Optimizing floor price in Real Time Bidding

Abstract

[EN] AdNetwork companies are very much a part of today's new digital marketing methods. This paper aims to develop an algorithm that solves the problems of AdNetwork companies in setting optimal floor prices. Establishing the optimal starting price for the bid is equivalent to setting the price that maximises revenue, which is optimal for the publisher and the AdNetwork company. In this market, that price will balance two opposite scenarios: a high floor price could lead to some impressions unsold, while a low floor price could be insufficient to reach profit margins. The contribution is twofold. First, this paper extends the problem of optimal price flor in real time bidding auctions for advertising in current scenarios where a DSP (Demand Side Platform) acts as a filter and only one bid is received by the AdNetwork and thus, the price paid corresponds to the reserve price. and, moreover, it is implemented in reality with a pseudo-algorithm (not provided for commercial reasons). It allowed to be implemented in a real case scenario for three publishers, obtaining an average increase of revenue of 127%.