd’s paper, I found the numerical example section very helpful in translating complex simulation and effect estimations into simple and understandable insights. One page of numerical examples can help me speed through another two pages of regression analysis and understand the previous four pages of background better. Using numerical examples not only clarified their approach but demonstrated their understanding of the methodology required to work with large datasets. During the interview, I couldn’t help but mention how much I like this section and wish many papers would do the same, as it makes complex methods more accessible and sophisticated concepts more understandable.
“This approach was also very helpful for us to make sure that we are calculating the economic loss correctly. Once everybody was on the same page with the different cookie types and the data, we showed what the results are. So, I thought that was actually a really good process for us, to understand what's going on and to communicate what we're doing.”
- Klaus Miller -
Are you ready to think about what will happen to online tracking technology in the future? If real-world experiments are infeasible, why not be more innovative in your approach? Start by looking at observational data to understand current phenomena and set up potential outcomes. Use simulations to illustrate the effects of these outcomes that have not yet materialized, and provide some numerical examples for readers to easily follow along.
Read the paper
Interested in finding how cookies' lifetime restrictions will affect the economy? Read the full paper here.
Meet Klaus Miller
Klaus Miller
What drives you to do the research / work you do?
“I had an experience working in the industry (media companies) and I really enjoyed that. But I think I enjoyed academia more. I can pick my own research topic which could be something that is relevant in a broader context. That is also what I like about this online tracking topic, it's relevant for regulators, the advertising industry, and users. It also has a broader impact than just doing profit maximization, for example. The research topic can be very diverse. Besides a marketing topic, It can involve a legal topic, it can involve a technology topic.”
If you were not an academic, what would you be?
“If I weren't in academia right now, I would try to become one. I think that's the best job in the world for me. It's not work, it's like having fun and working on things that I'm interested in. It's actually more like, oh, it's already the end of the week. But if being in academic is not an option, I would become an architect so I could build something that I can see and still have the conceptual creative part.”
This article was written by
T. (Ned) Choungprayoon
Ph.D. Candidate at the Stockholm School of Economics (Sweden)
Comments