Thursday, July 28, 2011

Scheduling of TV Advertisement: Theory vs Practice

I am working on a model for scheduling TV advertisements. Looking at the paper: this is easy to model (but difficult to solve). My model seems at least 10 times as complicated than what this academic study suggests. We needed more complexity to deal with:

  1. The number of spots to broadcast is not known in advance (depends on expected ratings of a break)
  2. There are complicated issues with spread over time
  3. There is a rather complex issue with the quality of the remaining unused space (this needs to be valued with respect to expected ratings and optimized)
  4. Different contracts have different requirements, and these requirements are not always easily modeled
  5. The model is huge (> 150K binary variables) for a monthly schedule
  6. The input data (from a database) is large and complex (this is sometimes called “data-intensive”)
  7. Numerous other things


  1. This sounds like a very interesting problem. I'm thinking of assigning it as a project in a short course on constraint programming that I'll be teaching early in November. Are you allowed to provide some more details of your realistic model? I'm especially interested in sizes (#spots, #clients, etc.) and, if possible, the description of some of the tough constraints that you mentioned. I'd like to make the theoretical model from the paper a bit more realistic and appealing to the students. Thanks in advance! Tallys.

  2. I am afraid the model can not be explained in just a few sentences. One would need to get familiar with GRPs, target audiences etc. The number of spots to be broadcast is variable in our model (it is a model result).

    I can give some order of magnitude numbers. The number of breaks is several thousand (this is a concept used a lot in Europe, where ads are shown in fixed breaks). The number of different spots is in the hundreds (I guess you call that client, but a single client can have several ad campaigns).

  3. Erwin, no problem. I imagined it would be hard to explain. Thank you for the order of magnitude numbers.

  4. Erwin: Speaking as an academic, if your model is that much more complex than the one in the paper, you simply need to make a bunch of "simplifying assumptions". Works for me all the time. Tell the client that all the complicating factors don't really exist. :-)

  5. New real world problems are like those hot, fresh and smelly breads in the bakery. It would be delightful to read about how you surpassed some of the difficulties you might be facing.