Programmatic optimization of ad pods for maximizing consumer engagement and revenue

Paper

Niranjan Kumawat, Manu Vajpai and Nitish Varshney

CTVs serve streaming content to users via the internet or a local network. These devices replicate the traditional TV experience and can serve sequential ads which are bundled together and then inserted into content. This bundling of ads is accomplished by ad-podding. However, un-optimized display of ads in these pods can impact consumer engagement and revenue. This work presents one of the first programmatic approaches to achieve optimal packaging of ads in pods with multiple constraints. We first establish the underlying problem as a multi-objective Knapsack problem and then explore established approaches of evolutionary algorithms and dynamic programming to tackle it. We further propose a new greedy approach which solves the problem efficiently for online deployment. Testing, with our in-house and public datasets, establishes the relative effectiveness and applicability of these approaches for optimizing podded creatives.