Perceptual Mapping Techniques

Perceptual Map
Need 2
SELF SONO

+20 Bu

SEMI SUL I SOLD

Pr Hi Si +20 Need 1

-20

SALT SIBI SUSI SAMA

Ot
SIRO

-20

Semantic Scaling
Research Illustration

‡ How sweet is your ideal cola ? ‡ How important is it to you that a cola have the proper sweetness ? ‡ How closely does brand X match to your ideal sweetness ?

Very=4

Somewhat=3

Not much=2

Not at all=1

Semantic Scaling
‡ Large samples (typically)
Survey-based methodology

‡ A priori selection of attributes
Unimportant attributes get low ratings Important attributes may be overlooked overlooked

‡ Limited rating scale
Constrained upper & lower ratings Gradients may not adequately differentiate Implicitly assumes linear relationships

‡ (Relatively) easy understand & apply

Conventional Mapping
Snake Chart
Describes it completely | | 0 1 | 2 | 3 Does not describe it at all | | 4 5
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Company provides adequate insurance coverage for my car. Company will not cancel policy because of age, accident experience, or health problems. Friendly and considerate. Settles claims fairly. Inefficient, hard to deal with. Provides good advice about types and amounts of coverage to buy. Too big to care about individual customers. Explains things clearly. Premium rates are lower than most companies. Has personnel available for questions all over the country. Will raise premiums because of age. Takes a long time to settle a claim. Very professional/modern. Specialists in serving my local area. Quick, reliable service, easily accessible. A ³good citizen´ in community. Has complete line of insurance products available. Is widely known ³name company´. Is very aggressive, rapidly growing company. Provides advice on how to avoid accidents.

¥©¨ ¤ ¨ ¡ §¤ ¡ ¡ ¥¡ © §¨¤ ©¨ §¡ # ¦¥¤£¢¡ ¥ ¡  ¦ ¨ £¤ © § © ¤ ¦ © § ¦¥¤£¢¡ ©¢¤¥ ¥ ¡¥ ¦ ©¨  © ¤ ¤ §¤ ¨¡ £ © ¥¤  ¥ ¡ ©¥ © © £¢¡ ¤( ¦ ¥¢ ¢¡ ¥ ¥© ¨¡¡ 2 ©  © ¤ ¦ ¤© © § © © ¤ ©  1 ¤© ¤ ¤ ¡ ¦ ¢ ¥ § © ¥ ¤ ©£ ¥ ©¨¡¢ ¤¥¡ © ¡ £ ¦ © 0 ¢ ¤ ¤ © © ¡ ©¢ ¥¡ ¤ © ¤ ©¤ ¡ © ¤ © ¢ ¢© £ © ¤ ) ¦ ¥¡ © © §¡ ¤ ¥¡ © ¡ © ¤ ¤ §¤ ©¥¥¡ ©£ ¤( © ¥¤£ ¢¡ ¡¢ ¥¤ © ¡ © ¤ © ¤ ¢ ¢© # ¦ ¤© ¥  ¥ ¤ £& ©¢¡  ¤¨ § ¨¥ ¡¤ © ¤ ¡   ¡¡ ¦ ¡ ©¤ © §¡ ¡ ¥¡¢¤ ¨¥¤ ©£¦ ¡¤ © §¨¤ ¨¡¡ ©¨ §¡ #   ¤©¨ ¡ ¨ ¤ ¥© ©¥ ¦ ¤ ¢¤ © © © ¤ ©¨ ¥¡ ¨¥¤ ¦ ¨¥©  ¢© ¡ £  ¤© ¡ © ¥© ©£© ¥©¨ ¤ ©¤ ¡ © ¤ © ¦ ¡£ © ¥¤ ¡¥  ¦¥¤£¢¡   ¤ ¦¢ ¡ ©¤ © §¡ © ¥¤  ¥ © ¤©¨¤ ©¨ §¡ £ ¦¥¤£¢¡  
T I S ttl s cl i s f irl . Is Is S T k s ri r r r ³ ill r is r sc s ick, r li ffici lt l i st i ci lists i s r i i i i r i i t c r l r f ssi l k rs s r l t, s citiz r t s r l l t li r ssi r l r ill c ti l i i l s r ic , ic r t s cl ´i c s. ³ tc si l/ s ,r s il t s ttl ic fi s r rl . ti r t . c l l l f r i l c c r . rt t i i it . l c l r sil s t i s r tt it . lic r c cl i . i f l c st cc ssi l . i sti ´. s r st c c cci . c . s ll c c cts s rs. ts. f i s. . il r rt ts f c , cci l . f r c tr . r t c r. t .

Describes it completely

0

1

2

3

4

Does not describe it at all

Snak

5

ti

20. 19. 18. 16. 15. 14. 13. 12. 11. 10. 1. . . . . 17.

' % $ " !  

. . . .

l

art

i

ri c , r

Perceptual Map
High Price

G
Low C Quality

E D B

A
High Quality

F
Low Price

Perceptual Map
High Price

G
Low C Quality

E D B

A
High Quality

F
Low Price

Perceptual Map
High Price

G
Low C Quality

E D B

A
High Quality

F
Low Price

Ideal Points
‡ Customer perceptions ‡ Aggregation of individuals
«Distributions around points

‡ Different shapes
«Optimal points, vectors

‡ Segment variations ‡ Evolutionary progression
«Nice to have => Must have

Preference Models
‡ Ideal points (individuals) ‡ Clusters (segments) ‡ Proximity (preference)

Perceptual Map
High Price

G
Low C Quality
1

E
2

A D
3

B F
Low Price

High Quality

In general ...
‡ Most of a brand¶s sales will come from the segments with the closest ideal points ‡ Most of a segment¶s sales (share) will go to the brands closest to its ideal point

Targeting Strategies
‡ Direct hit «
single product µright on¶

‡ Bracketing
multiple products µsurround¶

‡ ³Tweeners´
single product µsplitting the difference¶ to induce a new segmentation

Multidimensional Scaling (MDS)
‡ Rank pairs of products (brands) by degree of similarity
A is more like B than B is like C

‡ Statistically µreduce¶ the data to a 2-dimensional mapping
Usually a µblack box¶ application

‡ Judgmentally interpret the axes
Multi-dimensionally Mix of art and science

Beer Market
Perceptual Mapping ‡ Budweiser
Beck¶s Miller

Old Milwaukee Meister Brau

‡

‡
Stroh¶s

‡
Coors

‡ Heineken ‡

‡

‡

‡ Michelob ‡
Miller Lite

‡

‡

Coors Light

Old Milwaukee Light

Beer Market
Perceptual Mapping
Heavy Popular with Men Beck¶s Full Bodied Old Milwaukee Meister Brau

‡ Budweiser ‡

‡

‡
Good Value

Miller Blue Collar Stroh¶s

Special Occasions

‡ Heineken
Dining Out Premium

‡

‡

‡ Coors

‡ Michelob ‡
Pale Color Miller Lite Light

‡

‡

Coors Light

Popular with Women

On a Budget

Old Milwaukee Light

Less Filling

Beer Market
Perceptual Mapping
Regular
Full Bodied Heavy Popular with Men

Special Occasions Good Value Blue Collar Dining Out Premium

Budget

Premium
Popular with Women Light Less Filling

Pale Color On a Budget

Light

Beer Market
Perceptual Mapping
Regular
Full Bodied Old Milwaukee Meister Brau Heavy Popular with Men Beck¶s

‡ Budweiser ‡

‡

‡
Good Value

Miller Blue Collar Stroh¶s

Special Occasions

‡ Heineken
Dining Out Premium

‡

Budget

‡

‡ Coors

Premium

‡ Michelob ‡
Pale Color Miller Lite Light

‡

‡

Coors Light

Popular with Women

On a Budget

Old Milwaukee Light

Light

Less Filling

Beer Market
Perceptual Mapping
Regular
Old Milwaukee Meister Brau

‡ Budweiser
Beck¶s Miller

‡

‡
Budget
Stroh¶s

‡
Coors

‡ Heineken ‡
Premium

‡

‡

‡ Michelob ‡
Miller Lite

‡

‡

Coors Light

Old Milwaukee Light

Light

Multidimensional Scaling
‡ Smaller samples (than semantic scaling)
Very high cost methodology

‡ Requires extensive interpretation
By definition, results are equivocal

‡ Conventional wisdom: ³more precise´
How does anybody know?

‡ Separate effort to juxtapose preferences
Derived from brand rankings µJoint space¶ maps

Conjoint Measurement
‡ Pairs of tightly defined alternatives
Reduced attribute set Specific attribute values µOrthogonal arrays¶

‡ Computed µutility¶ weights
Based on pairwise preferences If added, reflect original preferences Basis for inferences re: attribute importance weights

Conjoint Measurement
‡ Smaller samples (than semantic scaling)
Very high cost methodology

‡ Requires extensive interpretation
Highly complex, hardly intuitive

‡ Basis for strong insights
Potentially dangerous if used literally