266  Forecasting Performance revenue drivers. Taking a fine-grained look at a company’s sources of growth will make clear what drives the company’s valuation. In new-product markets, the top-down approach is especially helpful but often requires more work than for established markets. For instance, consider the recent launch of June Life, a maker of web-enabled ovens. The company’s smart oven is marketed as many appliances in one, including a toaster, dehy- drator, and slow cooker. The accompanying smartphone app allows the user to control the oven remotely, check on remaining time, and even view the product cooking. Given the lack of history for the company’s products, how do you estimate the potential size and speed of penetration of this new product? You could start by sizing the more traditional products of Black & Decker and Cuisin- art. Analyze whether the new smart ovens, given their greater functionality, will be adopted by even more users than traditional ovens—or perhaps by fewer, because of their higher price. Next, forecast how quickly web-enabled products might penetrate the market. To do this, look at the speed of migra- tion for other electronics that have gone through a similar transition, such as the voice-only cell phone to the smartphone. Determine the characteristics that drive conversion in other markets; this helps you place your forecast in context. Next, assess the price point and resulting operating margin for the company’s products. How many companies are developing the product, and EXHIBIT 13.3  Costco: Sample Revenue Forecast1 $ million Historical Forecast 2017 2018 2019 2020 2021 2022 2023 2024 2025 U.S. revenues Revenue per square foot, $ 1,007 1,054 1,100 1,144 1,172 1,202 1,226 1,250 1,275 × Square footage per store, thousands 147 147 147 147 148 148 148 148 148 × Number of stores 514 527 543 558 566 574 582 590 598 = U.S. revenues 76,087 81,652 87,803 93,838 98,176 102,112 105,603 109,150 112,843 International stores Revenue per square foot, $ 904 958 968 997 1,027 1,058 1,089 1,122 1,156 × Square footage per store, thousands 142 142 144 144 144 144 144 144 144 × Number of stores 225 233 236 244 252 260 268 276 284 = International revenues 28,883 31,696 32,897 35,031 37,268 39,612 42,027 44,593 47,276 Membership fees Average fee per member 32 33 34 35 35 36 37 38 38 × Number of members, millions 90 94 99 102 104 108 110 114 116 = Membership fees 2,853 3,140 3,349 3,539 3,682 3,899 4,048 4,286 4,443 Ancillary businesses2 21,400 24,900 28,600 30,900 33,400 36,100 39,000 42,100 45,500 Total revenues 129,223 141,389 152,649 163,308 172,525 181,723 190,677 200,129 210,061 1 For better comparability across companies, data are presented on a calendar basis. Costco’s fiscal year-end is August 31. 2 Ancillary businesses include gas stations, pharmacies, optical dispensing centers, food courts, and hearing-aid centers. Source: Trefis, “Costco,” November 2019. Mechanics of Forecasting  267 how competitive will the market be? As you can see, there are more questions than answers. The key is structuring the analysis and applying historical evi- dence from comparable markets to help bound forecasts whenever possible. Whereas a top-down approach starts with the aggregate market and pre- dicts penetration rates, price changes, and market shares, a bottom-up ap- proach relies on projections of customer demand. In some industries, a company’s customers will have projected their own revenue forecasts and can give their suppliers a rough estimate of their own purchase projections. By aggregating across customers, you can determine short-term forecasts of revenues from the current customer base. Next, estimate the rate of customer turnover. If customer turnover is significant, you must eliminate a portion of estimated revenues. As a final step, project how many new customers the company will attract and how much revenue those customers will contrib- ute. The resulting bottom-up forecast combines new customers with revenues from existing customers.5 Regardless of the method, forecasting revenues over long time periods is imprecise. Customer preferences, technologies, and corporate strategies change. These often-unpredictable changes can profoundly influence the win- ners and losers in the marketplace. Therefore, you must constantly reevaluate whether the current forecast is consistent with industry dynamics, competi- tive positioning, and the historical evidence on corporate growth. If you lack confidence in your revenue forecast, use multiple scenarios to model uncer- tainty. Doing this not only will bound the forecast, but also will help company management make better decisions. A discussion of scenario analysis can be found in Chapter 16. Step 3: Forecast the Income Statement With a revenue forecast in place, forecast individual line items related to the income statement. To forecast a line item, use a three-step process: 1. Decide what economic relationships drive the line item. For most line items, fore- casts will be tied directly to revenues. Some line items will be economically tied to a specific asset or liability. For instance, interest income is usually generated by cash and marketable securities; if this is the case, forecasts of interest income should be tied to cash and marketable securities. 2. Estimate the forecast ratio. For each line item on the income statement, compute historical values for each ratio, followed by estimates for each of the forecast periods. To get the model working properly, initially set the forecast ratio equal to the previous year’s value. Your forecasts are 5 For more on company valuation using customer acquisition and retention statistics, see Daniel Mc- Carthy, Peter Fader, and Bruce Hardie, “Valuing Subscription-Based Businesses Using Publicly Dis- closed Customer Data,” Journal of Marketing 81, no. 1 (2018): 17–35.