Broadcasters generate a large part of their revenue through advertising. For CBS, the most-watched U.S. broadcast network, T.V. advertising accounts for two-thirds of total revenue. Major live events, such as election night, the Super Bowl, the Olympics, and the FIFA World Cup, strongly boost such revenues because advertisers are willing to pay a premium for their ads to air during the live broadcast of these events. In 2010, for instance, the cost of a 30-second spot during the Super Bowl was between $2.5 and $2.8 million, or 18 times higher than the corresponding prime-time advertising rates. Similarly, a 30-second spot during the Winter Olympics in the same year generated between $360,000 and $490,000, about three times the rate of an average prime-time spot. However, the selling and scheduling of advertisements in such an environment is a challenging task. The process of advertising in T.V. can be divided into four main phases: (1) selling of slots, (2) selection of advertisers, (3) optimal scheduling of slots, and (4) evaluating achievement of objectives.
The most critical challenges of radio and T.V. networks are:
- Advertisers want to buy periods, which are called advertising breaks and located within the T.V. programs, to persuade the audiences to purchase or take action upon their products or services. T.V. stations have to schedule programs interspersed with advertisements. It is essential to place advertisements so that the combination of advertisements would have the largest exposure possible to maximize the total returns from the adverts. One of the major problems faced by T.V. channel planners is to allocate the advertising time to advertisers to maximize revenue. The problem is complex due to the limitation of time slots for advertising. Some advertising requesters require the commercial to be seen by a given number of people, and the broadcaster will schedule the commercial to achieve this goal. Others want to have the broadcast seen by a particular demographic group or certain socio-economic classes of people. Some customers may want their commercials at the start/end of a given advertising break, and others may want the commercial to be associated with a given T.V. program. This makes finding optimal schedules computationally intensive problems that cannot usually be solved by hand. For this reason, naive approaches to scheduling lead to wasted resources and disenchanted audiences when ads fail to reach the interested consumers efficiently and must be aired repeatedly to meet impression targets.
2) In the Internet advertising space, available inventory is abundant. The problem is, only a subset of that inventory will work for their advertisers and campaigns. The challenge then becomes how do they spend a minimal amount of money to identify the subset of inventory that will perform the best but at the same time be the most cost-effective? In short, how can they successfully optimize Internet Advertising without breaking the bank or taking up all of their free time?
3) In live broadcasting, the break lengths available for commercials are not always fixed and known in advance (e.g., strategic and injury time-outs are of variable duration in live sports transmissions). Broadcasters actively manage their advertising revenue by jointly optimizing sales and scheduling policies.
4) The selling and scheduling of advertisements are challenging for sports events that involve unpredictable breaks during which advertisements can be shown. A case in point is cricket, whose matches have breaks of random duration. The uncertainty about break durations creates an obvious problem for the broadcaster: how to schedule (in real-time) the ads that have been sold while respecting the relevant constraints. The broadcaster must respect not only capacity constraints, under which the total duration of the ads scheduled during a break cannot exceed the length of that break, but also diversity constraints, under which two ads from the same advertiser (or from advertisers in the same industry) cannot be shown during the same break. Suboptimal or infeasible schedules have many undesirable consequences for a broadcaster. If the schedule violates diversity constraints, no revenue will be earned, and capacity will be wasted. A schedule that violates capacity constraints could lead to rescinding of the broadcast rights or other costly penalties. For example, cricket broadcasting rights require the broadcaster to guarantee live coverage of every ball of every match.
5) networks try to minimize the penalties incurred for not meeting the requirements of the advertisers. The penalty could be incurred due to the reasons, namely not reaching the specified audience rating points in the entire advertising campaign, irregularity in multiple airing of the advertisements, or not meeting the position goals. Advertisers check for their requirements weekly and put a penalty on the T.V. network for not meeting any of their requirements. As per specific rules, advertisers are also allowed to cancel orders for advertising in advertising breaks. The cancellation can be made up to 48 h before the airing of the program. Depending on the rules, the T.V. network may or may not charge a cancellation fee for canceling the order and may also refund the entire amount or part which was going to be spent by the advertiser. Moreover, several changes occur before the starting of the broadcast week, such as show format alterations, requests of advertisers for revision in the schedule of the commercials, and product conflicts owing to some cancellation of orders by advertisers.
6) Major T.V. networks typically have several shows available for broadcasting. Some of these belong to a series of half-hour shows, whereas others belong to a series of one-hour shows. There are shows of other lengths as well. There are a fixed number of 30-minute time slots, implying that some shows require a one-time slot, whereas others require two consecutive time slots. A specific rating can be expected if a particular show is assigned to a given time slot. The forecasted ratings may be based on experience with the show and the time slot; it may be based on lead-in effects due to the shows immediately preceding it. It may also be based on shows that competing networks assign to that same slot. The profits of the network depend very much on the ratings. So, one of the main objectives and challenges of the network is to maximize its average ratings.
7) Self-promotion is very important for T.V. stations; for instance, in 2016, the four Portuguese open T.V. channels broadcasted 123 298 self-promotion spots, which took up 997 hours. This means that, on average, about 41 minutes a day were used by each channel to promote their shows. Furthermore, in that same year, self-promotion accounted for about 15% of the total advertisement time (commercial and non-commercial). T.V. stations use self-promotion time slots for advertising the shows to be broadcasted to increase the shows’ audiences and thus, the value of the commercial breaks, leading to higher revenues from selling the latter breaks. The time made available for self-promotion is limited, and therefore, it needs to be effectively used. The effective use of time slots implies its effective scheduling, that is, the time sequence of shows to be advertised.
8) If a music scheduler likes some songs more than others on radio stations, there should be some songs in the music library that seem never to get scheduled.
9) In radio stations, song clustering means some songs consistently over-schedule near the same time of day, even as the music director’s formatting plan is for them all to schedule equally in all dayparts. Clustering is bad, and radio stations need rotation patterns for each song category to avoid song clustering by delivering equal and balanced play for every song with absolute reliability.
2. How Optimization Can Help
Broadcast programming is the scheduling of broadcast media shows, typically radio and T.V., in a daily, weekly, monthly, quarterly, or season-long schedule. The executive in charge of selecting the programs and planning the schedule is sometimes the director of network programming. At a micro level, scheduling is the minute planning of the transmission, what to broadcast and when ensuring an adequate or maximum utilization of airtime. T.V. scheduling strategies are employed to give shows the best possible chance of attracting and retaining an audience. They are used to deliver shows to audiences when they are most likely to watch them and deliver audiences to advertisers in the composition that makes their advertising most likely to be effective. In the scheduling of network T.V. programs, the scheduling horizon is typically one week, and the week consists of a fixed number of time slots. Several shows are available for broadcasting. These shows have to be assigned to different time slots to optimize a specific objective function. Moreover, the assignment of shows to slots is subject to a variety of conditions and constraints. For example, assigning a show to one slot may affect the contribution to the objective function of another show in a different slot.
Music scheduling systems are employed to sequence music at radio stations. Music scheduling is simply the function of generating a playlist. In the general radio broadcasting sense, scheduling is the placement of content against a linear timeline for transmission on a broadcast station. This content may include not only music but also commercial advertisements, station identifiers, and promotional jingles. Commercial advertisements, called spots in radio lingo, are scheduled by their own separate scheduling system, which keeps track of monetary considerations. The music schedule, non-music schedule (jingles, promos), and the commercial schedule are later merged into a single schedule (called the log) to guide what must be played on the station on a minute-by-minute basis.
With this background, we can use optimization techniques to help media companies in the following ways:
- We can consider the scheduling of advertisements in T.V. from T.V. network and advertiser perspectives. The objectives of optimization models from the T.V. network perspective could be the maximization of the audience ratings, minimization of the cost, or trading-off between the two. On the other hand, when the model is considered from the perspective of the advertiser, the objective could be the maximization of the reach and minimization of the cost of advertising. Apart from the optimization models, we can develop heuristics and population-based meta-heuristic optimization algorithms to solve the large-scale optimization models in case the models cannot be solved within a short time. Moreover, we can break a large problem into three stages to reduce the run time of optimization models. The following figure shows the schematic processes and information flow in a decomposed problem as an example.
We can consider the following types of optimization models in our advertising optimization tools:
- Minimum Space Requirements Models: these models are related to minimum space requirements between multiple airings of advertisements. Each advertiser wants that the multiple airing of the same commercials to be as much evenly spaced as possible.
- Penalty Minimization Models: The T.V. networks promise advertisers that they would try and meet their preferences such as that of location preferences (first or last position of the advertisement breaks) as these are likely to have higher audiences, avoiding product conflicts by placing their advertisements close to the advertisements of their competitors, etc. Although it is not always possible to meet all the preferences of the advertisers, the T.V. channel tries to meet the preferences as much as possible. Advertisers set a certain percentage of the audience that they would want to reach; if the T.V. network is able to meet that requirement, the advertisers are satisfied to a greater extent. The schedule for placement of advertisements is made such that the penalty incurred for not meeting the requirements of the advertisers is minimizes.
- Multi-objective Optimization Models: we can develop optimization models with multiple objectives while scheduling advertisements. The multi-objective nature of the models could either be considered from the perspective of the T.V. network or from that of the advertiser. For example, we can consider two conflicting objectives of maximization of audience and minimization of cost. In such models, the values of ratings and costs can be first forecasted by the regression model and then used further in the optimization model as inputs.
- Revenue Maximization Models: The T.V./radio networks generate a major part of their revenue by selling their advertising time in breaks of various programs to advertisers at a particular rate. While maximizing the revenue, the T.V./radio networks must consider increasing the audience ratings and at the same time meeting advertiser preferences. For example, we can develop optimization models in which the T.V. station accepts some advertisements and then schedules these advertisements with the objective of maximizing the revenue.
2) In live broadcasting, the break lengths available for commercials are not always fixed and known in advance (e.g., strategic and injury time-outs are of variable duration in live sports transmissions). Broadcasters actively manage their advertising revenue by jointly optimizing sales and scheduling policies. So, we can develop optimization models for T.V.\radio networks with a stochastic capacity to advertise airtime during a live event. This capacity consists of a number of commercial breaks of random duration. Breaks occur sequentially over a period of time and must be filled immediately upon arrival.
3) We can develop an optimization decision support tool that aims to plan and maintain the weekly self-promotion space for a T.V. station. The self-promotion plan requires the assignment of several self-promotion advertisements to a given set of available time slots over a pre-specified planning period. This tool should consist of a database, a statistic module, an optimization module, and a user interface. The input data should be provided by the T.V. station and by an external audiometry company, which collects daily audience information. The statistical module provides optimization models or algorithms that can find good solutions quickly. The interface should report the solution and corresponding metrics and can also be used by the decision-makers to change solutions and input data manually.
4) We can develop an optimization tool for scheduling music for radio stations. Our optimization tool can ensure each hour of the broadcast day is consistent with the station’s formatting plans and the songs and all other content are evenly and reliably scheduled. Moreover, it should have the following features:
- Accurate song spins and rotations. For example, it should favor songs that have not been played recently in or near the same hour and dayparts for the best possible rotation patterns.
- Preventing song clustering by providing optimal rotation patterns for each song category
- Scalable Rules rules with different preferences to match the priorities of a radio station. For example, “Always try to separate slow songs by three. If you cannot, two is acceptable. But never play them back-to-back.”
- Creating groups of rules that only apply in specific dayparts or when certain criteria are met.
- Rule tree wizard that analyzes the song library to get recommendations for rule settings. Or review the last scheduling sessions to find potentially problematic rules.
- Keeping the station fresh by scheduling a regular exchange of songs between active and resting categories, based on play count or days.
- Special library and history reports that help the radio station examine its overall content and performance by examining relative separation, viewing category turnover and weekly spins, etc.