Last month, Qwest joined a distinguished list of service providers that have tried to introduce some form of usage-based pricing for data services. Pundits immediately pointed out that network managers have resisted any shift away from traditional fixed pricing.
You may wonder how these carriers could be so dumb as to try something that has always failed. Well, maybe they know something.
The truth is, there are two different network markets. One market consistently rejects usage pricing and probably will forever; the other market is eager to embrace usage pricing. So depending on which market you happen to survey, you either believe that usage pricing is culture's only salvation or that it's the last bastion of international communism.
So far, the problem has been that the "we hate usage pricing" market is made up of nearly everyone who currently buys data network services. Traditional networking is supported by a network operations group that has a budget. This budget is based on the total cost of all network projects. It's a fixed budget, and network managers are accountable for it.
This is where the usage-price phobia comes in. If you have recently justified a specific cost with your carefully crafted plan, the last thing you want is for that cost to suddenly increase. Network users have an ugly habit of, well, using. If usage is what determines cost, using the network could make costs mount. Fixed pricing is safer.
Network managers who administer private enterprise networks are the easiest network users to find and survey. Thus, surveys always show a resistance to usage pricing. But this group isn't the only market for data network services.
How about the other market - the one in favour of usage pricing? For that, we have to look earlier in the evolution of network cost, to the point at which those network projects were first proposed. The typical network project starts with a business need that has a dollar benefit associated with it. Planners develop a network to the project's requirements and develop a cost figure for the network. If the benefit is significantly greater than the cost to meet corporate requirements for return on investment, the project gets approved.
This process has two shortcomings. First, it takes a lot of time to complete. That tends to disqualify using data networks to solve business problems or seize opportunities that spring up unexpectedly and require quick action. Second, the equipment part of network cost comes in the form of depreciation, which spreads the buying price of the net equipment over a multiyear period. If the benefits don't last as long as the depreciation does (for example, if a one-year crash project involves buying equipment depreciated over three years), users end up still paying for the equipment long after the project ends. This scenario, in short, only works for a few core applications - exactly what we tend to see running on private networks today.
Usage pricing would solve this problem. Give a buyer a network service that starts billing when you start sending and stops when you stop, and you have the perfect response to one of the short-term problem/opportunity applications of data networking.
Buyers report that they regularly disqualify data network solutions to business problems based on a combination of start-up delay and life cycle. In 1998, the total value of disqualified projects was almost 25% of the world's total data service revenue, according to my poll of 119 users worldwide. But even that figure understates the loss to service providers and businesses because almost 90 percent of all business problems that are too tactical for today's data networking aren't even presented to planners. The truth is, if we could get usage-priced, expense-style data services, we could create as much demand for data services as we do for voice.
Senior management and management planners in most large businesses know how important usage-priced data services could be to them, and that's why service providers keep coming around to this issue. That Qwest has done so at this point only validates that there's another opportunity out there - and a big one.
But we've got a long way to go, even with Qwest's pricing plans. Like others before it, Qwest has delivered usage pricing in the context of switched virtual circuits (SVC), and those killer applications don't support anything other than good old connectionless IP. Making SVCs useful in IP applications requires integrating special equipment with the service. That hasn't worked yet, either.
Usage pricing, from Qwest or others, also depends on the buyer's ability to get a local access connection that's fast and inexpensive. Otherwise, there's no way to get connected to those tactical usage-priced services. Access has been one of the downfalls of Sprint's Integrated On-Demand Network strategy, another attempt at providing flexible services to users. We may need to wait for the access problem to be solved before any form of usage-priced service can be effective.
Nolle is president of CIMI., a technology assessment firm in Voorhees, New Jersey. He can be reached at firstname.lastname@example.org.