Cẩm nang dữ liệu không dây P10 doc

could grow to 62 million by 2000 and 90 million by 2005.”2 Consultants, pundits, and the desperately needy have had a field day with projections of how large, and how quickly, the wirele

10

MARKET OPPORTUNITY

10.1 SECOND ERA OF LOW HANGING FRUIT

Most carrier business planning personnel and hardware providers have long been captivated by the perceived market opportunity for data In 1993 AT&T Wireless projected that “1 billion people will use wireless modems for electronic data communications by the year 2000.”1 Since that was one-quarter of the population of the earth, only seven years in the future, the number was bound to draw attention While that particular projection was patently absurd, for the next few years AT&T Wireless continued to espouse greatly reduced but still aggressive forecasts A 1995 example is:

“Paging and data could grow to 62 million by 2000 and 90 million by 2005.”2

Consultants, pundits, and the desperately needy have had a field day with projections of how large, and how quickly, the wireless data market would develop The peak fantasy year may well have been 1992 With ARDIS, BSWD, and OmniTRACS all operational and CDPD announced, the future never looked rosier Table 10-1 is a representative sample of very large subscriber forecasts made during this period More than half are “unanchored” projections: views of the future

in which the present was unknown, unstated, or unreported Lotus had the steepest slope; its forecast had 10 million in place within three years Jack Blumenstein’s 60 million for the year 2000 was later sustained by AT&T’s Kendra VanderMeulen The next four projections were richly similar to each other In 1994, when counterevidence was clearly available, BIS Strategic Decisions, after a prior, conservative 1993 forecast, joined the high flyer group in 1994, just as they began to waver in their convictions

Table 10-1 does not contain information that really helps one plan a business Given the vintage, the large quantities were almost certainly emotionally driven Key

ISBNs: 0-471-31651-2 (Hardback); 0-471-22458-8 (Electronic)

phrases flag such sentiments: “This is going to be an explosive market”3 or “mobile data will undergo explosive growth in the next decade” (do I detect a cliché?)4 or

“it has amazing growth potential.”5 There was also serious self-delusion: “Suddenly, the wireless market is taking off” (really?).6

By comparison, Table 10-2 is a model of prudence, even though no two forecasts agree But hidden in this seemingly random blur of numbers are useful insights into how the market has developed over time

With a few simplifying selections, an interesting view of the changing opportunity projections is shown in Figure 10-1 The IBM/Motorola enthusiasm of 1982—the first era of low hanging fruit—had barely dimmed seven years later with the ARDIS Red Book forecast But with each successive annual disappointment, the then-current install position was recalculated; the slope of the curves became less intense The

“realistic slope” line, suggested from this continuous adjustment of the installed base, turns out to be lower than the seemingly too-conservative Forrester Research projection of 1992

When will the ramp become a curve? In early 1998 Yankee Group believed a marked upward turn would begin in 1999 It had earlier predicted a truly “mass market” by 2002.7 Through the years Yankee Group has had problems with its forecasts Initially, there was the usual tendency to overestimate the current installed base But as late as 1996 Roberta Wiggins, Yankee Group’s Director, Wireless & Mobile Communications, stated, “I believe there are as many as 500,000 CDPD devices operating in pilot today.” This claim was later retracted.8

Table 10-1 Representative early subscriber projections (high)

Year of Estimate EstimatedUser Base

Projection

Lotus Portable Computing Group a 1992 350,000 1995 10,000,000 Jack Blumstein: ARDIS CEO b 1992 Unknown 2000 60,000,000

K VanderMeulen: AT&T Wireless c 1995 Unknown 2000 62,000,000

Arthur D Little (Cliff Bean) e 1992 <500,000 2000 13,500,000 PacTel Cellular f 1992 Unknown 2000 13,000,000 Yankee Group (Roberta Wiggins) a 1992 500,000 2000 13,000,000 BIS Strategic Decisions h 1994 Unknown 2000 15,000,000

10.1 SECOND ERA OF LOW HANGING FRUIT 145

Table 10-2 Representative subscriber projections (moderate)

Year of Estimate EstimatedUser Base

Projection

IBM/Motorola ARDIS

International Data Corp a 1991 600,000 1997 5,000,000 Forrester Research b 1992 <140,000 1993 147,210

1994 640,640

1995 1,188,480

1996 1,841,120

1997 2,625,000 Decision Resources c 1993 600,000 1997 5,000,000 Link Resources d 1993 Unknown 1998 4,208,000

BIS Strategic Decisions f 1993 1,000,000 2000 3,900,000

2003 5,600,000 Business Research Group

Business Research Group

1999 3,000,000

2000 6,000,000

2001 8,600,000

Figure 10-1 Shifting subscriber projections (medium)

In 1992 Yankee Group projected approximately 13 million users by the year 2000.

A year later the projections dropped to less than 10 million In 1995 their view altered still further9: “Why aren’t more businesses willing to transmit data over wireless networks? They have little business need Widespread demand for wireless data communications will take time to develop.” By 1998 the forecast for the year 2000 was down to 6,400,000.10 A sample of these forecast revisions is shown in Table 10-3

But Yankee Group is not simply creating forecasts that support a time-delayed ramp They believe that while the startup has been disappointingly slow, a huge opportunity is still there Subscriber growth rates are about to “hook” to realize that opportunity This is shown in Figure 10-2 with three representative Yankee Group forecasts

Table 10-3 Sample Yankee group projections

Year of Estimate EstimatedUser Base

Projection

Communications Week, 10-26-62 1992 500,000 1995 5,000,000

2000 13,000,000 ARDIS Lexington Conference,

1995 3,200,000

1996 4,600,000

1997 5,900,000

1998 7,300,000

1999 8,600,000

2000 9,900,000 Information Week, 2-3-97 a 1996 1,900,000 1997 2,750,000

1998 4,000,000

1999 5,800,000

2000 8,400,000

2001 12,180,000

2002 17,660,000 Global Wireless, Mar.–Apr 1998 1996 1,600,000 1997 2,200,000

1998 3,100,000

1999 4,600,000

2000 6,400,000

2001 9,000,000

a year over the next 5 years, reaching 10 million by 2002.” Note: a CGR of 45% does not produce the Yankee Group value.

10.1 SECOND ERA OF LOW HANGING FRUIT 147

10.2 SOME UNPLEASANT HISTORY

10.2.1 Job-Based Market Opportunity Approach

In 1987 a unique study of the U.S data radio opportunity was completed.11 This report had wide circulation, but its conclusion that the data radio market opportunity was quite finite was not lovingly received

The approach was to build a detailed database of occupations, with principal source information obtained from the BLS Occupation Industry matrix This information was supplemented by census data to make self-employment estimates The 579 detailed occupation codes were initially compressed to 408 categories; three sensitive occupation codes were expanded to 26 from other data sources The result was 434 detailed occupation codes organized for bottoms-up analysis Two levels of summary information were provided for ease of use Projections to 1991 and 1996 were made from the 1986 base

This work was aimed at public data service providers, not hardware vendors The object was to estimate the wide-area mobile (WAM) opportunity for business data

Figure 10-2 Representative Yankee Group projections.

users on public networks, that is, data radio users who ranged over a wide area, as opposed to in-building or campus, and who would use public services, as opposed to private networks The focus was entirely on business jobs since, except for the deaf, few nonbusiness data radio users were anticipated

There was no examination of fixed-position opportunity: no burglar alarms, credit card terminals, telemetry, or vending machines Truck drivers were counted, but the trailer portion of a tractor/trailer rig was not evaluated Nor was there much consideration of exotica such as wireless breath analysis systems.12

The approach used in a subsequent JFD Associates update of the 1987 work was

as follows:

1 All 434 occupations were first evaluated to determine the mobile content Thus, barbers, podiatrists, and dental hygienists, for example, were rejected as mobile candidates

2 Mobile jobs were then examined for classification as campus/in-building or wide area Thus, stock clerks, waiters, and lobby attendants, for example, are mobile but not in a wide area An obvious contrast is the long-haul tractor trailer driver who is clearly not campus confined These detailed judgments were not binary but probabilistic High school principals, for example, were judged to be 95% campus, but there are exceptional cases where the principal moves from school to school There were no double counts A worker mobile in both categories, campus and wide area, was classified as WAM

3 Judgments were then made as to which jobs use two-way radio, either voice or data Fishing vessel captains, police and fire fighters, and taxi drivers are all naturals

4 The two-way radio users were then classified for public versus private Health services managers were judged very likely to choose public systems, if available But subway motormen or fire inspectors will tend to be on private systems

5 Finally, the WAM, two-way, public radio users were examined to determine who fit data as opposed to voice (one can have both) Electronic repair personnel were high data; motorcycle mechanics have low data utility The result was a 125-page report with six statistical appendices that quantified the opportunity in both 1991 and 1996

10.2.2 1996 Projections: 1987 Work

So how did the 1996 projection pan out? First, consider the dramatic impact of the filtering process on the predicted 1996 jobs A simplified summary is shown in Figure 10-3

Remember, these were worker counts The consumer, as in personal cellular phones, did not appear Nor did these counts correlate directly to devices It is quite

10.2 SOME UNPLEASANT HISTORY 149

possible for the public mobile data user to have multiple radios—even multiple data radios Railroad workers may have several wireless data terminals spotted throughout

a yard (I once used ARDIS, BSWD, and data over cellular, but the resulting payload aggravated my tendonitis) This is an airtime service view The worker may be as loaded with devices as a mule—that makes device manufacturers happy—but that same worker will only operate one device at a time

A breakout of the then-seen 1996 potential is shown in Figure 10-4 Note that the final bar groups several small-volume categories as “other.”

In general, both the 1991 and 1996 projections were rejected by carriers as totally understating the market Actually, the 1991 forecast was greatly overstated We now

“know” that less than 20% of the market potential was met from all public providers

in 1991 The 1996 forecast was much closer: ~1.1 million actual subscribers began to pay for service versus the 1.4 predicted

This seemingly close fit occurred only because the true market potential was understated There were two main problems with the 1996 forecast:

1 The national economy performed wonderfully well in the period 1990–1996 Total civilian jobs at the end of November 1996 were 127.6 million,13 not the 124.3 million projected That is less than a 3% difference, but those extra 3.3 million jobs would have permitted another 380,000 public WAM data users, more than the total of all packet users combined at the close of 1996

2 The growth of business cellular users was woefully underestimated When the work was originally done, only selected year-end 1986 cellular statistics existed: <700,000 voice subscribers served by ~1500 base stations At the close

of 1996 there were 44 million voice cellular subscribers served by more than 30,000 base stations.14 The 10-year-old estimate of less than 8 million business cellular users was undervalued by at least 50%

The gross effect of these two errors is that it would have been possible for ~2 million WAM public data users to have existed by year-end 1996, not the ~1.1 million actually reached Thus, the industry clearly underachieved

Figure 10-3 JFD Associates market opportunity: 1996.

151

Table 10-4 WAM public data candidates: 1991 >10,000

Management

Managers, food service and lodging 11,185 All other managers and administrators 34,597 General managers and top executives 86,334

Professionals

Field sales

Nontechnical sales representatives, excludes scientific and related

products/services and retail 48,675 Sales agents, selected business services 11,534 Technical sales representatives, scientific and related products/

Maintenance

Data processing equipment repairers 31,243 Electronics repairers, commercial and industrial equipment 10,995 Station installers and repairers, telephone 15,390 Office machine and cash register servicers 21,546 Television and cable TV line installers and repairers 20,632 Industrial machinery mechanics 20,761 Automotive, bus, truck mechanics 16,297

Transportation

Truck drivers, light (includes delivery and route workers) 49,639 Truck drivers, heavy or tractor trailer 332,185

Security

10.2.3 Analyzing the Data

Assuming one gets it right, the power of this job base organization can be demonstrated with the following example: After the publication of a report one buyer asked the question: “In what occupations, and how large, is the 1991 Public Data Radio Opportunity when the market is segmented so that only classifications having

a job count greater than 10,000 are evaluated?” After combining fine grain job breakdowns that, individually, would have been missed, the Public Data file was sorted at the 10,000-job level This yielded the surprisingly compressed results tabulated in Table 10-4

Much to this client’s delight, 82.6% of the Public Data opportunity was covered in spite of the “nothing less than 10,000” constraint This boded well for a Top 500 marketing strategy Other general, high-level conclusions, which had relevance to

1991 business planning, could be quickly drawn:

1 The category “maintenance” is not particularly big Given ARDIS’ success in this target segment at year-end 1991, it suggested that ARDIS would branch into other application areas to continue its growth This clearly happened: for example, AVIS, Coca Cola, G.O.D (later a cropper), UPS, as well as nonmobile tasks such as the alarm systems of National Guardian

2 The size of the category “truck drivers, heavy or tractor trailer” opportunity was obvious Qualcomm could have installed OmniTRACS at its then current rate for years and not run dry Clearly competition would appear It did: for example, AMSC and HighwayMaster

3 Nearly half the opportunity—still virtually untouched—came from three super categories: “management,” “professionals,” and “field sales.” No single profile adequately describes these applications, but the users have by far the strongest need for laptops as opposed to special, often hardened, devices Their information needs are greater, and E-mail is a routine requirement These categories retain a marked need for voice communications and a residual need for facsimile Thus, multiprotocol modems, with provision for voice cellular access, will probably be a significant asset for these segments The origins of Smartphones lie in this type of analysis

10.3 A NEW(ER) LOOK AT JOBS

In 1995 the 1987 work was totally redone This was not exclusively a need for “fresher numbers.” Job growth (or decline) moves slowly enough so that directions change at

a relatively glacial pace Simple extension to a more distant time horizon, factored for any unexpected gains in the gross total, can produce workable results

The best new reason for a totally fresh update was to attack the job estimates by region so that, ideally, an individual city breakdown can be achieved There are few oil riggers in the extractive trades in Manhattan, but quite a few in Texas Customized

10.3 A NEW(ER) LOOK AT JOBS 153