Consumer Behaviour

A presentation on strategic marketing models

By

Group 1

Group Members

Sameer Sanghani

Neha Chaudhary

Atit Shah

Deesa Kamdar

Joanna Barsey

Sagar Thukral

Samina Papeya

Nirav Shah

Foundations of Consumer Behaviour Models

Consumer behaviour models are based on certain traditions:

• Behavioural Learning Under this theory consumers’ prior experience is the primary

determinant of future behaviour

• Personality Research Consumer attempts to reconcile his behaviour, others’ behaviour,

the state of his environment etc with prior beliefs.

• Information Processing

The characteristic nature of the individuals customers’ decision making process and changing belief

• Attitude Models Product attributes are drivers of the consumer decision process

Nature of Consumer Behavior models

•Consumers are Different

•Choice Decisions differ

•Context of Purchases Differ

Chapter Overview

-Variety Seeking Models-Satisfaction Models-Commu and Network models

Post-Purchase

-Multinominal Discrete Choice -Markov models

Purchase

-Perceptual mapping -Attitude models

Evaluation

-Individual Awareness Models-Consideration Models-Information Integration Models

Information Search

-Stochastic Models of Purchase Incidence-Discrete Binary Choice Models

Need Arousal

Stochastic Models

Brand Choice Model

Consumer Behavior

Multiple approaches for modeling consumer behavior

For low involvement products- little conscious random decision making takes place; STOCHASTIC models are appropriate

Concentrate on Random nature of the choice process than on a deterministic explanation

Stochastic Model-Brand Choice

B.C.model can be differentiated by how they deal with

Population heterogeneity

Purchase event feedback

Exogenous market factors

Purchase Feedback Models

Zero order model- assume no feedback

Markov model- assume only previous brand choice affect present event feedback

Learning model- assume entire purchase history affects current choice (recent having more effect)

Example Purchases of Brands Table Brand bought on Occasion 2 Brand bought on Occasion 1 Joint Probability Table Brand bought on Occasion 2 Brand bought on Occasion 1 Conditional Probability Table Brand bought on Occasion 2 Brand bought on Occasion 1

A B C Total A 137 47 19 203 B 41 179 12 232 C 22 10 46 78

Total 200 236 77 513

A B C Total A 0.267 0.092 0.037 0.396 B 0.080 0.349 0.023 0.452 C 0.043 0.019 0.09 0.152

Total 0.39 0.46 0.15 1.00

A B C Total A 0.674 0.232 0.094 1.00 B 0.177 0.772 0.051 1.00 C 0.283 0.128 0.59 1.00

p(i,j) = joint probability that a consumer will purchase brand i on the second purchase occasion and brand j on the first purchase occasion

P(i/j) = conditional probability that a customer will purchase brand i on the second purchase occasion given that brand j was purchased on the first occasion

Purchase Probability are related as follows:

p(i/j) = p(i,j) ----- (1) p(j) where p(j) = probability of purchasing brand j on

the first purchase occasion so p(j) = mj , the market share of brand j ∑p(i,j) = mi ----- (2) j ∑ p(i/j) = 1 i

Zero Order Model

1. The assumptions they make about consumer preferences and choice

2. The number of brand they consider

Moderately Heterogeneous Population (2)

Heterogeneous Population (1)

Extremely Heterogeneous

Population ( Mostly Brand Loyal (3)

f (p

), D

istr

ibu

tion

Act

ors

Pop

ula

tion

P, Probability of Purchase

Bernoulli Model

Simple Multiple Brand model

Joint probability of a consumer purchasing brand i and j on successive purchase occasions

p(i,j) = kmimj ----- (3)

where mi and mj are the market share of respective brand

p(i,i) = mi – kmi(1 - mi) ----- (4)

From (4),k = 1 - ∑p(i,i) ----- (5) 1 - ∑mi2

p(i) = mi ----- (6)

p(i/j) = kmi ( j = i ) ----- from (1) = 1 – k(1 - mi) ( j = i ) ----- from (1,4,6)

Brand Shares (mi) mi*(1- mi)

A 0.39 0.238

B 0.46 0.248

C 0.15 0.128

1-∑p (i,i)

k= 1- ∑mi²

∑p (i,i)= diagonal values of Joint Probability Table = 0.267 +0.349+0.09 = 0.706

k= 0.294/0.614 = 0.479

Example Contd.

Markov Model

Assumptions – Stationary (probabilities do not change) Homogeneity One purchase per time period

Zero Order Model – Brand choice is independent

Other models assume – Post purchase event feedback .

Markov Model assumes that only the previous brand purchase affects the present purchasing choice

Markov Models

Brand Choice – Pij

Given current market share , a Markov Model can be used to to predict how market shares change over time.

m = ∑ p m i,t + 1 ij it

Markov Models – Example

Consider two brand A and B with following switching matrix

A B

A 0.7 0.3

B 0.5 0.5

t + 1

t

2 Uses of Markov Model –

Forecasting of the market share with the help of transition matrix

How the effect of change in market structure can be evaluated

Markov Models

Price shift

Limitations

Stationarity - unrealistic a firm loosing market position will

take corrective action.

Post Purchase and purchase feedback

After purchasing and experiencing a product, a consumer’s reaction is important for future purchases.

Biehal (1983) showed that, for auto repair services, the outcome of prior experience is more important than external search in choosing the next service provider.

Post purchase behavior affects attitude which in turn affects the consumer’s behavior.

Post purchase affects how the consumer communicates about the product through WOM.

A purchase can affect future purchases through variety seeking.

Bearden and Teel (1983) used structural equation model to explain the post purchase effects.

Post Purchase and purchase feedback

Disconfirmation

Expectations

Attitude 1

Intention 1

Satisfaction/ Dissatisfaction

Attitude 2

Intention 2

Complain Reports

+

+

+

+

+

+

+

-

T1 Current T2 Future

+ Indicates Positive Effect

- Indicates Negative Effect

Lattin and McAlister modeled a consumer’s utility for brand on a given consumption occasion as a diminishing proportion

of the value of the features it shares with the brand the consumer chose on previous occasion.

Vi\j = Vi – λSijWhere,

Vi\j = Utility of i given that j was chosen previously

Vi = unconditional utility of i

λ = discount factor indicating consumer’s variety- seeking intensity.

Sij = value to the customer of all want-satisfying features shared by i and j

Pi\j = Vi\j(ΣkVk \j )

Applying Luce Model to this formulation gives the probability of purchase of i given a previous purchase

of j, Pi\j

~ Previous consumption alters the unconditional brand choice probability, Vi, that is,

~ Pi\j – Vi < 0, then j is a substitute for product i (the consumption of brand j lowers the probability of choosing brand i)

~ While Pi\j – Vi > 0, then j is a compliment for product i (the consumption of brand j increases the probability of choosing brand i)

(B) Noncompensatory Models1) Conjunctive Model :-

In a conjunctive model a consumer prefers a brand only if it meets certain minimum, acceptable standards on all of a number of key dimensions. If any one attribute is deficient, the product is eliminated from contention. Let

YJK = perceived level of attribute k in brand j

Tjk = minimum threshold level that is acceptable (negatively valued attributes such as price that have a maximum level can be multiplied by -1)δjk = 1, if brand j is acceptable on attribute k

0, otherwise

Aj = 1, if it is a preferred brand overall 0, otherwise

under the conjunctive model, we have δjk = 1, if Yjk ≥ Tjk

0, if Yjk < Tjk

Aj = Πk δjk

Thus, Aj will be nonzero if and only if Yjk ≥ Tjk for all attributes.

2) Disjunctive Model :-In this model, instead of preferred brands having to satisfy all of a number of criteria, they have to satisfy one of a number of criteria. The conjunctive model is often called the “ and ” model, while the disjunctive model is called the “ or ” model. Under the conjunctive model the consumer may insist on a product that has lots of memory and software. Under the disjunctive model the consumer may want to settle for either a product with a lot of memory or a lot of software.

Mathematically, we can express this as

Aj = min(Σkδjk, 1)

Where Aj and δjk are defined as before.

3) Lexicographic Model :- This model assumes that all attributes are used, but in a stepwise

manner. Brands are evaluated on the most important attributes first; then a

second attribute is used only if there are ties, and so forth.

Mathematically, if we assume that the attributes are arranged in order from most important to least important, then brand j is the preferred brand if:

yji > yjk for all brands k,k = 2,…,k