Uncovering the Hidden Data Costs of Mobile YouTube Video Ads
*Co-primary authors
The ACM Web Conference 2024 (WWW ’24), Singapore, May 13–17, 2024
Data17,600 main videos · 46,600+ ads · ~8,225 hours · 8 countries
Summary
Video platforms monetize via ads, but users also pay in mobile data, which is especially costly in developing regions. This paper presents the first independent empirical analysis of YouTube in-stream video ad data costs from the user perspective.
We built a crawl/stream pipeline and released a public corpus of 17,600 main videos and 46,600+ ads (~8,225 hours) across eight countries, four developing and four developed. Buffer-state analysis surfaces latent wastage:
- Users often still download 80–100% of a skippable ad’s data even after skipping (~31% of cases in developing regions, ~52.6% in developed regions).
- Mid-roll ads can force re-download of as much as ~71 seconds of main video on average early in playback.
- Excess losses average about 6.7% of a 2GB mobile plan.
We discuss implications for platforms, ABR/ad-insertion policies, and affordability/inclusion.
Longer write-up: The hidden mobile data cost of YouTube video ads
BibTeX
@inproceedings{atique2024uncovering,
title = {Uncovering the Hidden Data Costs of Mobile YouTube Video Ads},
author = {Atique, Emaan and Alam, Saad Sher and Ahmad, Harris and Qazi, Ihsan Ayyub and Qazi, Zafar Ayyub},
booktitle = {Proceedings of the ACM Web Conference 2024},
series = {WWW '24},
pages = {1138--1148},
year = {2024},
month = may,
publisher = {ACM},
doi = {10.1145/3589334.3645496},
url = {https://doi.org/10.1145/3589334.3645496}
}