With the emergence of programmatic and new sales channels, advertisers have become extremely sophisticated, making smaller and more focused buys targeted to a narrow demographic audience. To accomplish this, more and more advertisers are using real-time data and Artificial Intelligence to improve their campaign efficiency and effectiveness. In addition, major technology trends such as the emergence of a mobile-first millennial generation as well as internet based subscription models (YouTube, Spotify) have disrupted the media industry. What do these trends mean for the broadcast media industry? An exponential increase in complexity, intense competition, reduced ad budgets, and cost cutting across the board.
The Road Ahead – The more things change, the more they stay the same
Below are the steps most media companies take when it comes to generating revenue for ad sales. Given the disruption in the industry, unfortunately not much has changed.
Each of the above steps has traditionally been an extremely manual, time consuming process without a lot of science or analytics behind it. Pulling data from disparate traffic systems into Excel, manual data manipulation in pivot tables, intuition based determination of ratecards, and manual spot placement using complex UI screens of traffic systems. All this results in gut-base decision making. Needless to say, these outdated practices result in most broadcasters leaving millions of dollars on the table. Add to this, the compounding effect of fresh challenge, tectonic shifts in demographics and media consumption, most broadcasters are in the midst of a perfect storm.
Delivering the right ad, at the right time, to the right audience, at the right price
Leading broadcasters use the latest innovations in Machine Learning & Artificial Intelligence methods to accurately forecast ratings as well as demand, and to determine optimal prices for ratecards based on customer sensitivity analysis. They use state-of-the-art algorithms from Integer Programming & Combinatorial Optimization (IPCO) – a branch of Operations Research & Data Science – during the sales process to create optimal proposals that match advertiser requirements, and to determine optimal spot placements on the log.
IPCO methods can help Account Executives (AEs) come up with better deal plans, and traffic managers optimally schedule spots on the log.
Ad Schedule Optimization for Spot Placement
Most broadcast media companies use relatively simple heuristics to place spots on the traffic log on a nightly basis. A team of traffic managers then sift through these ‘system generated’ logs and swap spots around to improve inventory sellout levels or better adhere to a variety of business rules. Talking about business rules – there are over 25 different categories of rules some of which include flight date restrictions, advertiser / industry separation requirements, constraints on which spots can be assigned to different breaks, exclusions, but also have a fair and equitable distribution of spots.
The basic problem is to allocate available inventory slots to demand from all on-the-books orders and place as many spots on the log as possible, while preserving high-value inventory for future demand. This is like a Tetris puzzle – orders are like Tetris pieces, each with a different shape & color – each order has different requirements along multiple dimensions. All orders must be packed into the available inventory slots as tightly as possible, so that we maximize the value of remaining inventory while fitting as many Tetris pieces as possible.
IPCO based modeling approaches can be used to solve this complex scheduling problem. Revenue Analytics has leveraged decomposition based methods, along with IPCO approaches, to solve challenging ad placement problems in a matter of minutes. Typical models have over 3 million decision variables and over 3 million constraints. The system throws all available spots to be scheduled in the air when it is run, evaluates billions of potential solutions and then picks the best solution that balances advertiser requirements with the broadcaster goals of preserving the value of the remaining inventory. When comparing optimal ad placements using IPCO based methods to relatively simple heuristics, the difference is pretty dramatic – greater than 1% improvement in revenue on the log that translates into tens of millions of dollars of revenue, significant improvement in preserving high-value inventory & frontloading, as well as dramatic (>50%) reduction in manual ad placements by the traffic teams.
Proposal Optimization for Deal Planning
Advertisers provide buy requirements to AEs in the form of target CPP, demo, budget and a host of other parameters including flighting, spot length mix & daypart mix among others. AEs have traditionally built proposals manually to satisfy all RFP requirements, with little to no attention to impacts on the value of remaining ad inventory. This is a very time-consuming process, taking even experienced AEs hours to come up with a proposal. This problem has over 20 different categories of constraints and multiple conflicting objectives – it is almost impossible for any human to find even close-to-optimal solutions in a reasonable timeframe. The problem is highly nonlinear, with multiple peaks and valleys that make it extremely difficult to identify the best solution. The figure below illustrates this – relatively simple heuristics tend to get stuck in local optimal solutions, without finding the true global optimal.
The result – suboptimal deals that either over-deliver (thus leaving money on the table for the broadcaster) or under-deliver (leading to additional liabilities and costly make-goods) on the RFP, enormous time spent by AEs to create proposals. There is a better way.
AEs input these requirements into the Proposal Optimization model which evaluates hundreds of millions of scenarios and calculates the best proposal to meet all RFP requirements, while preserving high-value prime inventory as much as possible. Typical problems have over 100k decision variables and 100k constraints. State-of-the-art methods from IPCO can be used to solve this complex proposal optimization problem in a matter of seconds. Partnering with our clients we’ve seen IPCO based methods generate optimal proposals that better satisfy all RFP requirements while significantly preserving the value of ad inventory, for RFPs going out up to 18 months in the future, across multiple channels / stations.
Media companies can turn current headwinds of intense competition, increasing complexity in their business and reduced ad budgets into a competitive advantage – the key is to search for revenue increases in unexplored areas of the business. Some of the savviest broadcast media companies use automated tools and cutting edge analytics to better optimize ad placements and create proposals – leading to a dramatic reduction in manual work and better utilization of available inventory, ultimately resulting in organic revenue growth.