Many business leaders are wondering how to forecast sales through the rest of the year. They also wonder how to forecast next year’s sales, which is the usual starting point for annual planning. It’s a tough job, but without a sales forecast the management team doesn’t know whether to conserve cash aggressively or continue current operations. It’s not an easy problem to solve, but some insights are possible.

As of this writing (early July 2020), the economy is in recession and the path of the Covid-19 pandemic is highly uncertain. Let’s start by bracketing the range of possibilities. The upside is easy: a quick return to the old forecast, the one we started using in January 2020. That would require a vaccine or treatment being developed quickly, so it’s certainly the optimistic end of possibility.

The downside risk can be estimated by the worst-case historically. Major sectors of the economy are tallied up by the U.S. Bureau of Economic Analysis’ National Income and Product Accounts. Here’s a table showing the worst years since World War II for major parts of the U.S. economy:

Spending % Change

Consumers

Durable goods -7.9

Nondurable goods -4.3

Residential -25.3

Nonresidential -20.2

Business Capital Spending

Equipment -20.7

Intellectual property -4.8

Government (ex transfers)

Federal, defense -67.0

Federal, nondefense -16.0

State & local -0.4

More detail is available, but this table illustrates some important generalizations. Durable goods have a greater downside than nondurables and services. We can postpone buying a new car or sofa easier than we can postpone buying groceries or paying utilities. Business investment spending can drop pretty steeply. This format does not show business spending on operations—those expenses are incorporated into the final products that they sell. We know from other sources that business purchases of new machinery can go down very steeply, more so than purchases of grease to keep the old machinery running.

Could the 2020 or 2021 economy be even worse than the table shows? Records are made to be broken, the athletes say, and the economy could decline more than in the past. Still, worst-on-record is a reasonably pessimistic scenario for planning. Running such numbers through a cash model will likely be a grim experience. If the company can develop tactics to survive worst-on-record, however, the leadership can then focus their attention on growth strategies.

Other scenarios that the finance team might model are two-thirds of worst and one-third of worst. The revenue paths of the different scenarios will drive different levels of cost-cutting.

Most companies will need a deeper dive into data. For businesses selling directly to end-users, the table above is a good starting point. Each of these categories can be broken down in more detail so that consumer durables can be split among motor vehicles, furniture, and other household goods, recreational goods, and other. Then each category is further subdivided, so motor vehicles is broken down to new vehicles, used vehicles, and parts and accessories.

Businesses that sell to other businesses for resale or further processing have greater data challenges. Lawyers sell some services directly to consumers (divorce and estate planning.), but most services are provided to businesses. Similarly, hydraulic assemblies may be sold for use in cars, construction equipment, agricultural implements or industrial machinery.

For raw materials or intermediate goods and services, companies must understand where their products are going to identify the end-user. Sometimes sales functions are divided by market, so this information is readily available. Often times trade associations gather this sort of information. Then the company’s worst-case scenario combines the worst cases for several different sectors.

The challenge is greater when sales are mostly through wholesalers or other intermediaries. Understanding the company’s business requires working with the wholesalers to understand the end-users—for it’s the end-users who trigger the company’s sales cycle.

Statistics that reliably foretell the future are much searched for but less often found, as I argued in Leading Economic Indicators Are Misleading. A good approach has companies searching for the drivers of final demand for their goods or services. For consumer goods, business analysts can monitor employment, incomes, and consumer attitudes. Sharp analysts will understand just which consumers are the biggest buyers. Some products are aimed at low-income families (think dollar stores), while others target the middle class and others the luxury market. Most of the data on income by class are released with long time lags, but we can infer some patterns with employment data. The general principle is to abandon the idea of finding ideal data, and use what data we have.

For example, a company selling to lower-income consumers might use the monthly employment report for the U.S. to see how people with just a high school education are doing finding jobs. A business selling luxury goods might monitor the stock market.

In the end, a sales forecast will be judgmental. It may come out of a computer model without further adjustment, but some human-created the model and decided to use it rather than a finger in the wind.

With the coronavirus pandemic, an understanding of the cases and possible paths for the virus will help forecasting. Sales forecasters may warn that they are not epidemiologists, but good forecasters grab whatever data are relevant, learning the appropriate theory along the way.

Although sales forecasting is possible, highly accurate forecasting is not. So the sales forecast can be used as a rough idea, with plenty of humility add to the construction of business plans. Flexibility in decision-making will benefit those companies that develop resilient plans. Leaders are often committed to their prior plans, because of emotional attachment to their creations as well as the prestige that comes from plans respected by others. Losing those emotional commitments will allow sales forecasts to be rough guides rather than set-in-concrete constraints on future changes.