In a past life, when I computed Internet costs for hosted customers based on their usage, I wrote a paper explaining how 95th percentile billing worked. This type of billing is used for commercial customers and bills for near peak volumes. It is unique in that it allows for a certain amount of unbilled usage spiking. At the time, and I expect even now, few customers truly understand it or know how to tailor their business processes to reduce their Internet bills.
Internet billing is a funny thing. Very few people have an intuitive feel for the units of measurement used, but this in itself is not that unusual. Do you really know how much water your washing machine takes to run a load of laundry or how much electricity your DVR uses when you aren’t watching TV? But the idea that everyone should pay the same amount for electricity and water is clearly not accepted. So why should Internet billing be different?
If we look at common thought at the beginning of the nuclear power age, many experts predicted that electric meters would become extinct. The cost of Nuclear power would be so cheap on a per unit (KWh) basis that measuring it would be pointless. Sadly this vision was not realized. However, it is true for today’s Internet service. The cost of Internet bandwidth is in creating and maintaining a total capacity. The 95th percentile billing used for commercial Internet billing reflects this reality, in its own limited way so does our current residential billing practice. At the lowest tier dial-up providers are still providing low-cost, low-capacity service. As you move up the chain you can find different rate plans from broadband providers for different levels of service. So why has there been a backlash against metered billing plans? Robert Charette offers insight into the issue.
Time Warner’s plan would have charged usage not on their peak consumption, but on their total consumption. This means that Time Warner would not be billing on what it cost to deliver the service, but would be profiting excessively from their most dependent customers. Additionally, at a time when margins are still good on broadband service, their plan was to make more money off the heavy demand customers without passing the savings down to the casual usage customers.
We need to understand that metered billing can be very useful. By aligning metered billing with the true cost of delivering a service we can shape customer behavior through self management instead of using artificial means. Take as an example telephone billing in the past. Usage costs for residential long distance were higher during business hours but cheaper in off peak time. This resulted in users moving their personal calling to off peak times freeing up network capacity for business calls. In this case, the metered billing accurately reflected the age old business principals of supply and demand.
Metered Internet billing should only be applied in the same way. Currently some ISPs use traffic shaping to artificially constrain use during peak demand time. They are forced into this because the user’s have not been induced through the rate structures to either limit their bandwidth usage or to move their activity to off peak time. It is fairly easy to create tools to measure usage and display costs in real-time taking into account peak and non-peak rates. Once presented with that information it would be fairly easy for the users to alter their behavior, moving their file sharing to off peak hours for example. The biggest reason this is not done, in my opinion, is because the 95% of the users who are currently subsidizing the 5% who use 50% of the bandwidth would find out they are being overcharged for their service.
I would ask you to remember the “web hog” commercials of the early 2000’s and the implications that the network didn’t have the capacity for all the activity. With the age of Internet video having arrived it is important that we provide a monetary incentive for ISPs to provide adequate bandwidth yet avoid the pitfalls of the network neutrality mess and ensure that people are paying their fair share. Correctly implemented metered billing accomplishes all of these goals.
In conclusion, a backlash against Time Warner’s plan was a good thing, however metered billing can be a good thing and the correct implementation should be welcomed. However, such billing needs two things to be successful. It needs visible metering so that user’s can conceptualize how their consumption is related to their cost and it needs to be structured to provide an incentive to modify behavior and thus optimize overall network usage.