Surprisingly, the modern economy of Internet services is just as vulnerable. Digital vendors selling intangible wares such as IP telephony, videoconferencing, public key encryption, digital whiteboards, e-mail and text are beset with precisely the same problem: how to quickly and accurately calculate prices for digital goods consumed.
For a solution, digital vendors must turn to a key aspect of their Internet billing system-the rating capability. The rater takes usage input (sometimes represented by a network detail record) and calculates pricing based on a rate table or predefined set of rules established by product management. Without this per-transaction rating capability, product managers must offer services at a flat rate.
To date many IP service providers have settled for flat cyclical fees in lieu of usage rating. For example, consider an ISP’s core service, Internet access, which is often priced flat-rate as a function of the access speed (e.g., a monthly rate of $19.95 for dial-in, $1,500 for T1 and $30,000 for T3 access), not usage.
No longer. The IP platform can now offer secure, QoS-based service suitable for on-demand voice, video, application hosting, fax, secure remote access, content distribution or any just about any digital application. With the brutal competition in the flat-rate consumer markets, digital service providers are compelled to offer these premium services to sustain their business. Unlike access, premium services cannot be profitably priced at a flat rate; they must be priced by usage.
Consequently, rating goes from obscurity to the spotlight within the billing system. Without a robust, flexible rating infrastructure, IP service providers cannot capitalize on IP protocol advancements to offer and bill for the premium services mentioned above. This article introduces the challenges and issues inherent to rating IP services, and outlines the elements IP providers should look for in an IP rating and billing system.
IP Rating: A Brief System Perspective
Let’s consider the rater’s role within an IP billing system. As illustrated in Figure 1, the rater operates at the core of the billing system. It accepts network transactions from the usage collection subsystem, assigns a price, and provides the rated transaction to the billing database. From there account balances are adjusted, and the records are stored for later use by the invoicing, reporting, accounting, and customer care systems.
Figure 1 also shows the rater’s role in provisioning prepaid service. Here, the rater’s job is to take an account balance and determine exactly how much service a customer is entitled to (so-called “prerating”). Based on this calculation, the provisioning system authorizes the network elements to provide exactly the amount of service that corresponds to the customer’s balance.
IP-optimized rating engines share many of the same engineering challenges facing other telecommunications billing systems. The rater and related interconnection subsystems must be engineered to keep up with transaction load. Otherwise, real-time posting and provisioning capabilities are compromised. Likewise, as a mission-critical mechanism, a high degree of fault tolerance is required. Finally, since it is tightly coupled with the financial system, the rater and biller must record a complete audit history of rating and price change events (revenue assurance).
IP Rating: The Price Definition Challenge
In spite of the similarities to a traditional telecommunications rater, IP transactions present a unique challenge for price definition. Offering competitive services in “Internet time” requires a rater with enormous flexibility and agility. A premium IP service with little or no competition can become a commodity within a few short months. Consumer expectations about pricing also fluctuate quickly, as new competitors enter the market. A critical success factor for businesses in the Internet environment is the ability to rapidly adjust prices to accommodate these influences.
The task of keeping up with fast-changing industry price trends typically falls upon product management. Product managers depend on the rater’s price setup interface to define new, innovative, differentiated services. The price setup interface must satisfy the strenuous, potentially competing demands of the product manager:
Supply all of the power needed to develop potentially complex pricing rules.
Provide intuitive ease of use to avoid confusion and subtle pricing errors.
Permit the product manager to introduce new means of measuring costs.
The interface must also enable the product manager to contend with opposing forces: the consumer desire for low cost versus the business’s desire to keep prices higher than costs, and the consumer desire for simple pricing versus the business’s limited ability to correlate usage to cost.
Typically, businesses wish to ensure that revenues exceed expenses for every service. In order to enforce this magic formula, a business must first know how usage correlates to cost. This is the first hurdle that a business must overcome in order to intelligently manage pricing (covered in greater depth in the next section, “Selecting Pricing Criteria”).
Even if exact costs are known, the product manager must strike a balance between the other forces described above. Consumers prefer simple rating and package plans. They like to be able to calculate their expected prices in advance. However, overly simple prices are a formula for network saturation in peak periods and unprofitable operation.
Customers also want to pay as little as possible overall. The product manager is responsible for making critical business decisions as to what losses are acceptable in order to meet the marketing objectives of simplicity and competitive appeal. The price setup interface must empower the product manager to bridge this gap between actual costs and consumer expectations by defining a simple, appealing and profitable rate structure. The sidebar example illustrates this principle.
Let’s consider a simple example where costs correlate directly to the time of day of usage, with no other significant influences. The cost curve can be seen in Figure 2a.
The simplest possible price plan is a flat rate. A seemingly reasonable flat price is shown in Figure 2b. This rate appears to be profitable for 80 percent of the day, with significant losses only from 1 p.m. to 3 p.m. A well-informed product manager might make the perfectly reasonable decision to incur losses over that period. Without adequate information, however, this decision could prove fatal. Suppose that 50 percent of daily traffic occurs during the period 12 p.m.-3 p.m. This means that the entire service will be unprofitable, even though it is priced right for 20 hours per day!
Figure 2b: Potentially unprofitable flat-rate price plan.
A better rate structure is shown in Figure 2b. Pricing is slightly more complex, but guarantees profitability for the business at all times of day. Furthermore, it features 8 hours (4 p.m.-12 a.m.) at a lower price than the flat rate in 2b.
The right data and the right price setup tool enable product managers to define a rate structure that gives customers the lowest prices
guarantees profitability
is simple enough to please the customer.
IP Rating: Selecting Pricing Criteria
In the sidebar example, the sole determinant of cost was time of day. Real life is never so simple. Every new IP service offers a unique set of attributes that may be used for pricing. Examples include bytes, bandwidth, quality of service, security level and recipient count.
To illustrate this point, consider a provider that wishes to tier pricing for video service based on session length and QoS provisioned (e.g., “$1 for the first minute, $0.50 each minute thereafter”). The provider may also wish to increase or decrease the price as a function of the actual size of the video-stream played. Thus, the basic video service might be defined to support low-end videoconferencing (e.g., 384 Kbps throughput, with a guaranteed end-to-end delay of 100 ms and a loss rate of less than 5 percent). What if a customer wants a higher resolution? Or a faster frame rate?
Here, a provider may wish to define this service and charge based on actual bandwidth used above the base rate. Likewise, the provider may wish to discount the service if the customer sends at a lower rate, or if the QoS delivered violated the service level agreement (SLA).
Such pricing is impossible unless the high-resolution cost metrics are exposed in the rating interface in such a way that the product manager can incorporate them into a pricing plan. Without a high-resolution costing interface, it is impossible to capture the true cost per transaction.
To further illustrate this point, consider the simplest of all IP services: e-mail. You might think that direct costs of e-mail are negligible, so that a simple flat-rate model would suffice (e.g., $5 per month for all the e-mail you can send). In fact, the pertinent cost metrics are far more complex, involving such elements as:
Message size: The “Hi, how’s it going?” message represents negligible bandwidth and storage overhead. Here, a flat rate makes good sense. The “Hi, check out my latest PowerPoint presentation” attachment that is 10 MB long is a completely different story. The overhead of the latter message is equivalent to at least 10,000 of the former.
Distance: Delivering a message within the office is not a big deal. Delivering a message to India, however, incurs the cost of traversing an expensive transatlantic link and interconnecting with an ISP that is under regulatory constraints.
Disk space: If a user doesn’t check e-mail for a while, it simply spools up on the mail server. Disk space is cheap, but if large messages spool up and consume 100 MB, the cost to the service is no longer negligible. QoS: The cost to deliver a message does vary with the time of day. For example, it is less expensive to deliver e-mail in the middle of the night than mid-day, when demand for network resources peak. Would a user be willing to pay more to have messages delivered instantaneously?
POP3 queries: E-mail is stored on a server until a client (typically using the POP protocol) asks for it. Each query consumes server and network resources. The period at which a POP client checks for new e-mail is configurable, and often clients are configured to check every minute, or less, resulting in several thousand queries per day. This places a huge load on the server, even if a single e-mail was not sent or received.
With such complex cost metrics, how can a provider price e-mail service at a flat rate? No doubt, most providers are actually losing money in some instances. The containment strategy is to control costs by restricting access to resources. For example, ISPs may limit message sizes to 1 MB, assign a 5 MB e-mail disk quota and schedule the transmission of messages based on size and current load. Likewise, they actively search for spammers and terminate their service. These parameters are adjusted so that the costs for most customers fall below the flat rate collected.
By bounding usage, the carrier controls its exposure but foregoes potential profits. For example, what if e-mail customers are willing to pay extra to have larger disk quotas, send huge messages, get instantaneous delivery, avoid spam [<
To fully capture this available revenue and delight the customer, the rater must be able to support flexible pricing capabilities and expose high-resolution cost metrics in the rating interface. Otherwise, the provider risks losing high-margin incremental revenue and alienating customers who desire services outside the base packages.
Additional Sources of Rating Complexity
Thus far, we have discussed the selection of pricing criteria, the price setup interface, and the business decisions facing product management. These issues constitute only the most basic challenges of IP Rating. A large number of other variables can contribute to the complexity of the rating calculation, including:
Taxation-Application of taxes may affect the base rate. The applicability of telecommunications tax law to IP services is still being decided in the courts and legislatures.
Cross-service discounts-Product managers want to discount rates of one service based on usage of another.
Comparison rating-A competitive advantage in pricing can be driven home by showing customers how much they saved relative to another published rate.
Customer hierarchies-A customer account may be affected by the pricing or discounts of a corporate sponsor.
Incentive programs-Some fraction of a customer’s charges may be redirected to the VAR or salesperson who acquired the account.
Multiple currency-Multinational offerings, or offerings made to multinational corporations, may require rating in multiple currencies and/or performing automatic conversion.
These and many other potential complications make rating one of the most interesting challenges in the IP OSS arena. A comprehensive analysis of the capabilities of currently available raters and billers will appear in the “Internet Rating and Billing Functionality Report” that Billing World will publish in June.
Conclusion
As new IP services are brought to market, the only constant is change. Each new service offers a unique set of usage metrics to control rating. As each market matures, competition forces providers to rapidly adapt from high-margin, premium-service prices to simpler, high-volume consumer pricing. Service providers who cannot control rating and pricing to change with the technology and the market will be left behind.
As new protocols are introduced, the future of rating can only grow more complex. Future protocols might support non-determinant rating: rates negotiated in real time between the user and the provider. Some day soon, your computer may able to instantaneously switch from one carrier to another when it detects a lower price.
Billing designed for usage-based services with simple cost components are readily available. However, many of these systems won’t extend to the fast-changing IP environment without substantial redesign of their rating architecture. Billing systems that don’t address the unique cost components of IP services-and provide a flexible, accurate mechanism for product managers to develop new services quickly-may become an Achilles’ heel.
To plumb the challenges and solutions of IP optimized rating, provisioning and billing, visit the half-day workshop conducted by the authors at the Billing’99 trade show in Dallas, Texas, June 22-28. A detailed conference agenda is available on page 57.
