THE ECONOMICS OF INTERNET ACCESS (The Sequel).
Source: Mckinsey
TELECOMMUNICATIONS
The last mile
to the Internet
Jed Dempsey, Guido Frisiani, Rishabh Mehrotra,
Nagendra L. Rao, and Andrew O. White
THE ECONOMICS OF INTERNET ACCESS
The basic components of the business
system that delivers Internet access to
consumers are the national transportation
backbone; the local point-of-presence (POP)
infrastructure with access lines, servers,
routers, and modem banks; and customer
service and support. Many players interact and
compete in complex ways in this space, and
the relationships between them are evolving
rapidly. Four key roles can be identified:
Backbone providers (notably UUNET and
Sprint) own the backbone infrastructure and
act primarily as wholesalers, but also sell
directly to end users, particularly medium-sized
and large corporations.
Large Internet service providers (ISPs)
(both national players such as EarthLink,
MindSpring, and MCI, and regional and
local players such as Erol?s, FlashNet, and
the RBOCs) have relationships with 100,000
or more end users, and either purchase
capacity from others or own local or regional
infrastructure (in which case they often also
act as wholesalers to smaller ISPs).
Small Internet service providers are
local ?mom and pop? operations that
typically provide dial-up access to a few
thousand customers.
Online service providers (such as
America Online or MSN) provide an integrated
offering of Internet access and exclusive and
proprietary content.
The Internet access industry as a whole has
lost money ever since its inception. In 1997,
the losses amounted to some $400 million on
revenues of roughly $6 billion. Although AOL,
the leading online service provider, reached
profitability in 1997, and small ISPs were
profitable on average, backbone providers,
and in particular large ISPs, posted heavy
losses. There were five main reasons for this:
The limited penetration of the Internet.
In 1997, penetration stood at about 23
percent of US households ? too small a
number of users to pay for fixed infrastructure
and G&A costs.
The rapidity of growth. This pushes up
acquisition costs to $60 to $100 per customer
(sums that are put down to expenses rather
than capitalized), and makes for the inefficient
use of network capacity, with additions coming in lumpy increments that force players
to spend ahead of demand.
The competitive pricing pressures and
the resulting ubiquity of the ?all you can eat?
flat-price model, which allows a small number
of customers to abuse the system and
consume a huge chunk of capacity without
bearing the cost. Among AT&T WorldNet?s
subscribers, the 3 percent of heaviest users
consumed over a third of the total capacity
until restrictions in the form of usage-based
fees were introduced in late 1997.
The immaturity of revenue sources,
with transaction and advertising revenues
per online household lagging far behind the
levels achieved by media such as cable TV
and TV home shopping. This is despite
superior demographics, better targetability,
and continuing increases in time spent
on line (AOL?s customers spent up to 48
minutes per day on line in 1997 compared
with under 20 minutes in 1996).
The high customer churn, with well over
40 percent of subscribers switching provider
in the course of a year. By contrast, even the
highly competitive world of long-distance
telephony experiences churn rates of only
20 to 30 percent.
Over the next two to three years,
improvement can be expected in all five
of these problem areas. By 2000, Internet
penetration may have reached 33 percent of
US households, while growth is likely to have
slowed considerably. Pricing pressures have
started to ease, with AOL increasing its
rate from $19.95 to $21.95 per month and
AT&T WorldNet limiting ?all you can eat?
to 150 hours per month to stop abuse.
Non-subscription revenues have started to
increase (advertising revenues per online
household rose from $17 in 1996 to $39
in 1997), and churn rates are falling as
competition stabilizes and providers invest
in customer support.
When taken together, such changes could
make the average player profitable by
2000. But industry structure will be critical,
with a few players capturing scale and
scope economies and pricing levels being
determined by the extent of fragmentation
or consolidation in the industry. |
