The future of the New Media -- which variously such disparate markets as CD-ROM, video and computer games, PC On-Line services, and the Internet -- depends on the installed platform of personal computers that can play/run these kinds of programs. That's a simple point, but easily overlooked in the application hype.

The exact character of the consumer installed base is probably even more important right now because the whole push with all these services is into the consumer market and toward consumer applications. Who are these consumers? What are their household characteristics? And how fast is this base growing? These are the questions the PC Tracking Project from Alexander & Associates is focusing on.

Summary of Findings

Based on 10,000 total interviews conducted over the period late May through the first of August, we estimate the number of consumer households (abbreviated HH s) that have a personal computer at home for their home use at 28.3 million. This number is lower than many other estimates -- most of which are based not on consumer research but on PC shipments. Interesting ownership patterns suggest where the "stickiness" is in the rate of adoption by consumer households. These results are summarized in the graphs below.

A Methodological Note

Alexander & Associates fields the "Computer/Videogame Benchmark Study" on an annual basis in early December each year. This in-depth study emphasizes use and ownership patterns and focuses on cross platform issues -- including video games, computer applications (including on-line), cable television, and VCR use. Awareness and intent to purchase questions are part of this extensive profile of the consumer installed base. This study is based on 2,000 total interviews with a representative cross section of the US household population, using a questionnaire of over 150 interview points and a standard demographic battery.

Beginning in the Spring of 1995, A&A also began tracking PC ownership on a weekly basis because of the fast changing nature of the consumer installed base. Each weekly survey is an independently drawn random sample of 1,000 households and each sample is representative of the US household population. Independently drawn random samples are additive. So the cumulative sample on which our June/July estimate is based is 10,000 households.

The mathematics of statistical sampling imply that the standard error of estimate (which is the band of "uncertainty" around the projection) on an estimate of 29.8 percent (which is the 28.3 million households out of a population of 95.6 million) when that estimate is based on 10,000 interviews is 0.87 percent -- less than one percentage point. Relative to the quantity being estimated (the 28.3 million consumer households with a PC at home) the range is plus or minus 820 thousand households or about 2.9 percent. Other factors than sampling methodologies, however, can influence the accuracy and reliability of survey results.

Exhibit 1: Early 1995 Forecasts

The Benchmarrk study in 1994 measured the installed base of consumer households at 22.7 million, up 1.8 million from December of 1993, which was estimated at 20.8 million. These numbers, which when announced were both considerably lower than estimates in the trade and consumer press, subsequently became generally accepted as other studies reporting similar results were made public. Our estimate in early '95 of year-end '95 installations was 29.1 million, up 7 million during the year. We forecast total sales of PC s to consumers of about 9 million.

Exhibit 2: Underlying Factors

From 1993 to 1994, we estimated that about 6.3 million PC s were acquired by consumers for home use. Of this number, however, only about 1.8 million went into new PC homes; the remainder were upgrade and replacement units that did not expand the total size of the consumer installed base. Key to the split between new households and upgrade households in 1994 were the rapidly falling prices on 486 based computers that made it reasonably economical for a number of families to consider replacing legacy systems such as Commodore, IBM AT s, and 286 and early 386 processor based systems. From 1994 to 1995, based on purchase intent data in the 1994 Benchmark study, we estimated that consumer households with PC s would increase to about 29 million, an overall increase of 7 million. Additionally, we estimated that about 2 million households would upgrade or replace their systems for an estimate of total PC sales to consumers of about 9 million units.

Exhibit 3: Recent Results

Our June/July estimate can be interpreted as an average for the period or an estimate of the mid point of the period. Thus, for the end of June, we are estimating that the total number of consumer households with a PC is 28.3 million. This increase suggests that the rate of adoption among new consumer households may be higher than anticipated and that the consumer base by early December of this year (which excludes late year Holiday purchases) may be higher than 30 million.

Comments

Increased adoptions are likely to have been driven by falling 486 system prices under pressure from new Pentinum systems. We don't see any evidence that consumer enthusiasm for Windows '95 has caused anticipatory buying.