How to come up with a sales forecast that works

For example, if you’re forecasting a web application, show it as a function of projected inbound traffic, clicks from paid search and visitors from organic search, then conversions, subscriptions, renewals and attrition. Granularity makes a forecast feel more solid.

As another example, if you’re forecasting a restaurant business, break it down into chairs and tables and meals served, drinks, dinners, lunches, by day of the week and, at least as a sample point, hours of the day.

Always acknowledge capacity issues.

The restaurant has chairs and tables, and the web application has bandwidth and users. Don’t ever get caught forecasting beyond capacity.

Look for patterns from other situations, similar products or business offerings, similar markets.

There are introduction patterns available for televisions, colour televisions, personal computers, faxes, cellphones, etc. Is your business going to be like any of them?

Never pretend there won’t be competition, even for the newest of the new.

If you can’t find anybody competing already where you want to go, look harder. Look for how people are already solving the need.

Before I did the early business plan templates I started publishing in 1984, there was nobody else in the market but I was competing with books, classes, and magazine articles.

I could also anticipate that if my stuff caught on, others would appear in the same market.

Distrust the data analysis.

I had great success once with an epidemiology model, predicting penetration of personal computers in Latin America would behave like the spread of disease; one user infecting others, with different economic strata having different propensity for infection.

But even that, and most of the mathematical analytical methods, are always subject to the problem of using the past to predict the future.

Be realistic

They’re good as background, to temper your thinking and educate your guessing. But they don’t stand alone very well.