FEEDBACK LOOP AMPLIFICATION IN ALGORITHMIC BUDGET ALLOCATION: A DISCRETE LOGISTIC MODEL OF THE ADVERTISER DEATH SPIRAL

Abstract

Modern automated bidding systems on Google Performance Max and Meta Advantage+ behave as contextual bandits whose reward channel is the advertiser's own conversion stream. 

References

1. Calvano E., Calzolari G., Denicolò V., Pastorello S. Artificial intelligence, algorithmic pricing, and collusion. American Economic Review. 2020. Vol. 110, No. 10. P. 3267–3297. DOI: https://doi.org/10.1257/aer.20190623.
2. Association of National Advertisers. Programmatic media supply chain transparency study. ANA, 2023. URL: https://www.ana.net/miccontent/show/id/rr-2023-12-ana-programmatic-media-supply-chain-transparency-study (accessed: 06.04.2026).
3. Angluin D., Laird P. Learning from noisy examples. Machine Learning. 1988. Vol. 2, No. 4. P. 343–370. DOI: https://doi.org/10.1007/BF00116829.
4. Natarajan N., Dhillon I. S., Ravikumar P., Tewari A. Learning with noisy labels. Advances in Neural Information Processing Systems. 2013. Vol. 26. P. 1196–1204. URL: https://papers.nips.cc/paper/2013/hash/3871bd64012152bfb53fdf04b401193f-Abstract.html (accessed: 06.04.2026).
5. Cover T. M., Thomas J. A. Elements of information theory. 2nd ed. Hoboken: Wiley, 2006. 776 p. DOI: https://doi.org/10.1002/047174882X.
6. Ivitskiy I., Savchenko D., Sydorenko D. Data integrity as the terminal constraint in AI-driven advertising: an information-theoretic analysis of conversion fraud and agentic threat evolution. Zenodo. 2026. DOI: https://doi.org/10.5281/zenodo.18675362.
7. Ivitskiy I. API-native marketing operations: a reference architecture for automated decision loops. Global Trends in Economics, Finance and Management: proceedings of the III International scientific and practical conference (Kyiv, September 30, 2025). Kyiv: Research Europe, 2025. P. 101–104. URL: https://researcheurope.org/wp-content/uploads/2025/10/re-30.09.2025-101-104.pdf (accessed: 06.04.2026).