How are Internet Forecasts Made?

Brill's Content finds out.

An important aspect of the Internet economy is its own growth estimates. The reports of private research organisations such as Jupiter Communications or Forrester Research are especially influential. Their growth forecasts of the future size of the Internet population or the ecommerce market are often quoted in business plans, IPO documentation or in the press.

Forecasts are very important for the Internet industry because so few firms have actually made profits. One example for the significance of estimates is in earnings per share (EPS) calculations, used frequently to compare stock valuations. An EPS based on expected profits in 1999 or 2000 is used for firms such as Excite, Lycos or Amazon.com in the USA or Intershop and Brokat in Germany.

But how do research organisations arrive at their predictions? Jennifer Greenstein asked this question in the November issue of Brill's Content (the article is not available on-line), an American magazine recently started as a monitoring instance for the quality of old and new media content. Greenstein discovered that some of the currently available forecasts for growth of the Internet population and ecommerce have been calculated using data obtained in 1996. And some of this data was actually derived from interviews with the management of the Internet companies themselves. Forrester Research, for example, interviewed the executives of 52 ecommerce firms, asking them for their future market size estimates. The forecasts for the on-line markets are, therefore, partially the necessarily optimistic expectations of the market participants themselves.

Figures obtained through interviews with company executives are supplemented with data generated through consumer surveys. Greenstein reports that Forrester Research surveyed only 100 to 300 people for its 1996 American on-line shopping estimates. In December 1997 Forrester initiated a more extensive survey of 170,000 consumers; it will use this data in its new ecommerce study to be released this month. The work of Jupiter Communications appears in a better light than that of its competitor. Already in 1996 Jupiter supplemented 237 company interviews with data collected from 3,000 consumers. For its newest ecommerce reports Jupiter surveyed 65,000 people four times in November 1997.

Greenstein's article in Brill's Content does not refer to the European estimates generated by American research companies. One of the best known multi-country estimates of European ecommerce is a report by Jupiter Communications released in July 1998. The study compared the on-line markets for travel, books, music and software in the UK, Germany, France and Scandinavia in the years 1997 to 2002. The report predicts that Germany will have a significant lead in all four categories. Jupiter arrived at these estimates with the obligatory executive interviews, an analysis of traditional European shopping patterns as well as a telephone-based consumer survey of 150 people in each region.