Forecasting Without Data

When it comes to preparing a forecast for a new product that has no historical or corollary data, pure analytical techniques don't help much. So what do you do when you have a new product/service idea and are trying to figure out how much of it you can sell? The answer is to develop a disciplined, methodical process that best identifies and approximates the many influencing factors and market conditions facing your product launch. Here are a few approaches to consider:

Delphi Method
This method utilizes several expert panels (functional or cross-functional; preferably a mix of insiders and outsiders) who are asked to provide a forecast. Each forecast is weighted equally and the results are shared between panels, who then continue the discussion in an effort to fine-tune their forecast with points raised by other panels. If the panels' forecasts begin to converge, a final forecast can be drawn from the average of all the forecasts. BASES Model If you have a new product to be distributed in supermarkets, drug stores, or discount stores, you might find that AC Nielsen's BASES database of 25 years of product launch history might prove to be a good starting point. The methodology looks at thousands of products with similar characteristics, consumer receptivity, and marketing plans, and forecasts sales under flexible sets of assumptions. Unfortunately, BASES is limited to just these kinds of products. It doesn't help with new paints, sporting goods, electronics, etc.

Diffusion Model
This widely used mathematical model uses rates of product adoption and usage spread from other established products to predict the rate of new product acceptance into the market. It is very applicable to both durable and non-durable goods across a wide variety of categories.

Monte Carlo Simulation

This analytical method, used in conjunction with a spreadsheet, is a form of "what-if" analysis. The name is derived from Monte Carlo, Monaco, known for its casinos and games of chance. Monte Carlo simulation is based on the probabilities of an outcome, much like the probabilities of rolling dice, and the predicted outcomes that are expected over time. Monte Carlo methods randomly select values from a defined set of possible outcomes to create scenarios of a problem. The process is repeated many thousands of times (via computer simulation) to establish probabilities of various outcomes.

Agent Simulation

Monte Carlo on steroids, Agent Simulations are highly advanced statistical models that identify and isolate the likely behavior of "agents" who are integral to the success of the product (e.g., consumer segments, competitors, distributors, etc.) based upon sets of rules found to explain past behaviors. Used for everything from air traffic modeling to war games, these sophisticated tools are very insightful where there are a large number of variable behaviors among a moderate number of agent parties. These approaches can be combined and supplemented with market research to develop a custom process that best suits your needs. Be careful to check your preferred approach with someone experienced in new product forecasting before you go too far down the road to ensure your efforts are sound before you invest too much of your time, money, or reputation.

Glossary of Forecasting Terms

Bayesian Probability:
A probability based on a person's subjective or potential judgments and experience.

Casual Forecasting:
This method involves finding factors that relate to the variable being predicted and using those factors to predict future values of the variable.

Conjoint Analysis:
A statistical technique in which respondents' utilities or valuations of attributes are inferred from the preferences they express for various combinations of these attributes.

Decision Calculus Models:

The quantitative models of a process that are calibrated by examining subjective judgments about outcomes of the process.

Demand Analysis:
A study of the reasons underlying the demand for a product with the intent of forecasting and anticipating sales.

Exponential Smoothing:
A forecasting technique to estimate future sales using a weighted average of previous demand and forecast accuracy.

Forecasting Models:
In forecasting sales, share, or other marketing objectives, a variety of methods have been used, including time series models (e.g., moving averages, exponential smoothing), econometric models (e.g., regression, input-output), and judgmental models (e.g., Delphi technique).

Markov Model:
Uses a matrix to find the probability that the user of a brand in a category will switch to another brand in the category.

Projective Technique:
A qualitative research technique used to investigate attitudes and perceptions not described by verbal answers.

Qualitative Forecasting:
Methods that rely on one or more persons to generate forecasts without using mathematical models.

Quantitative Forecasting:
Methods use historical date to forecast the future.

Regression Analysis:
A statistical technique used to derive an equation that relates a single continuous criterion variable to one or more continuous predictor variables.

Statistical Demand Analysis:

A method of developing a sales forecast that attempts to determine the relationship between sales and the important factors affecting sales, typically using regression analysis, and the use of that relationship to forecast sales for the future.

Time-Series Forecasting:
Projecting future values of a variable based on the past and current observations of the variable.

Tracking Study:
Market research to measure brand positions and the long-term effects of marketing activity.

Overall, three forecasting models have been presented: moving averages, exponential smoothing, and linear regression. Each model has advantages and disadvantages. Simply comparing MAD across all three models as seen in Table 3, it can concluded that the linear regression model tend to predict sales with more accuracy. In addition, the rules of thumb regarding regression performance all indicate the regression model is performing well.
Table 3: Sales Data Set for Rib Sales Model MAD 3 Year Moving Average 8.09 5 Year Moving Average 9.90 Exponential Smoothing Alpha = 0.1 8.52 Exponential Smoothing Alpha = 0.3 14.13 Linear Regression 6.94