From concept to consumer

Editor's note: Jerry W. Thomas is CEO of Decision Analyst. He can be reached at jthomas@decisionanalyst.com.

Most of the books and articles on new product sales forecasting published over the last 50 to 60 years have focused on consumer packaged goods (CPG). These are the packaged foods, beverages, health and beauty aids, soaps and cleaners, over-the-counter drugs, etc., sold in supermarkets, grocery stores, drugstores and mass merchandisers such as Walmart, Target and Amazon.

While the lessons from forecasting sales of new CPG products might give us some clues to forecasting the sales of new durable goods, for the most part the CPG new product forecasting models and methods are ineffective when it comes to forecasting new durable goods sales.

Consumer packaged goods tend to be used up or consumed quickly and purchased frequently (typically multiple times per year), whereas durable goods tend to last a very long time (a minimum of three years, according to the Bureau of Economic Research). Washing machines, refrigerators, lawn mowers, chainsaws, automobiles, computers and power tools, etc., are examples of common durable goods. As these examples suggest, durable goods tend to be:

  • Physical, tangible products.
  • Long-lasting (three or more years).
  • Non-consumable (that is, they are not used up or consumed in the short run; these products can be used over and over again).
  • Tools or machines. Most durable goods can be thought of as “tools” or “machines” to help the human race do work, accomplish tasks, store things or move things.
  • Expensive, compared to CPG products. Most CPG products cost a few dollars, whereas durable goods often cost hundreds or even thousands of dollars.
  • Constantly evolving. For example, if a manufacturer launches a new electric lawn mower, that lawn mower will likely undergo annual updates and improvements, so the product gets better and better year after year.

Given the characteristics of durable goods, how can a manufacturer go about forecasting sales of these products with some hope of reasonable success? Let’s begin with a confession and a little humility. Accurate durable-goods forecasting is extremely difficult and the longer the time horizon, the less accurate the forecast will be. No mathematical model, no artificial intelligence or machine learning model, no soothsayer or industry expert can sit in an ivory tower and accurately predict sales of a new durable-goods product. Sales forecasts must be based on evidence and facts from the likely purchasers of that new product.

Typically, a new durable good is very expensive to develop, manufacture and take to market. The costs include: millions of dollars in product design, development and engineering; millions of dollars in tools, machines and production lines to manufacture the new product; millions of dollars in sales, marketing and distribution costs to take the new product to market.

First, with so much money at risk, a reasonably accurate sales forecast will help the manufacturer make an informed and educated decision about whether the new product makes any sense. Second, the disciplines and thinking involved in developing a good forecast will identify ways to reduce risks and increase chances of success. Third, the analytics and research involved in the forecasting will help develop the long-term evolutionary plan for keeping the new product relevant and competitive in the future.

Within the broad and complicated world called durable goods, there are tens of thousands of major companies, each with different strengths and weaknesses. There is no one best-practices paradigm that makes sense for all of these disparate companies. Accordingly, the following suggestions are a rough guide and the guidelines must be tailored to each company and its unique characteristics and situation.

Struggle to identify

Some companies are very creative and have lots of new product ideas to evaluate, while others struggle to identify viable opportunities. Regardless, the process of developing new durable-goods ideas can generally benefit from:

  • In-depth qualitative research. Focus groups and in-depth interviews among target-market groups can be invaluable. If we can develop a detailed, in-depth understanding of the target audience, their motivations and wishes, their fears and failures, their frustrations and struggles, we may discover new product opportunities. We can’t ask a target audience what new products they need (they don’t know), but if we truly understand their emotions, behaviors, and frustrations related to a product category, we may see or infer new product possibilities.
  • Ideation and brainstorming. Once we know something about the physical and psychological needs of our target audiences, we can use ideation and brainstorming sessions to create new product ideas to solve the problems or frustrations and seize the opportunities identified. These same ideation methods can be used with in-house teams from new product development, marketing, sales and engineering, to help guide the new product development process and create new products that address target-market opportunities.

Evaluating early-stage new product ideas

Once you have some new product ideas to evaluate, marketing research techniques can be applied. If you have 20, 30 or more ideas, then we would recommend new product screening studies to identify the ideas with the greatest market potential. Typically, such screening studies can be conducted at modest costs.

Once we have identified the better new product opportunities, we can do volumetric concept tests to give us an approximation of the sales volume potential represented by each of the top-rated new product ideas. The concept test with volumetric questions gives us multiple ways to assess market potential and determine which of the new product concepts warrant further development. Since each new product concept is evaluated monadically (that is, evaluated by an independent sample of target-market consumers), the cost per new product concept is somewhat expensive – but much cheaper than a failed new product.

Evaluating late-stage new product ideas

This is where it gets difficult. We must have some type of model or prototype of the new product so that we can get accurate feedback from target-market users. If it’s too expensive to build a prototype, then a demonstration video might be a solution so that all the details of the new product can be fully communicated to the target market. We must be able to fully communicate what the new product is and how it works, so that accurate measurement of market potential is possible. The more realistic the presentation of the new product prototype, the more accurate the marketing research will be.

In most instances, some type of product clinic is the next step. Target-market users are recruited to come to a central-location facility where a model or video of the new product concept can be shown and demonstrated to potential purchasers. This process might begin by only showing the new product model or video demonstration to get initial consumer feedback. Then, the model or video presentation itself would be tweaked and refined.

Then, if there is a well-defined product category, major competitive products would be introduced into the product-clinic mix, so that potential purchasers are evaluating the new product in comparison to competitive brands. Potential purchasers would be asked what they like and do not like about the new product, relative to competitive brands. They would be asked how likely they would be to purchase the new product, if it were available, and they would be asked about current product features, plus potential new features that could be added in the future, styling, pricing expectations, etc.

If budgets permit, the product clinic is a wonderful opportunity to learn much more. Choice modeling is a powerful technique that can be used to optimize design, functionality, features and pricing of the new product. In choice modeling, participants are asked to select which of the products in the clinic they would be most likely to buy, given a set of features, prices and functionality for each brand (called a scenario).

Once a choice is made, then the variables change for each brand (creating a new scenario), and participants are asked to choose which one they would buy in this new scenario. This process is repeated following an experimental design until six to 10 scenarios are completed. The experimental design allows decision scientists to infer the importance of different features, functions and prices for each brand, including the new product.

Choice modeling provides a reliable way to predict how well the new product will sell once it’s introduced into the market, given a set of features, functions, prices and competitive brands. The equations derived from the choice modeling results are then used to create a decision simulator, so that analysts can play “what-if” games on the simulator. What would the new product’s market share be at the end of the first year, given a particular set of features, functions, price, distribution level, competitive assumptions and estimated brand awareness? Choice modeling provides a solid basis for forecasting a new product’s first-year sales volume. Second-year forecasts can also be estimated by the simulator but the accuracy goes down as the time horizon stretches into the future.

Qualitative investigations 

Given the challenges of forecasting sales of new durable goods, it’s often a good idea to include qualitative investigations as a part of the product clinic. The depth of understanding that comes from good qualitative research can help refine the sales forecasts produced by the simulator.

Another really important factor for durable goods is the nature of the distribution system. For example, power tools may be distributed through large retailers like Home Depot, Lowe’s and Amazon but a small manufacturer of lawn mowers may go to market via a hodgepodge of small dealers scattered across several states. The simulator forecast must be adjusted for these types of distribution variances. The role of advertising varies greatly from product to product. Apple Computer may spend $100 million on advertising to introduce a new computer, while the small lawn mower manufacturer may spend $100,000. The sales forecast must take into account the length and strength of media advertising.

New product sales forecasting is expensive but it can dramatically reduce the risks of failure and increase the likelihood of long-term success.