Balancing ROI in Computational Advertising: Multi‑Party Game Theory and the Role of Information Asymmetry

Computational advertising is fundamentally a game of ROI balance between advertisers and ad platforms, where each side seeks the most advantageous point in a multi‑party bargaining process. The evolution of pricing models—from early CPM "slot" sales to CPC, CPA, and the current oCPX hybrid—reflects this ongoing negotiation.

Different ad formats (brand, e‑commerce, game, SMB) share common characteristics: they are transactions whose success is measured by ROI, which includes both the advertiser's return on spend and the platform's return on traffic sales.

In the early "slot‑selling" era, abundant traffic and low competition allowed advertisers to achieve high ROI, but as internet content homogenized and user growth slowed, traffic quality declined, prompting platforms to shift toward more precise CPC and CPA models to protect their ROI.

The shift from coarse CPM to fine‑grained CPC and CPA was driven by the need to track conversions more accurately; however, true ROI calculation only became feasible with the introduction of CPA‑style active conversion tracking, later refined into the oCPX model that balances platform and advertiser interests.

Information asymmetry is identified as the commercial essence of advertising: advertisers exploit gaps in user knowledge to create value. Historical examples, such as regional price differences in physical goods, illustrate how information gaps generate bargaining power, a principle that persists in digital ad ecosystems.

Even today, a sizable portion of e‑commerce ads target users who still lack full awareness of online shopping options, creating opportunities for higher ROI through tailored messaging.

From a technical perspective, data and algorithmic models are essential for quantifying and exploiting these information gaps. Targeting (explicit user interest labeling) and ad ranking (eCPM and CTR prediction) are the primary mechanisms that translate data insights into higher conversion rates and, consequently, higher ROI.

Business‑level strategies—such as customized campaign tactics, DMP utilization, and platform tools—complement technical methods in extracting information asymmetry.

The article concludes that the entire computational advertising ecosystem revolves around ROI, with continuous cycles of data‑driven optimization and strategic adjustments to maintain the balance between advertiser and platform interests.