Start from the Future

In forecasting the performance of high-growth companies like, don't be constrained by current performance. Instead of starting from the present—the usual practice in DCF valuations— start by thinking about what the industry and the company could look like when they evolve from today's very high-growth, unstable condition to a sustainable, moderate growth state in the future, and then extrapolate back to current performance. The future growth state should be defined by metrics such as the penetration rate, average revenue per customer, and sustainable gross margins. Just as important as the characteristics of the industry and company in this future state is the point when it actually begins. Since Internet-related companies are new, more stable economics probably lie at least 10 to 15 years in the future.

But consider what had already achieved. Its ability to enter and dominate categories is unprecedented, both in the off- and the online worlds. In 1998, for example, it took the company just three months to banish CDNow to second place among online purveyors of music. In early 1999, it assumed the leadership among online sellers of videos in 45 days; later that year, it became the leading online consumer electronics seller in 10 days.

Let us create an optimistic scenario based on this record. Suppose that were the next Wal-Mart, another retailer that has radically changed its industry and taken a significant share of sales in its target markets. Say that by 2010, continues to be the leading online retailer and has established itself as the overall leading retailer, both off- and online, in certain markets. If the company could take a 13 percent and 12 percent share of the total U.S. book and music markets, respectively, and captured a roughly comparable share of some other markets, it would have revenues of $60 billion in 2010, when Wal-Mart's revenues will probably have exceeded $300 billion.

What operating profit margin could earn on that $60 billion? The superior market share of the company is likely to give it significant purchasing power. Remember, too, that will earn revenues and incur few associated costs from other retailers using its site. In this optimistic scenario,, with an average operating margin in the 11 percent range, would most likely do a bit better than most other retailers.

And what about capital? In the optimistic scenario, may well need less working capital and fewer fixed assets than traditional retailers do. In almost any scenario, it should need less inventory because it can consolidate its stock-in-trade in a few warehouses, and it won't need retail stores at all. We assume that's 2010 capital turnover (revenues divided by the sum of working capital and fixed assets) will be 3.4, compared with 2.5 for typical retailers.

Combining these assumptions gives us the following financial forecast for 2010: revenues, $60 billion; operating profit, $7 billion; total capital, $18 billion. We also assume that will continue to grow by about 12 percent a year for the next 15 years after 2010 and that its growth will decline to 5.5 percent a year in perpetuity after 2025, slightly exceeding the nominal growth rate of the gross domestic product.1 To estimate's current value, we discount the projected free cash flows back to the present. Their present value, including the estimated value of cash flows beyond 2025, is $37 billion.

How can we credibly forecast 10 or more years of cash flows for a company like We can't. But our goal is not to define precisely what will happen but to offer a rigorous description of what could.

1 Real GDP growth has averaged about 3 percent a year for the past 40 years, and the long-term expected inflation rate built into current interest levels is probably about 2 to 2.5 percent a year.

Weighting for Probability

Uncertainty is the hardest part of valuing high-growth technology companies, and the use of probability-weighted scenarios is a simple and straightforward way to deal with it. This approach also makes critical assumptions and interactions far more transparent than other modeling approaches, such as Monte Carlo simulation. The use of probability-weighted scenarios requires us to repeat the process of estimating a future set of financials for a full range of scenarios—some more, some less optimistic. For, we have developed four of them (Exhibit 15.1).

In Scenario A, becomes the second-largest retailer (off- or online) based in the United States. It uses much less capital than traditional retailers because it is primarily an online operation. It captures much higher operating margins because it is the online retailer of choice, even if its prices are comparable to those of other online retailers, it has more purchasing clout, and lower operating costs. This scenario implies that was worth $79 billion in the fourth quarter of 1999.

Scenario B has capturing revenues almost as large as it does in Scenario A, but its margins and need for capital fall in the range between those of that first scenario and the margins and capital requirements of a traditional retailer. This second scenario implies that had a value of $37 billion as of the fourth quarter of 1999. becomes quite a large retailer in Scenario C, though not as large as it does in Scenario B, and the company's economics are closer to those of traditional retailers. This third scenario implies a value for of $15 billion.

Exhibit 15.1—Potential Outcomes

Exhibit 15.2—Expected Value

Finally, in Scenario D, becomes a fair-sized retailer with traditional retailer economics. Online retailing mimics most other forms of the business, with many competitors in each field. Competition transfers most of the value of going online to consumers. This scenario implies that was worth only $3 billion.

We now have four scenarios in which the company's value ranges from $3 billion to $79 billion. Although the spread is quite large, each scenario is plausible.2 Now comes the critical phase of assigning probabilities and generating the resulting values for (Exhibit 15.2). We assign a low probability, 5 percent, to Scenario A. Although the company might achieve outrageously high returns, competition is likely to prevent this.'s current lead over its competitors suggests that Scenario D, too, is improbable. Scenarios B and C—both assuming attractive growth rates and reasonable returns—are therefore the most likely.

When we weight the value of each scenario, depending on its probability, and add all four of these values, we end up with $23 billion, which happened to be the company's market value on October 31, 1999. It therefore appears that's market valuation can be explained by plausible forecasts and probabilities.

Now, however, look at the sensitivity of this valuation to changing probabilities. As Exhibit 15.3 shows, relatively small variations lead to big swings in value. The share prices of companies like are extremely volatile because small changes in the market's view of the likelihood of different outcomes affect the current value of these shares quite significantly. Nothing can be done about this volatility.

2 We capture cash-flow risk through the probability-weighting of scenarios, so the cost of equity applied to each of them shouldn't include any extra premium; it can consist of the risk-free rate, an industry-average beta, and a general market-risk premium.

Exhibit 15.3—Volatility of Expected Values

Exhibit 15.3—Volatility of Expected Values

Customer Value Analysis

The last difficult aspect of valuing very high-growth companies is relating future scenarios to current performance. How can you tell a soon-to-be-successful Internet play from a soon-to-be-bankrupt one? Here, classic microeconomic and strategic skills play a critical role because building sound scenarios for a business requires knowledge of what actually drives the creation of value. For and many other Internet companies, customer-value analysis is a useful approach. Five factors drive the customer-value analysis of a retailer like

1. The average revenue per customer per year from purchases by its customers, as well as revenues from advertisements on its site and from retailers that rent space on it to sell their own products.

2. The total number of customers.

3. The contribution margin per customer (before the cost of acquiring customers).

4. The average cost of acquiring a customer.

5. The customer churn rate (that is, the proportion of customers lost each year).

Let us see how could achieve the financial performance predicted by Scenario B and compare this with the company's current performance. As Exhibit 15.4 shows, the biggest changes over the next 10 years involve the number of's customers and the average revenue for each. In Scenario B,'s customer base increases from 9 million a

Exhibit 15.4—Customer Economics, Scenario B

0 0

Post a comment