Dynamic pricing. It’s a term used throughout many different industries, airlines, example2, example3, and sports to name a few. Typically however, dynamic pricing when it comes to sports is applied to ticket sales. Depending on what a variety of variables look like (opponent, day of the week, inventory levels, etc.), pricing on tickets can change hourly. Oddly enough, you don’t see much of this happening within corporate partnerships – at least not yet. Why? I’m not totally sure. As such, I’ve decided that it’s my next big project at work. There are certainly many different ways in which we could go about this, but let’s walkthrough a few different ways I can tackle this relatively quickly.
1) Price based on peak times
2) Price based on demand (inventory level)
3) Price based on inventory level and time (decay model)
Price based on peak times
First, let’s define what a peak time is. For our purposes, let’s define this as a time in which we can confidently predict substantially increased impression traffic. This could be during rush hour, or more specifically to that of sports, the hours leading up to and after an event occurs. As an example, let’s consider Out of Home boards – for those of you that don’t know, these are the huge LED screens that you may see outside of some arenas or stadiums. The look something like this:
These type of boards typically run throughout the day, and play host to many different clients that pay to have their creatives shown. As you can imagine, clients are paying for these creatives to play on these boards in the hopes that a large number of impressions may lead to goal conversions (this may be increasing brand recognition, increasing awareness of a new product, or an actual purchase of a product). Regardless of the goal, dynamically pricing these boards higher during peak hours makes a lot of sense.
The problem as it relates to our business model is that at this point in time it would be a logistical nightmare to sort through. Many of our contracts are structured in such a way that would not allow us to pull clients from certain hours in order to run a different set of clients during these peak times. That, but logistically stopping a current playlist of creatives and running a new set may be too taxing to take on with current technology and personnel bandwidth. For now this may be out of the question, but certainly something to look towards the future with.
Price based on demand (inventory level)
Create a straightforward supply and demand based model that sets price levels based on the amount of inventory available. The concept is generally the less inventory available, the higher the price. The reverse is also in play. Potential problems with this method is that the model doesn’t factor in the many variables that exist and come into play.
Price based on inventory level and time (decay model)
How about a hybrid to the former? Instead of pricing dynamically on the asset and its past historical data, the goal should be to minimize the amount of time inventory is available. Simply put, the strategy is to determine the optimal price point based on a mix of variables that includes both the amount of inventory available and amount of time the inventory has been available. The other key is to set a price floor. What are potential variables to consider?
· Amount of inventory available
· Amount of time it has been available
· Seasonality/ time of year
· Number of projected events (i.e. peak times and events)
· Pending contracts
· Upcoming Area events
Creating this model is more complex than it sounds, but not by much. The biggest challenge in building out this type of strategy is having both the granular data to allow for inclusion of variables such as time assets have historically been available before being sold, and having enough of it. Ideally you would want several years of consistent data to comb through. In our case we have nearly 3 full years, so although not perfect, at least there is a basis to base recommendations off of.
So which is the best route to go? Let’s not kid ourselves, there are a lot of different ways in which one could approach this problem of dynamically pricing assets. Above a few of them are outlined. The best route to go should be determined, in part, by the amount of data you have available. The flexibility you have in controlling the assets themselves is also very important, as factors such as existing contracts could add difficulties to implementations.
For our business practices, I’m building and implementing the latter. I’ll keep this group updated as to how that ends up going.