Mark Blessington is President of Mark Blessington Inc., a sales and marketing consulting firm, and the author of two recent books: Sales Forecasting: A Practical Guide (2015) and Sales Quotas: An Analytical Approach to Quota Setting (2014). He started as a day trader and later expanded into the marketing and sales arena. His article, “Sales Quota Accuracy and Forecasting,” appears in the Winter 2016 issue of Foresight. What follows is his Forecaster-in-the-Field interview, from the Spring 2016 issue.
Mark, you started your career in general management consulting and then added sales consulting to the mix. How does forecasting fit in?
My consulting career began in 1978, and for many years my client work was mostly in marketing where I did not encounter or use forecasting. That process was housed in finance, manufacturing, and strategy.
In 1985 I moved into human-resources consulting with a focus on sales effectiveness. We did a lot of work in sales compensation and plan design, but rarely concerned ourselves with quota setting, assuming it was rather inconsequential: just allocate the sales target from the annual business plan down through the sales organization. That’s simple math.
About 2000 I began to dabble in stock trading, eventually becoming a full-time technical trader, predicting short-term trend deviations using neural nets and other AI software. I believe I developed some particularly effective routines, but started to question the ethics of trading. Was I adding value to the world as a trader? Plus, my trades were being impacted by frontrunners (as described in Flash Boys by Michael Lewis).
In 2010 I eagerly returned to sales and marketing consulting. On my first “new” assignment, it hit me: traditional sales-quota setting is a rudimentary form of forecasting. What if more-sophisticated techniques were applied to produce more accurate quotas? I embraced a mission: integrate forecasting science into sales and marketing organizations.
You also seem captivated by forecasting in product hierarchies.
Hierarchies permeate sales and marketing organizations and present a unique opportunity: can we use aggregation to increase accuracy? Most investigators ask, “Which aggregation approach is best for a given hierarchy: top-down, bottom-up, or middle-out?” I call this the “one size fits all” aggregation approach.
I am exploring a subtly different angle: which aggregation method is most accurate for each part of the hierarchy? For example, is it more accurate to select the best aggregation method per territory or per product? Preliminary findings for a pharmaceutical client favor “pick and choose” over “one size fits all.” I am now creating an R module to automate my method in three dimensions (time, geography, and product). Then I’ll test it on a wider variety of sales and marketing hierarchies.
What motivated you to write your recent book Sales Forecasting – A Practical Guide?
Most forecasting books are overwhelmingly daunting, which is partly why forecasting is rare in sales and marketing departments. These people process so many transactions per hour that they have no time to sit back and decipher statistical hieroglyphics.
I teach forecasting from the learner’s point of view, starting with the novice and then progressing to intermediate forecasting. Most executives I know are beginners. Forget the formulas for all but the experts. Don’t talk to executives about complex formulas unless you want to be sidelined from the decision- making process.
The first stop on the forecasting learning curve should be testing and error measurement, without which you have no business making a forecast and you have no business assessing a quantitative forecast – regardless of your title in the organization. We must find simple and efficient ways to teach executives forecasting basics.
I do not support the current oxymoronic trend toward “blind analytics.” Forecasting software is often blindly trusted, even by the very research organizations paid to rigorously examine it. Sometimes these packages are sold as if all you need to do is buy and install the program, and then accurate forecasts start popping out. This is terribly naïve. Software should be rigorously assessed until you have independent proof of its accuracy.
Forecasting has wide utility across every business organization. Its value won’t be fully realized until the experts stop communicating with arcane symbols and start conveying basic concepts in simple and concise language. Otherwise, blind analytics will continue unabated.