When do changes in consumer preferences make forecasts from choice-based conjoint models unreliable?

Sheik Meeran, Semco Jahanbin, Paul Goodwin, Joao Quariguasi Frota Neto

Research output: Contribution to journalArticle

7 Citations (Scopus)
122 Downloads (Pure)

Abstract

Forecasting the sales or market share of new products is a major challenge as there is little or no sales history with which to estimate levels and trends. Choice-based conjoint (CBC) is one of the most common approaches used to forecast new products’ sales. However, the accuracy of forecasts based on CBC models may be reduced when consumers’ preferences for the attributes of products are labile. Despite this, there is a lack of research on the extent to which lability can impair accuracy when the coefficients estimated in CBC models are assumed to be constant over time. This paper aims to address this research gap by investigating the prevalence of lability for consumer durable products and its potential impact on the accuracy of forecasts. There are reasons to expect that lability may be particularly evident where a product is subject to rapid technological change and has a short product life-cycle. We carried out a longitudinal survey of the preferences of 161 potential consumers relating to four different types of products. We established that for both functional and innovative products: (i) the CBC models vary significantly over time, indicating changes in consumer preferences and (ii) such changes may cause large differences in forecasts of the probabilities that consumers will purchase particular brands of products. Hence employing models where coefficients do not change over time can potentially lead to inaccurate market share forecasts for high-tech, short life-cycle products that are launched even a short time after the choice-based modelling has been conducted.
Original languageEnglish
Pages (from-to)512-524
Number of pages13
JournalEuropean Journal of Operational Research
Volume258
Issue number2
Early online date1 Sep 2016
DOIs
Publication statusPublished - 16 Apr 2017

Fingerprint

Forecast
Sales
Life cycle
Model
Life Cycle
Consumer preferences
Technological Change
Time Change
Coefficient
Inaccurate
Forecasting
Attribute
Vary
Modeling
Market share
New products
Product lifecycle
Coefficients
Estimate

Cite this

When do changes in consumer preferences make forecasts from choice-based conjoint models unreliable? / Meeran, Sheik; Jahanbin, Semco; Goodwin, Paul; Quariguasi Frota Neto, Joao.

In: European Journal of Operational Research, Vol. 258, No. 2, 16.04.2017, p. 512-524.

Research output: Contribution to journalArticle

@article{326a384e2a3a42d096b5d2ce0e3ed31f,
title = "When do changes in consumer preferences make forecasts from choice-based conjoint models unreliable?",
abstract = "Forecasting the sales or market share of new products is a major challenge as there is little or no sales history with which to estimate levels and trends. Choice-based conjoint (CBC) is one of the most common approaches used to forecast new products’ sales. However, the accuracy of forecasts based on CBC models may be reduced when consumers’ preferences for the attributes of products are labile. Despite this, there is a lack of research on the extent to which lability can impair accuracy when the coefficients estimated in CBC models are assumed to be constant over time. This paper aims to address this research gap by investigating the prevalence of lability for consumer durable products and its potential impact on the accuracy of forecasts. There are reasons to expect that lability may be particularly evident where a product is subject to rapid technological change and has a short product life-cycle. We carried out a longitudinal survey of the preferences of 161 potential consumers relating to four different types of products. We established that for both functional and innovative products: (i) the CBC models vary significantly over time, indicating changes in consumer preferences and (ii) such changes may cause large differences in forecasts of the probabilities that consumers will purchase particular brands of products. Hence employing models where coefficients do not change over time can potentially lead to inaccurate market share forecasts for high-tech, short life-cycle products that are launched even a short time after the choice-based modelling has been conducted.",
author = "Sheik Meeran and Semco Jahanbin and Paul Goodwin and {Quariguasi Frota Neto}, Joao",
year = "2017",
month = "4",
day = "16",
doi = "10.1016/j.ejor.2016.08.047",
language = "English",
volume = "258",
pages = "512--524",
journal = "European Journal of Operational Research",
issn = "0377-2217",
publisher = "Elsevier",
number = "2",

}

TY - JOUR

T1 - When do changes in consumer preferences make forecasts from choice-based conjoint models unreliable?

AU - Meeran, Sheik

AU - Jahanbin, Semco

AU - Goodwin, Paul

AU - Quariguasi Frota Neto, Joao

PY - 2017/4/16

Y1 - 2017/4/16

N2 - Forecasting the sales or market share of new products is a major challenge as there is little or no sales history with which to estimate levels and trends. Choice-based conjoint (CBC) is one of the most common approaches used to forecast new products’ sales. However, the accuracy of forecasts based on CBC models may be reduced when consumers’ preferences for the attributes of products are labile. Despite this, there is a lack of research on the extent to which lability can impair accuracy when the coefficients estimated in CBC models are assumed to be constant over time. This paper aims to address this research gap by investigating the prevalence of lability for consumer durable products and its potential impact on the accuracy of forecasts. There are reasons to expect that lability may be particularly evident where a product is subject to rapid technological change and has a short product life-cycle. We carried out a longitudinal survey of the preferences of 161 potential consumers relating to four different types of products. We established that for both functional and innovative products: (i) the CBC models vary significantly over time, indicating changes in consumer preferences and (ii) such changes may cause large differences in forecasts of the probabilities that consumers will purchase particular brands of products. Hence employing models where coefficients do not change over time can potentially lead to inaccurate market share forecasts for high-tech, short life-cycle products that are launched even a short time after the choice-based modelling has been conducted.

AB - Forecasting the sales or market share of new products is a major challenge as there is little or no sales history with which to estimate levels and trends. Choice-based conjoint (CBC) is one of the most common approaches used to forecast new products’ sales. However, the accuracy of forecasts based on CBC models may be reduced when consumers’ preferences for the attributes of products are labile. Despite this, there is a lack of research on the extent to which lability can impair accuracy when the coefficients estimated in CBC models are assumed to be constant over time. This paper aims to address this research gap by investigating the prevalence of lability for consumer durable products and its potential impact on the accuracy of forecasts. There are reasons to expect that lability may be particularly evident where a product is subject to rapid technological change and has a short product life-cycle. We carried out a longitudinal survey of the preferences of 161 potential consumers relating to four different types of products. We established that for both functional and innovative products: (i) the CBC models vary significantly over time, indicating changes in consumer preferences and (ii) such changes may cause large differences in forecasts of the probabilities that consumers will purchase particular brands of products. Hence employing models where coefficients do not change over time can potentially lead to inaccurate market share forecasts for high-tech, short life-cycle products that are launched even a short time after the choice-based modelling has been conducted.

U2 - 10.1016/j.ejor.2016.08.047

DO - 10.1016/j.ejor.2016.08.047

M3 - Article

VL - 258

SP - 512

EP - 524

JO - European Journal of Operational Research

JF - European Journal of Operational Research

SN - 0377-2217

IS - 2

ER -