Sales organizations drive Industrial product sales
March 29, 2018
Much of the focus in the literature has been on demand forecasting for consumer product companies. There has been less written about demand forecasting for industrial product companies that market technically savvy products to businesses, largely via a direct sales force. Thus, a lot has been written about using historical, statistically based forecasts that incorporate promotional and new product impacts on demand, and that are adjusted by marketing intelligence. However, sales organizations are the major “demand-shapers” in the industrial-product industries, and what sales does needs to be more seriously considered when developing a demand forecast.
How industrial manufacturing companies shape demand
Much of the literature discusses the demand forecasting process for products sold to consumers. Thus, it focuses more on forecasting the demand impacts from the promotional and new product activities of marketing organizations, with much less on sales organization activities. In contrast, sales organizations of industrial product companies play the most important role in creating and shaping demand, because their customers are largely businesses to which they sell directly. As I think about the various topics I’ve explored in this ongoing JBF column, I have to admit that the lion’s share of them have been consumer-products oriented. Yet, my first major project at a product company, Data General (DG), led to building a model to forecast computer sales based on sales force size and other characteristics.
Sales Forecasting at a Computer Manufacturer
I joined DG, a Fortune 500 industrial-products manufacturer (of minicomputers), in the early 1980s as a management science analyst in an internal consulting group. The group did in-house analytical projects for various departments at the company. As a marketing person (at the time), my boss and I looked to do my first project with the commercial-side of the company, in one of the sales or marketing organizations. We found a sales support director who was interested in analyzing sales force productivity.
We started the project by collecting data on individual sales reps and how much they sold each fiscal quarter, as well as other information about them. After much data crunching, we uncovered a “correlation” between the amount of time a sales rep was with the company and his/her sales performance—a “learning curve” per se. For example, we had found that newly hired sales reps sold very little in their first six months at the company. It turned out there was a steep “learning curve” in selling our computers. As a rapidly growing high-tech company with significant sales rep turnover in a “hot” sales market, many of our reps were fairly new to the company.
The VP of Sales bought into the learning-curve concept and agreed that the “correlation” between employment time and a rep’s selling performance over time was indeed “causal. ” Based on his agreement, our project team used the empirical learning curve as the basis of a computerized sales simulation model that we developed. The VP used it to make decisions about hiring and the size of the sales force that was needed to achieve sales targets. The learning-curve relationship was updated each quarter and then used as the basis for forecasting computer sales. The future sales rep hiring plans in conjunction with the employment time of current sales reps were the major inputs to the model. The model “aged” sales reps with respect to their time at the company over the forecasting horizon. This created a profile of the numbers of reps in each age category over time, which was then used with the learning curve to forecast quarterly sales. The model was successful, and it became the VP’s decision-support tool for evaluating hiring plans.
Later on, the model had its greatest success. It was in the year that it was used to advocate for hiring more sales reps than the company’s executive team thought would be necessary to meet sales targets. The model’s results were brought into the company’s board meeting to help them finalize hiring decisions. It is not often that quantitative models make it into a boardroom. As a quantitative modeler, I considered it to be a great success! Later in my career, I’ve come to realize that using a sales-based forecast to project industrial-product sales is critical.
Industrial Product Selling
Most industrial products are highly engineered and therefore are sold on their merits in terms of their technical specifications and features. Often these are specified in a Request-for-Proposal (RFP) sent out by the customer. At DG, we sold our computers through two major channels: 1) direct sales to “user” companies that would develop their own customized software applications, and 2) direct sales to Value-Added Resellers (VARs) that would bundle our computers into turn-key systems—that were comprised of VAR-developed software as well as other computer hardware components. Direct sales using our own sales force was the only way we marketed our products—because it was a technical sell. Our products were not commodities, as many consumer-products are, and therefore needed a product-knowledgeable sales team to sell them. Generally little advertising, distribution, or promotional activities are required to market industrial products.
At many industrial product companies, the engineering-types rule the roost. Executive teams believe that if one designs and builds a great product, customers will flock to your doors to buy them. They feel that there is little need for sophisticated marketing and sales programs. Sales reps are viewed as order-takers not order-makers. One year, I heard that a few sales reps were going to be compensated more than the CEO, and that the executive team was lamenting the fact. The team was thinking about changing sales compensation schemes in order to prevent this from happening again. (Frankly, as a mere analyst, I was hoping that many more of our sales reps were highly compensated, so they could afford to drive big, fancy, and expensive cars. Sales reps brought in the revenue that helped pay my salary!).
While these executive beliefs may or may not be true, these order-taking sales reps do drive when and in what quantities customers purchase industrial products. So, while they might not create demand, they certainly help to shape it. The sales rep has to configure the requisite equipment and place a sales order that is accurately configured to meet the customer’s RFP-based specifications.
Thus, sales activities need to be seriously considered when forecasting industrial-product customer demand, of course, in the context of other considerations. My research over time has uncovered a few anecdotes that reinforce this point.
- I once met a consultant who had an extensive consulting engagement with a large high-tech company’s sales organization. He had developed a standalone sales forecasting system for it. And like the one we developed at DG, it evolved to become an important decision-support tool for making sales management decisions. It turned out to be so successful in forecasting, that the company’s supply chain group decided to build on it to support its operational forecasting.
- I was once briefed by a supply chain manager from Alcatel-Lucent, a telecommunications equipment manufacturer which marketed business systems. The manager was responsible for demand forecasting and inventory management in the company. Organizationally, he originally had his forecasters and inventory analysts co-located in the corporate office. After seeing that his forecasters were underperforming, he distributed them so as to be co-located in key regional sales offices. He found out that when they worked among the sales reps they got the visibility (into future sale opportunities) they needed to produce more accurate forecasts. As he said to me (paraphrasing): “I increased my forecasters’ salaries, had them live day-to-day with the sales force, and they then shipped me more accurate forecasts each month. ”
3. A VP of supply chain at IBM had gone through a successful major overhaul of its supply chain operations, including the order management function. As he was initially evaluating the latter function, he found sales reps complaining that they were spending inordinate amounts of time entering orders for integrated computer systems. It was taking too long to put orders into the system that would definitely pass the “system integration” phase when the order was filled. To fix this problem—and enable sales reps to spend more time selling to customers—the VP implemented an improved order management system and support team. It was such an important element of the major overhaul, that sales rep productivity was one of the major goaled-objectives in his annual performance review by his superior.
A sales forecast needs to be developed - not assigned
In my JBF column in the Spring 2000 issue, “Forecast Reconciliation: Whom Do You Trust? ” I discussed the issue of having multiple forecasts (for example, forecasts from sales, marketing, as well as a historical statistical baseline forecast) that need to be reconciled. The advice I gave at that time was that Sales’ forecasts are better at ascertaining geographical trends, while Marketing’s are better at detecting brand and product line trends. Meanwhile, a statistical baseline forecast is a great “sanity check” for total demand being forecast at a national level across all products.
However, with regard to industrial products, I would add that a sales forecast might potentially be more useful than a baseline forecast at the national level. However, this would be true only if it is really a sales forecast, and not merely a sales plan. Typically, when a sales organization produces a so-called sales forecast, it is really what they plan to sell, not forecast to sell. A typical approach to developing a sales plan is to start with an annual national sales target (in dollars), and then portion it out to sales regions, which then assign portions to individual sales reps in the region. The resulting numbers are then used to set each rep’s annual sales target, upon which his/her commission will be determined. Thus, the real intent of the so-called sales forecast is to align a national target by assigning it to sales reps in the regions—not to forecast sales.
The problem with these types of sales forecasts is that they are developed based on gut feelings, rather than on established scientific forecasting methods and principles. Like the DG model, sales forecast models ought to incorporate the quantitative impacts to sales of various factors about sales force activities, as well as the status of the future sales pipeline of prospective customers. Generally, historical sales data (drawn from transactional order management systems) is critical in consumer products forecasting. Meanwhile, in industrial-product forecasting, future customer information (drawn from Customer Relationship Management—CRM—systems, such as Salesforce.com) is critical.
Generally, many business forecasters are skeptical of forecasts produced by a sales organization, mainly because they are not professionally developed. Forecasters who feel this way, especially in the industrial-product industries, should not just disregard the sales forecasts. Instead, they should roll up their sleeves and work more closely with sales managers to develop credible sales forecasting models. This will produce more accurate demand forecasts, as well as provide sales management with a decision-support tool to help them be more productive.
Dr. Lapide is a lecturer at the University of Massachusetts, Boston, and an MIT Research Affiliate. He has extensive experience in industry, consulting, business research, and academia as well as a broad range of forecasting, planning, and supply chain experiences. He was an industry forecaster for many years, led supply chain consulting projects for clients across a variety of industries, and has researched supply chain and forecasting software as an analyst. He is the recipient of IBF’s inaugural Lifetime Achievement in Business Forecasting and Planning Award. He welcomes comments on his columns at email@example.com.