I have never seen (nor heard of) a sales forecast that was accurate, except by accident.  The reason is simple: the politics of self-interest makes accurate forecasts virtually impossible.  Here’s how it works inside most firms:

  • Step #1: Top management needs a story to tell investors, and thus asks the sales manager for a forecast.
  • Step #2: The sales manager asks the sales reps what they think they’ll sell, based upon their current gut feeling.
  • Step #3: The sales reps make a wild guess at what they might actually sell, then subtract 10 percent as a fudge.
  • Step #4: The sales manager compares that forecast to the sales quota and adjusts both to match.
  • Step #5: The marketing group issues its own forecast, based upon statistics from a “market research” firm.
  • Step #6: The manufacturing group ignores all of the above and forecasts what it plans to manufacture.
  • Step #7: Top management looks at the three forecasts, throws them out, then makes up numbers that will look good.
  • Step #8: The accountants jigger the books so that, regardless of what was sold, the quarterly report resembles the CEO’s story.
  • Step #9: If, for some reason (like the laws of mathematics) Step 8 isn’t good enough, the CEO fires the sales manager.

The above routine is slightly less complicated inside smaller firms, but the essential silliness remains, because the desire for accuracy remains clouded by wishful self-interest.

Technically speaking, it IS possible to generate accurate sales forecasts! This entails building a mathematical model that predicts future buying behavior based on the following five factors:

  1. Previous years’ sales
  2. Seasonal changes in buying patterns
  3. Historical impact of marketing campaigns
  4. Overall state of the economy
  5. Fluctuations in currency exchange rates.

You then test the model against historical data to confirm that, had it been in place, it would have accurately predicted sales. 

Yeah, right.  That’s going to happen.

Companies almost never build these mathematical models 1) they don’t have the requisite skills to manage the mathematics and 2) in the case of startups especially, they don’t have enough data.

How, then, to come up with an accurate, useful forecast? The key is to remove as much of the politics as possible.  Here are three simple strategies: 

  • Decouple forecasts from quotas.  There's nothing wrong with having sales quotas, but the time to talk about those quotas (and compensation) isn't when you're trying to come up with a meaningful projections.  This takes some mental discipline on the part of the sales manager, who must resist the temptation to conflate the two activities.
  • Reward accuracy rather than penalize it.  While there's no question that the sales team should be able deliver the sales that are needed to make the company successful, the forecasting process needs to reflect reality, not wishful thinking.  Shooting the messenger is dumb.  What’s needed is “just the facts” attitude with rewards for getting the facts right.
  • Remove the amateurs from the process.  Part of the problem with the standard scenario is that everybody gets involved, bringing along their own, weird organizational baggage. Marketing, manufacturing, accounting and top management should react to the sales forecast, rather than drive it.

READERS: Have any of you ever seen an accurate forecast, other than by accident?  If so, please leave a comment, because you’re a real rarity in the business world.