Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TD20

5 - Self Pricing Beating in a Market With Preference Interdependence and Uncertainty Ting Luo, Assistant Professor, California State University Fullerton, Mihaylo College of Business and Economics, Fullerton, CA, 92831, United States, Lijia Shi Self price beating pricing strategy promises the early buyers that if the seller lowers the price in the later period, an ex post price refund that is more than the price difference will be refunded to them. We propose self price beating as a pricing policy under market externality and uncertainty. It combines the advantages of both price commitment and no price commitment. Based on a two- period model, we find that self price beating can sustain the high 1st period demand as price commitment does, while it can also generate high 2nd period profit as in the no price commitment case when there is market uncertainty. The total profit from self price beating surpasses both price commitment and no price commitment. 6 - Designing Shipping Policies for Online Retailers: The Role of Topping-up Behavior to Qualify For Free Delivery Lifei Sheng, Assistant Professor, University of Houston Clear Lake, Houston, TX, United States, Guang Li, Dongyuan Zhan We study a widely used shipping policy for online retailers. Under such a policy, a flat shipping fee can be waived if a customer’s total purchase amount in a single order exceeds a certain pre-determined threshold. The threshold may promote some customers to top up their order size to qualify for free shipping. We model this “topping-up” behavior and design the optimal shipping structure under both monopolistic and competitive settings.

4 - Applications of Mixed-Integer Programming in Media Industry Oguzhan Ozlu, Revenue Analytics, Atlanta, GA, 30318, United States In this talk, we’ll present the applications of mixed-integer optimization models in scheduling television/radio commercials and a high-level architecture of such scheduling solutions. Advertisers purchase several spots to air their commercials for a specified time period. The goal of an ad-scheduling solution is to optimally place all advertisements to maximize the advertiser and broadcaster revenues while minimizing the violations of advertisers’ order requirements and reducing the need for manual ad placements by scheduling teams. The ad-scheduling solution automatically retrieves key data inputs from a variety of enterprise data marts, solves a series of interacting MIP’s and publishes resulting schedules to the enterprise scheduling system. We will present the challenges and key learnings that helped us to develop automated cloud-based scheduling systems for large media enterprises. 5 - Enterprise Revenue Management System Pratik Mital, Revenue Analytics, Atlanta, GA, 30318, United States Developing an enterprise revenue management system is a complex task that requires significant resource investment (IT infrastructure and people), and time investment. The task becomes even more complex where entities from different parts of the world are involved and a common platform is being developed to serve all of them. Systems thinking and developing the right team to work on this task is key to its success. This talk will present a case study from developing an enterprise revenue management system for one of the largest cruise companies in the world. Revenue Management in Advertising Sponsored: Revenue Management & Pricing Sponsored Session Chair: John G. Turner, University of California-Irvine, Irvine, CA, 92697-3125, United States 1 - Managing Digital Advertising Campaigns Sami Najafi Asadolahi, Santa Clara University, Leavey School of Business, 500 El Camino Real. Lucas Hall, Santa Clara, CA, 95053, United States, Naren Agrawal, Stephen A. Smith Advertising agencies, a key intermediary in the digital advertising supply chain, manage numerous ad campaigns for multiple clients in real-time. Because of uncertainties in the demand from campaigns for viewers, and the rate at which the target viewers visit websites, ensuring that campaigns proceed according to plan is a difficult challenge. We describe a methodology to manage such campaigns. 2 - Dynamic Revenue and Project Management with Application to Online Advertising Hossein Jahandideh, Google, Mountain View, CA, United States We consider a provider who makes dynamic pricing decisions for deadline-based projects. The provider has the flexibility of processing a project in different time periods as long as all projects are completed by their respective deadlines. Because of the additional flexibility, this amounts to a non-traditional dynamic network revenue management (DNRM) problem. We convert this problem into the format of a traditional DNRM, approximately solve the problem, and provide a real-time pricing strategy. We discuss the application of this model to online advertising, where advertising campaigns correspond to projects. 3 - Effect of Complex Multimedia Advertising Campaigns: A New Automated Method for Big Data Will Wei Sun, University of Miami Business School, Miami, FL, United States, Pengyuan Wang, Guiyang Xiong, Jian Yang The proliferation of online advertising not only brings new opportunities for advertisers to reach the consumers but also allows them to efficiently track large- scale individual-level data on ad exposures and conversions. Advertisers often question whether multimedia campaigns are actually effective and it remains unclear how to maximize their success. This study introduces a novel tree- structured causal inference model, which is nonparametric, flexible, computationally efficient, and suitable to analyze complicated nonlinear effects. The model enables automatic segmentation of consumers, allowing advertisers to better optimize the allocation of ad resources via accurate targeting. 4 - Competitive Real-time Policies for the Allocation of Online Display Advertising Ali Hojjat, University of New Hampshire, 10 Garrison Ave., Durham, NH, 03824, United States, John G. Turner We propose a new bid scaling mechanism for real-time matching of online guaranteed targeted display advertising with a website’s users. Using primal-dual analysis, we derive the competitive ratio and show the conditions under which it beats the commonly known best bound of 1 - 1/e. n TD20 North Bldg 129A

n TD19 North Bldg 128B

Novel Revenue Management Models Sponsored: Revenue Management & Pricing Sponsored Session Chair: Pratik Mital, Revenue Analytics, Atlanta, GA, 30318, United States 1 - A Novel Approach for Multiple-round Bid Pricing

Iman Nekooeimehr, Revenue Analytics, Atlanta, GA, United States In B2B pricing, it’s very common for the sellers to propose their prices in multiple rounds of bids. The sellers have multiple chances, therefore they can price their products/services higher and gradually decrease their prices in the next rounds of bids. In this presentation, we would like to present a new framework which assists companies find the optimal price when participating in multiple-round bids. 2 - Mathematical Modelling for Time-of-Use Pricing of Electricity in Monopoly and Oligopoly Goutam Dutta, Professor, Indian Institute of Management, Wing 3, Room No 3H, Production Quantitive Methods Area, Ahmedabad, Gujarat, 380015, India, Nidhi Kaicker, Debamanyu Das, Subhashree Banerjee This paper attempts to study the efficiency gains for time-of-use (TOU) pricing over flat-rate pricing in the electricity sector. The electricity market may be characterised by a monopoly in some cases, where a single firm continues to enjoy market power, or an oligopoly, where two or more firms compete against one another by strategic interactionThis study establishes the feasibility condition for efficiency gains to arise from time-of-use pricing in a monopolistic set up using constrained optimization.The strategic behaviour of firms in a duopoly, generalizable to n firms, is modelled in this study using constrained optimization. 3 - Approaches to Cancellation Forecasting in the Travel Industry John Harvey, Senior Manager, Carnival UK, Southampton, United Kingdom In the Cruise Industry, understanding Cancellation behaviour is key to maximising revenue, as we can identify mitigation strategies as well as overbooking strategies. But new patterns can take time to emerge so relying on previous cancellation curves is insufficient in a dynamic market. We propose a statistical approach of modelling Cancellations using a Poisson Process at an appropriate level of aggregation. These will to allow us to generate confidence intervals and establish the probabilities of a new cancellation trend emerging, which can then be used to update the Poisson models via Bayesian methods.

353