All Probabilities are the Same

Quick Summary: Sales probabilities are always 50%, an order is received or not.

Abstract:

One only has to review quarterly results of public companies to see how difficult it is to accurately forecast sales. These companies have incredible resources at their disposal and most have detailed historical data to take into account in developing their predictions.  Most companies assign probabilities to each sale and then determine a weighted average in determining their forecasts.  The wide variance from period to period and company to company suggests that this technique is less than effective.

Most sales organizations forecast sales based on the probability of receiving an order.  Some systems are highly sophisticated using algorithms that take into account a number of factors, perhaps even the phase of the moon.  Meanwhile, others are nothing more than the gut feeling of each sales person.  Independent of the technique used, all probabilities are the same.  The probability of receiving an order is 50%; either you receive it or you don’t.  To test this hypothesis, have you ever received 80% of a purchase order?  Instead of it being written on 8.5 by 11 inch paper, was it written on 6.8 inch tall paper (80% of 8.5 inches)?

On the other hand, if the number of opportunities is large enough, consistent probability assignments are made, orders are roughly the same size, and time is not a factor, perhaps determining a weighted average of the dollar amount times the probability can yield a reasonable guess that could be used as a forecast.  Unfortunately, to be useful, time cannot be ignored, different people with different views on the weighting factors will occur, and order sizes can vary significantly.  So, the weighted average technique is most likely problematic.

An exception can be forecasting a very large quantity, low cost product or service with mass market appeal.  A movie theater popcorn machine is a good analogy to use to describe this model.  Although there is a certain probability that each kernel will pop, it is extremely hard to predict which one will pop at any given time due to the variance of their position next to the heating coils.  However, one can fairly accurately predict that a certain number of kerns will pop during some specific time frame.  As long as there are enough kernels, heat, and there was no requirement for any one kernel to pop at any one time, then the average probability model can provide an acceptable guess.

Unfortunately, only a few of us are in the popcorn business.  Instead, individual focus is given on each sales opportunity and each will be impacted by any number of variables that can change their value and importance at any time.  Obviously, sales forecasting is a very hard problem to solve.  Quarterly results from public companies and private companies alike show the difficulty of accurately predicting sales.  Very few companies regularly meet their forecasts.  Although many will deny it, many of the companies that do consistently meet their forecasts seem to be able to start the next quarter with a substantial backlog.  Perhaps they were lucky or perhaps they were “sandbagging” and delayed reporting certain sales.  With the financial market’s fixation on short term results and to validate their own predictive models, who can blame companies for using any legal means at their disposal to meet their forecasts.

For many companies that serve markets that are characterized with short product life cycles or products that fall into the category of “nice to have”, such as new cell phones, rather than “must have”, such as groceries, forecasting can be even more problematic due to external influences that are well beyond the control of the company.  Relying on simulations, complex algorithms or other confidence building techniques to create three digit accuracy probability models is like measuring with a micrometer, marking with caulk, and cutting with an ax.

Probabilities set by sales reps or those closest to the prospect can be useful as one tool in developing a forecast.  The best way to use them is to focus on the very low and very high probabilities only.  For example, do not consider any potential sales within the period of interest with a probability lower than 50%.  Similarly, provide extra scrutiny to any potential sales opportunity with a probability of great than 85%.  For those high probability opportunities, focus on what could go wrong to derail or delay the sale.  Once identified, the focus should shift to working on preventing the negative factors to minimize their occurrence.  The percentages associated with this technique may need to be adjusted based on who is setting the probabilities for each potential sales event.  Naturally, some individuals are more optimistic while others are more pessimistic.  Also, just as companies may “adjust” their performance, sales rep may do the same by adjusting their probabilities.

Remember, the probability of the probabilities being correct is 50%.  Either the probabilities are correct or they are not.

Article Number : 5.010202

A Handy Reference Guide for Executives and Managers at All Levels.

9 Volumes 36 Chapters ~500 Articles