What Took You So Long to Call?

When Linda Manchester gets stuck in a really dismal day, she sneaks over to her computer and opens a special folder called "nice words." Inside are dozens of archived email messages from customers in California, Texas, and other states.

The notes are instant cheer-me-ups -- full of phrases such as "very, very impressed" and "great company." Look closely, though, and something striking emerges. Most of the upbeat emails have arrived recently. There's almost nothing from before early 2000.

Ironically, Manchester put in some of her hardest work in 1998 and 1999, when no one from the outside was cheering her on. She has built a 19-year career at Lucent Technologies, where she is now VP and general manager of the MultiVoice division. It is her mission to help develop phone technology that handles voice calls as if they were packets of Internet data. But getting there -- for Lucent and for everyone else -- has been an ordeal.

After years of stumbles, it looks as if Internet telephony will finally make it into the mainstream. Why has such an exciting and powerful technology taken hold so slowly? Three big factors stand out. Each has proven surmountable over time -- but only if people know what they are up against and develop appropriate strategies. The effort to overcome each of these obstacles contains a broader lesson for anyone who is championing new technology to transform a major market.

Lesson #1: Brace yourself for many months -- perhaps years -- of extra work, until that nifty lab prototype works reliably in the field every time. Ask Howard Balter, CEO of Net2Phone, a Newark, New Jersey company that has been promoting Internet phone service for the mass market since the mid-1990s.

"In 1995," Balter recalls, "we were all getting excited about the Internet." He was a senior executive then at IDT Corp. in Newark, which developed Net2Phone as a subsidiary. "I saw what people were doing with voice clips, sending them around the Internet, and I asked our engineers, 'Why not do the same thing but on a bigger scale, allowing people to make voice calls from PC to phone?'" When his engineers said that it might be hard to do well, Balter told them to go ahead anyway.

His company devised software and services that allowed people to place dirt-cheap phone calls by using their modem connections. But signal quality was so shabby that it sounded as if callers were under water. What's more, 20% to 30% of calls were being disconnected midconversation, because various data packets that contained snippets of the call were getting lost somewhere on the Internet.

In a controlled IP network, Internet telephony performed much better. But there, it was possible to control packet creation, transmission, and reassembly. On the public Internet, it was a lot more difficult to guard against transmission snags and delays.

The remedy for many problems has been better signal-compression technology. When Net2Phone got started in 1995, it was using what is now regarded as a pretty primitive "codec," or system for coding and decoding speech into digital packets. After several upgrades, the company is on its fourth-generation codec and says that performance is much better.

Lesson #2: Woo customers patiently and persistently -- because inertia won't give way quickly. Big businesses aren't eager to rip out existing circuit-based voice systems, says Bob Hafner, an analyst with consulting company Gartner Inc. That's especially true if they have recently done major upgrades and have not yet recorded the value of those assets completely.

Seemingly minor issues can slow down customer acceptance too. Can Internet-based systems interact well with the 911 network? Will big savings on calling rates be undermined by steeper costs for new handsets that just sit on people's desks? Does the decentralized nature of the Internet mean that telecom equipment will be dispersed to far more locations throughout an enterprise, instead of being clustered in one communications center? If so, who is going to maintain it?

In such situations, there is an art to picking out the customer segments that are most amenable to trying something new. Universities, for example, are some of the earliest adopters of Internet-based calling systems, because they are rewiring their entire campus anyway to provide high-speed Web access.

A year or two ago, Lucent thought that it had found the right early adopters for its VOIP equipment: young upstart phone companies that were taking on the established giants. "They spent very little time kicking the tires," Manchester recalls. "They just wanted to put in the equipment and start driving revenue." That was great while it lasted. But many of those companies ran up unsustainable losses and went bust or had to retrench. So now Lucent is concentrating its efforts on the Bell operating companies and other dependable -- but harder-to-please -- members of the telecom carriers' old guard.