IP Telephony: Realities and Billing Opportunities

The number one myth circulating in the telecommunications industry is that IP telephony will replace circuit switching. Last month’s Publisher’s Letter hopefully debunked that myth. This month’s letter focuses on the marketplace realities of IP telephony and associated billing opportunities. First, here are my top 10 IP telephony realities:

1. Hybrid Networking Is In: No one network technology can do it all, and old network technologies die slowly. Circuit switching was designed and optimized for plain old telephone service (POTS). Except for intra-company or private network traffic, it would be shocking if more than 1 percent of switched voice were transported over a pure, public IP telephony network independent of circuit switches in the next 5 years. So when planning future voice billing systems, think of gathering billing records from both circuit switches and IP gateway or router-based switches.

2. No Free IP Voice: Voice traffic, including in-band data and fax, generates more than 80 percent of the carrier industry’s revenues today. Carriers can’t afford to give it away or allow fax traffic to pass for free over the Internet! Data is and for the foreseeable future will be the differentiator, but voice revenues will carry the day for many years to come. Besides, clever users could find ways of using free voice access 24 hours a day, 7 days a week, over the Internet, thereby saturating networks with traffic. In reality, voice will be part of a data deal, but it will have to be metered and be part of the end user’s bill.

The fallacy about voice-for-free stems from the confusion over profit, revenue and bandwidth growth. Yes, data traffic is growing faster than voice, but revenue and profits are not. Breaking down the cost of a 10-cents-per-minute long distance call over the circuit-switch network to dollars-per-bit yields $26 per gigabit. If you do the same for a T-1 private leased circuit, at 1.5 MBPS it comes out to 25 cents per gigabit, assuming the average T-1 goes for $1,000 per month. If carriers based voice prices on T-1 bit prices, the 10-cents-per-minute call would be 0.1 cents per minute or, in essence, free. Free voice would require T-1 revenues to increase 100-fold, if carriers’ growth depended on IP services. Data is growing in terms of bandwidth utilized, but not in revenues and surely not in profits. Long distance switched-voice is profitable, while leased T-1s in the United States are not. Bottom line: no carrier will offer value-added services for free, even though transport costs will drop to zero.

3. Nobody Knows What Networking Costs: When the FCC was pressed to come up with a proxy rate for the wholesale price for local telephone resale after the passage of the 1996 Telecom Act, “unofficially” the FCC bureaucrats took the discount the CLECs wanted over the ILEC retail prices (40 percent) and what the ILECs thought were avoided costs of billing, customer care and marketing (2 percent), added them and divided by 2 to get a 21 percent discount rate: ((40 + 2)/2 = 21). Almost every state Public Utility Commission (PUC) adopted 21 percent or close to it as the wholesale discount rate for local service resale. Even though the Telecom Act says the states are supposed to come up with a discount rate based on detailed financial analysis or avoided costs, every FCC bureaucrat knew that nobody knows what networking really costs.
IP-centric carriers believe that IP networking will reduce the cost of networking by an order of magnitude. This is mostly hype (see Billing World, January 1999). The reality for IP telephony billing is that the price of service is likely to be based on access, network elements and OSS, and delivered to a customer in plain English on an invoice.

4. Not Everything Over IP: The Department of Defense funded the development of IP in the late 1970s and early 1980s. DoD’s marching orders were to design a platform that could send data over any type of network, regardless of network quality or reliability. However, since the IP telephony hype started, the paradigm has shifted from “IP over anything” to “anything over IP.”
Just as an individual highway is not designed to carry all kinds of traffic (residential, interstate) or vehicles (cars, garbage trunks, bicycles), neither are networks. IP is great because the numbering scheme allows efficient multiplexing and inter networking, but it’s not as good when it comes to quality of service (QoS). ATM, on the other hand, is great for QoS, but not good for multicasting.
IP telephony introduced the concept of efficient integration of voice and data. But other than incoming call center switching, customers don’t care about traffic integration. Billing for IP telephony will have to reflect actual usage, and that means sniffing IP packets to differentiate traffic type, content, and other usage factors.

5. Think Billing Applications – Not Bits or Bytes Transmitted: The thought of presenting users with bills based on the number of bits or bytes transported is ridiculous. First, users think in terms of applications, and that’s what they want to see on the bill – how many calls placed, to whom and for how long. Second, providing low-cost bandwidth is not a viable business strategy. Selling IP network applications is. So carriers need to think billing applications, including voice, right from the get go!

6. Business versus Residential IP Networking Markets: If ever there were a night-and-day difference in markets, it’s with IP QoS and bandwidth management in the business and residential markets. In the intra-company or private networking market, a carrier can control all information flow from an end-to-end perspective. Therefore, Internet service billing can be service-level focused, (that is, QoS-based pricing). As such, it will have to be presented to the corporate user for department charge back. Also, IP telephony will likely be pitched to corporate users in terms like “give us your bills for the last 3 months, and we will show you how we can save you 30 percent.”

The residential market will be different. Here, IP will focus on access. Whether it’s IP telephony over xDSLs or cable modems, as soon as that IP telephony call comes into a telco or cable TV headend, it’s off to a circuit switch (see Billing World, January 1999). Billing will be based on gathering information from various systems, filtering events and correlating these with other data.

7. In Reality It’s About Outbound Telephony: Why would a customer want to use IP telephony technology to process incoming calls? There are some cases – incoming call center traffic, for example. But, in general, IP telephony makes sense only for outgoing calls. For starters, customers don’t pay usage for incoming calls other than toll free, 800 traffic. Second, customers view incoming traffic as more important, and thus service must be more reliable. Finally, IP telephony can’t match the Intelligent Networking (IN) features of today’s circuit-switched telephony. So when designing IP telephony billing systems, think outbound.

8. The Edge Is Where It’s At: The real benefits of IP are for integrated access. For the business user, it’s the elimination of separate T-1s for local, long distance, Internet, Frame Relay and other network access. For the residential user, it’s the ability to send voice and data over xDSLs or cable modems. Regardless of the market, billing systems will have to capture usage data at edge multiplexing devices and correlate that data with network switch data.

9. Think Statistical IP Billing: IP proponents dream of all information (voice, data, video and e-commerce) passing through one access pipe, and of gathering all this data to do convergence billing. Creating such a billing system in one fell swoop would be asking too much. Billing for IP will likely evolve from usage-based data from other networks to statistical-based event gathering using network probes.

10. No One IP Telephony Market: Just as “anything over IP” is unrealistic, so is thinking that all markets are going to be addressed with the same IP networking solution or IP protocol platform suite. TCP/IP is the protocol glue that holds the public Internet together, serving website access, e-mail and e-commerce very well. UDP is the protocol glue for IP telephony and, in general, is not needed for public Internet applications. Wireless IP will require something more than TCP or UDP. Cable IP for video and telephony will drive a new IP approach as well. Bottom line: A single approach to IP billing won’t fit all markets when it comes to IP in general and IP telephony specifically.

The New IP Telephony Billing Challenges and Opportunities

Not all IP markets are the same, so expect different near-term carrier requirements for billing, customer care and other OSS. Here are the major strategies, capabilities and challenges for carriers in this new IP world:
1. Established IXCs: The big three, AT&T, MCI WorldCom and Sprint, have basically identical backbone network architectures for data and dedicated circuits: ATM/SONET with ATM/Frame Relay at the backbone edges, along with IP networks. They also have “circuit switch centric” mentalities and believe IP telephony makes sense only for Intra-company IP networking (Intranets), residential xDSLs or cable modem access. Of course, their CEOs sing praises to IP telephony in public, to keep their companies’ stock prices high. Yes, the established IXCs have their internal IP telephony camps, and, yes, they are spending development dollars. But the established IXCs are generally skeptical about IP telephony but bullish on IP. Here’s how they stack up in the IP revolution, regarding marketplace strengths and weaknesses: AT&T’s strength lies in having the most POPs (points of presence), sites where they meet ILECs and CLECs. The more sites a carrier has, the less costly access is for users. My guess is that AT&T has more than 1,000 POPs in the United States, and it is only a few miles away from 80 percent of U.S. business users. In addition, with AT&T’s acquisition of TCG, a local fiber CLEC, AT&T probably has direct fiber access to roughly 10,000 buildings in the United States. Not as good as MCI WorldCom, with 40,000 in the United States and 10,000 in Europe, but still better than Sprint. AT&T also has a major presence in the residential ISP market (more than 1 million subscribers), whereas Sprint and MCI WorldCom have relatively few. Finally, with the pending purchase of TCI, AT&T will have more than 10 million cable subscribers and about 12 million partially controlled subscribers through other TCI joint ventures. Along with its stake in @Home (an Extranet for the cable industry), AT&T would play a major system operator role for other cable operators regarding IP telephony access.

MCI WorldCom has a very strong presence as a facilities-based CLEC. Its strengths lie in its role as an IP backbone provider or Network Service Provider (NSP) to other ISPs, its peering relations with other NSPs and its corporate Internet access user base. Its weaknesses, by choice, are the lack of residential presence in IP compared to AT&T and its lack of mobile wireless presence -- which won’t be important in the IP game for several years to come.

Sprint’s strengths are its IP NSP presence regarding ISP customers served, its peering relationships with other NSPs and its corporate IP access customer base. Sprint’s weakness is its small residential IP customer base, which, like MCI WorldCom, is by choice and may not be important in the consumer market for years to come, if ever. However, its lack of a strong facilities-based presence regarding fiber access to buildings has forced Sprint’s Integrated On-Demand Network (ION) approach.

The original ION concept called for Sprint to lease xDSLs and fiber access from the RBOCs and CLECs. A December announcement indicates, however, that Sprint now plans to form a separate business unit to install co-location cages in more than 1,000 ILEC COs in order to get direct access to unbundled loops, as today’s data CLECs do. MCI WorldCom has similar plans for gaining access to unbundled loops under its NSP operation, UUNet.

Big IXCs and IP Billing: Except for the residential market, where AT&T will likely dominate with its cable partnership and ISP presence, the business market IP billing challenge and opportunities are identical for the big three IXCs. Each has a massive circuit-switch infrastructure. Each will move to place ATM multiplexers or edge devices at customer premises or multi-tenant office buildings linked to their POPs. What will they need for commercial success? – An ATM metering device! The plan is to convert dedicated voice-circuited traffic, IP and Frame Relay packets to ATM cells at the customer premises. A supercomputer would be needed to store all those ATM cells and create a bill. No workable standards are in place, and no technology for usage-based billing has arrived. So for now, it will be little more than flat-rate access with billing data gathered by other networks.

So What About IP Telephony?

In reality, for the large, established IXCs, IP telephony plays an insignificant role today in the business marketplace. Telecom managers are nowhere near convinced that IP telephony will ever deliver circuit-switched quality, and they see no long term role for IP telephony other than call center, fax over IP and perhaps integrated messaging.

From the established IXC carrier’s perspective, the name of the game is integrated access to gigabit backbone networks, circuit-switched voice infrastructure, frame relay data and various forms of Internets, both private and public. Billing, customer care and OSS challenges will center around ATM, not IP. 2. The New Breed of IXCs: The game plan for the new IXCs, such as Qwest or Level 3, is different. These carriers have money, transmission capacity (fiber) and seasoned telecommunications managers. But they have little else that’s required to create a customer and collect a monthly check for service. They do have, however, lots of ideas on next-generation networks based on IP. Vendors need to look no further for new billing, customer care and OSS challenges.

While the new breed IXCs are IP-centric, they believe in a dumb-backbone network with service provisioning and intelligence located at the edges of the core IP network. They also believe in the eventual phase out of both circuit-switching technology and ATM/SONET. Each carrier brings to the table a different IP vision or, alternatively, a different take on how it is going to eat the established IXCs’ lunch. No new breed carrier focuses more than Level 3 on IP telephony and associated OSS differentiation.

Level 3 has access to nearly $6 billion in cash and wants to create a 21st-century architecture built around a dumb, but high QoS, IP network while establishing an open IP network architecture. The latter point explains why Level 3 spearheaded the creation of the Technical Advisory Council to work out a road map for seamless integration of circuit-switched and IP networks.

This group realizes, and I agree, that (a) Circuit-switch networks will be around for a long time, and IP telephony has to be seamlessly integrated with today’s PSTN infrastructure; (b) Carriers can’t grow circuit switching technology to handle all IP requirements at the desk top; and (c) Carriers can’t add telephony and real-time video applications to existing IP networks that use routers designed for packet data-based applications. In other words, rather than trying to carry IP telephony over today’s multi-service IP networks, create a robust IP architecture for voice, and add other services later.

In no other area of telecommunications are more R&D dollars being spent, and perhaps no other company is lavishing more R&D dollars on next-generation, IP networking. These new IP networks will see 30 percent of total network costs spent in billing, customer care and other OSSs. For Level 3 alone, that’s nearly $2 billion up for grabs. Again, look no further for challenges and opportunities.

3. RBOCs and the xDSLs: The RBOCs want to offer the xDSLs through the regulated business, but want to avoid offering unbundled xDSLs to CLECs or having to resell the integrated xDSL/DSLAM infrastructure at a wholesale price. Putting aside regulatory issues, the bottom line is, the RBOCs only wish to offer xDSL bundled with their own ISP retail services. Even after the regulatory dust has settled, the xDSLs will remain problematic for the RBOCs, particularly regarding profit cannibalization.

First, the data CLECs are offering HDSL connectivity (1.5 MBPS), unbundled loops from customer premises to ISPs, for $100-$200 per month. This is a fraction of what the RBOCs charge for T-1 connectivity ($400-$1,000 per month), which, incidentally, is RBOC-provisioned with HDSL technology. Thus the RBOCs will experience revenue cannibalization.

Second, xDSLs change the way people use the Internet. Users stay online 24 hours a day, 7 days a week. To find the cannibalization issues for the RBOCs, just look in the Yellow Pages. If users have to dial into an Electronic Yellow Page server and spend several minutes to retrieve the desired information, it obviously would be faster to fetch the 4-inch thick Yellow Pages book and let their fingers do the walking. However, online Yellow Page information is available almost instantly when linked via xDSL. So where is the RBOC cannibalization? RBOCs make more than $5 billion in Yellow Page ad revenue per year, and it’s almost all profit. In fact, Yellow Page ad sales are the most profitable business activity RBOCs have going today, revenue, which the xDSLs threaten to cannibalize.

Third, it’s inevitable that the ITSPs and data CLECs will use xDSLs for switched-voice access via IP telephony to connect to long distance networking services. That’s a 2.5-cents-per-minute access charge not passed on to the RBOCs.

Bottom line: For the RBOCs to make money or, more to the point, avoid lost revenue, they will have to do IP networking and usage-based pricing with the xDSLs. Even if the xDSLs are used only for public Internet access, the RBOCs will have to go to usage-based pricing. Why? Just look at a typical dedicated Internet access user day:

TeleStrategies and Billing World share a T-1 circuit to our Internet Network Service Provider (NSP). We pay Bell Atlantic $400 per month for the T-1 and $1,800 to our NSP for unlimited 1.5 MBPS access. Residential users are not going to spend $2,200 per month for Internet access. The xDSLs can reduce the T-1 charges, but what about the $1,800 NSP charge! Besides, if carriers offer unlimited, flat-rate, high-speed Internet access to consumers, people will find ways to saturate that access link 24 hours a day. Bottom line: In the long run, the xDSLs will have to be metered, and information (content) will have to be differentiated from plain access or IP telephony in order to develop the xDSLs into a profitable venture.

4. Cable TV, IP Telephony and Billing: The cable industry is hot on IP telephony. A number of standardization activities are headed by Cable Labs and others. A number of cable systems will start offering IP telephony in 1999; the first is likely to be Videotron in Quebec. No other industry segment has its act together on IP telephony more than the cable industry. IP telephony billing, customer care and other OSS developments will likely be dominated by the routing vendors and their billing vendor partners. Cable companies, as in the past, will look for service bureau or perhaps AT&T for IP telephony billing.

5. Mobile Wireless Carriers and IP: As mentioned in last month’s Publisher’s Letter, its going to take some time before IP technology is pulled into the wireless world. The techniques used for saving bandwidth (low rate codes and silence suppression) are already being implemented without the need for IP multiplexing. Also, high-speed, packet data and value added services are several years away, along with new-generation data terminals designed for fast processing, small size (palm size) and low cost. But this is the number one opportunity from IP telephony for billing, customer care and OSS vendors in the long run. This revolution goes by the name of network computing. post-PC devices or information appliances. But no computer vendor, except perhaps Microsoft and Intel, believes that the cost of a PC can come down to $100, so a new approach is needed for the masses. The computer people believe that the intelligence must be put into the network, since the post-PC devices will have limited functionality in order to realize a new billion-dollar marketplace for terminals.

While many of these devices will be wired into a home LAN or telephone line, many will also be wireless and will use speech-recognition processing. Wireless IP telephony devices will use common radio carrier frequency bands, creating opportunities and challenges in billing, financial clearing houses, customer care and OSS.

Bottom line: IP will drive network architectures in the 21st century, and IP telephony will be important. But the transition from circuit-switching to IP will be evolutionary, not revolutionary.

Drs. Jerry and Matthew Lucas will conduct a seminar on IP telephony and integrated access for non-engineers, March 1-2, 1999, in Washington, D.C., and April 28-29, 1999, in San Francisco, California. Call (703) 734-7050 or check out www.telestrategies.com for more information.