Crystal Balls and Other Forecasting Imprecision


How can you predict what size the Web is going to be when
there are so many contradictions about how big it is today? Three years into the
mass-marketing of the World Wide Web, there's no definitive gauge, despite dozens of
measurement services.

What you believe may depend on who you believe when it
comes to measuring the interactive audience.

Year-end tallies generally agreed that about one-fourth of
U.S. homes (give or take 5 percent) have access to the Internet. That translates into
about 24 million households, up more than 30 percent during 1997. (Barely 0.5 percent
accessed the Internet via cable modems.)

One thing that the census-takers agree upon is the fact
that the upward trend will continue. They also acknowledge that high-speed is
'in.' But the variations in user-base estimates may come back to haunt the
industry -- especially since today's numbers are the launchpad for predictions about
the online interactive audience of tomorrow. Product and technology decisions are being
made based on these data.

For example, Find/SVP's 'American Internet User
Survey' identified 57.4 million adults who log onto the Internet; this study looks at
individuals, rather than households, so its total user base represents about 21 percent of
American adults. Among Find's interesting factoids: The Web's gender tilt has
almost been leveled -- 58 percent of Internet users are male, and 42 percent are female.
However, this study also identified 15.9 million people who had tried the Internet in the
past year but who are not currently using it -- an interesting first sign of churn and Web

Relevant Knowledge Inc.'s Web census comes up with
55.4 million Americans over age 12 using the Web at year-end. Its age segmentation put
'38 percent of all Web users' in the 18- to 34-year-old category.

Meanwhile, Decisions Analysts Inc. puts the number of
Internet households at 21 million, but it estimates that only 28 percent are in the
18-to-34 bracket; it calculates that nearly 57 percent of Internet households are in the
35- to 54-year-old category -- probably parents of Web-heads, as well as surfers

Other head-counters are looking for trends. IntelliQuest,
for example, found a shift away from commercial online companies. The IntelliQuest data
show that in early 1997, 70 percent of users first accessed the Internet through a service
such as AOL, Prodigy or Microsoft Network; by year-end, only 61 percent of surfers were
using such commercial online packagers. IntelliQuest says that when asked what type of
provider they would prefer, current and would-be users first mention an ISP, followed by
their local telephone company and then a commercial online service. Iconoclast offers a
low-ball estimate of 39.6 million North American online users, and it includes ethnic
breakdowns (about 1.4 million African-Americans, 1 million Hispanics and 1 million
Asians). Other services -- such as Net Metrics, NPD Group, I/Pro, Odyssey Partners and
Forrester -- also publish usage data.

You can take your choice among any of these estimates. Or
you can use other gauges about the status of the Internet today. For example, Network
Solutions Inc., which has been the sole registrar of Internet domain names (a monopoly
that is about to end), accepted 960,000 new sites in 1997, almost double the 489,000
domains set up in 1996.

If electronic commerce is the preferred yardstick, the Web
generated well under $1 billion in sales last year -- almost insignificant compared to
other shopping avenues. Cable and broadcast teleshopping (including infomercials) does at
least five times that volume, and direct mail/catalogs are 100-fold larger. Not to mention
actual brick-and-mortar retail sales, which are in the hundreds of trillions of dollars.

Despite the questionable and inconsistent numbers, Web
optimists are eager to apply their exponential growth curves to these data. Some of their
calculations show Internet audience penetration in the 50 percent range and transaction
volumes in the $5 billion to $6 billion region within four years.

If you work the math right, you can get to those numbers.
Of course, if you let logic get in the way, the size comes down fast.

For one thing, as you climb that bell curve past early
adopters and early followers, the slope gets steeper. There are few truck-chasers in this
race. The next 10 million or 20 million prospects are likely to be less tech-tolerant than
the first cohort of users. Price may matter more, and these late adopters' appetites
for electronic commerce may be thin.

Such factors will have a big impact on Internet users and
usage -- far greater than price competition of $30 or $40 per month for cable-modem or
telco digital subscriber line high-speed access. And, of course, there's the basic
measurement issue: Where will the new customers come from as high-speed access
proliferates? Maybe the migration from today's platforms will just level off the
frenzied growth.

What will you do with all of the rosy predictions? It
depends on who you trust for your forecasts.

I-Way Patrol columnist Gary Arlen gets his Internet numbers
the old-fashioned-way: guesswork, plus a slide rule.