Basic Information of Forecasting the Adoption of a New Product
Author: Elie OfekPublisher: HBR
Case Number: 505062
Publication Date: Feb 15, 2005
Revision Date: May 20, 2008
Course Category: Marketing
Case Summary of Forecasting the Adoption of a New Product
The Bass Model or Product Diffusion Model- Created by Frank Bass in 1969 as an analytical framework for modeling the first-purchase growth of a new product
- The model makes assumptions about how information is passed between individuals in a social system, and how this affects their timing of adoption
- Assumes that a consumer can adopt a new product only once
- Innovators – individuals who adopt a product independently of others
- Imitators – individuals who adopt a product only after observing that others have done so first; they respond to influences
- Variables in the Bass Model:
N(t) is the total cumulative # of consumers that have already adopted the new product through period t;
N(t-1) is the cumulative # of adopters for the new product through the previous time period (i.e. t – 1)
S(t) is the # of new adopters for the product during the time period t, and can be expressed as N(t) – N(t - 1)
m is the total market size; provides the scale of the demand forecast; gives a total consumer base or terminal value of total adopters that will not be exceeded
p is the coefficient of innovation; represents the rate or probability that an innovator will adopt at time t
q is the coefficient of imitation; accounts for “word-of-mouth” effects that result from interpersonal communication between adopters & non-adopters
- The Bass model asserts that the likelihood of an initial purchase being made at time t, given that a purchase has not been made before, is a function of the # of previous adopters, such that:
p + (q/m)N(t - 1) = likelihood of purchase by a new adopter in time period t
m – N(t - 1) is the # of consumers that haven’t previously adopted by the start of time period t; this is the pool from which new adoptions in the current period can occur
- The Bass model in its simplest form gives us the number of new adopters in time period t by multiplying the rate of new adopters by the # of consumers that have yet to adopt:
S(t) = [p + (q/m)N(t – 1)][m – N(t – 1)] -> the Bass model
- Ex. A1 in the reading shows possible adoption curves for new products, where a product with a high coefficient of innovation will have a sudden increase in the adoption curve before leveling off. Conversely, a product with a low coefficient of innovation and a high coefficient of imitation will have a gradually sloping adoption curve.
- Market research surveys that assess consumer interest or likelihood of purchase is one way to gauge total market size
- p & q may be determined through the analysis of p’s & q’s for previously launched analogous products
- Several analogous products may be considered, in which case a weighted average of their p & q values is appropriate
XM Satellite Radio and its launch:
- XM used a national telephone survey to gauge interest and determine total market
- There are other factors that influence total market though, such as the % of the market that’s willing to purchase new radios and pay a monthly fee, so it is important to do a sensitivity analysis
- to estimate p & q, XM assumed AM/FM automobile radios, portable CD players, & mobile phones & satellite TV to be analogous products and used weights to determine p & q
- XM forecasted a gradually sloping demand curve after launch due to a model dominated by imitators rather than innovators.
- Of course, market penetration may be affected by price decreases or advertising spend, in which case the ff. generalized Bass model (GBM) holds:
S(t) = [p + (q/m)N(t – 1)][m – N(t – 1)]Z(t)
Where Z(t) = 1 + α[P(t) – P(t – 1)]/P(t – 1);
α is a coefficient that indicates the percentage increase in the speed of diffusion that results in a 1% decrease in price
P(t) is price in period t
And the ff. assumptions hold:
1. a price decrease affects the # of adoptions in that period only
2. the actions of the firm through it’s marketing mix affects imitators and innovators the same way
OTHER MODELS DISCUSSED BRIEFLY IN THE READING:
The Discrete Choice Model
- Use this if 2 or more versions of the new product are already in marketplace, or if prototypes exist
- Develop a set of key attributes by which to compare competing versions of the product, perform a conjoint analysis, & use this data to estimate market share for each product
Data Driven Demand Forecasts
- used when historical sales data is available to forecast future sales, therefore cannot be used with new products
- A Moving Average forecast of demand uses the average sales from previous periods to forecast future sales
- The Exponential Smoothing Method uses an exponentially weighted moving average of previous sales; best used when drastic changes have recently occurred in marketplace; i.e. recent observations are weighted more heavily than older ones
- Be careful not to forecast past the terminal value of m (total market size)
No comments:
Post a Comment