consumer driven product development for export markets

Consumer Driven Product Development For Export Markets

Chris Findlay February 18, 2015 Alberta Agriculture/CIFST

•  Sensory Science as a Tool •  Sensory Maps provide guidance for NPD •  Case Studies

–  Red Wine – Cabernet Sauvignon –  Whole Grain Bread –  Asian Noodles -‐ CIGI

Overview

2

3

Consumer Driven PD

The success of products developed for export markets increases greatly when developers have specific consumer-‐driven sensory targets to work with.

The Scallop Story

SoU as BuVer, Tough as Rubber It’s all in the cooking

4

Good fun! Poor science!

6

The Beef Story

Tenderness and Cooking

7

•  The Coke-‐Pepsi Challenge – Difference Tests

•  Product profiles – Descrip]ve Analysis using trained panels

•  Consumer Tes]ng – Product liking with larger groups

What do we really do in Sensory?

9

Compusense Panel

10

The Infamous Beer

Accelerated Batch Fermenta]on (25 days to 15 days) Accurate Balanced Fermenta]on

11

PC

2

PC 1

Product Liking Clusters

X2

B1

G8

G1

G4 X1

G2

Products that are found close together are in the same liking cluster

CLUSTER 1

CLUSTER 2

Strawberry Jam Consumer Liking Clusters

Chunks of Fruit

Smooth and Sweet

12

•  Product taste tests were once the staple of product development or quality control departments.

•  The ques]ons were simple and the results were basic.

•  Sensory science has moved forward from then through the applica]on of basic science to build a robust understanding of sensory evalua]on, consumer response and consumer choice.

Start off simple…

13

•  Discrimina]on –  Difference Tests

•  Formula]on Changes •  Process Improvements •  Mul]-‐plant produc]on •  Ingredient Subs]tu]on

Product Matching

14

•  Analy]cal sensory profiles of products that are both accurate and precise.

•  A library of the sensory proper]es of prototypes

•  Reliable measures of sensory shelf life.

Calibrated Descrip]ve Analysis

15

The Ballot

16

The Response

17

Immediate Feedback

18

FCM Introduced in 2006

|

2000’s |

|

1990’s

2010’s

40 hours

20 hours

6 hours

The Effec]veness of FCM Training

19

What we know about products •  There is no product that is “universally” liked, even water.

•  Op]miza]on of products is essen]al to achieve efficiency and market success

•  To op]mize, you must have a clear target. •  Unless you segment your consumers based upon their sensory preference you will not have a clear target.

20

What we know about consumers •  There is no such person as an “average” consumer.

•  There is no product that is “universally” liked, even water.

•  Tradi]onal demographics are no indicator of consumer preference

•  Consumers Lie!

21

•  Cluster consumer based upon their liking of products and their behaviour.

•  Consumer-‐driven product development; crea]ng new products based upon sensory design, targeted on the desires and needs of specific consumer groups.

22

Beyond Tradi]onal Segments

Finding the right respondents

23

Data collection at stores

24


25


26

Make sure they are the real consumers!

27

•  Sensory and consumer research has taken advantage of the progress in compu]ng and communica]ons to be able to take its tests to consumer, wherever they are and to permit sensory laboratories to collaborate on a global scale.

Exploi]ng digital technology

30

Web-Based Data Collection

Global Platforms Consistent methods Sophisticated tests Central management SaaS Management

31

Web-Based Data Collection

The world has changed:

• DA now performed at home

• HUT tests done on the web •  Smart phones allow tests to be done online anywhere

32

•  Consumer segmenta]on is important to understand liking

•  Consumer-‐driven product development works

•  Large consumer tests are expensive

•  Large consumer tests take ]me and resources

Consumer Category Tests

33

•  The Risks –  Fa]gue –  Carry-‐over effect –  Boredom. –  Consumers behaving like experts –  Resources

•  The Remedies –  Tes]ng at a single event –  Incomplete Block Designs

•  The Challenges –  Missing data –  Valida]on

Considera]ons for Large Studies

34

The Effect of Order and Day on Consumer Liking 12 White Wines, 115 Consumers, CBD 12:12 over 3 Days

0

10

20

30

40

50

60

70

1 2 3 All

Day 1 Day 2 Day 3 Combined

Posi]ons 1st 2nd 3rd 4th

35

Sta]s]cal Challenge •  A valid approach to segmenta]on of consumer BIB data •  Using a combina]on of sensory best prac]ce, experimental

design and advanced sta]s]cal analysis

Sensory Informed Design Method Development

36

Cabernet Sauvignon Study

•  A study of 12 Cabernet Sauvignon wines was conducted using over 600 recruited consumers and tested for liking

•  Consumers sampled 3 of the 12 wines in a BIB design

•  Data was analyzed for liking clusters with missing data replaced with consumer’s individual mean

•  Four liking clusters successfully demonstrated different sensory liking profiles

37

3.00 4.00 5.00 6.00 7.00 8.00 9.00

W1

W2

W3

W4

W5

W6

W7

W8

W9

W10

W11

W12

Cabernet Sauvignon Mean Liking

38 38

3.00 4.00 5.00 6.00 7.00 8.00 9.00

W1 W2 W3 W4 W5 W6 W7 W8 W9

W10 W11 W12

3.00 4.00 5.00 6.00 7.00 8.00 9.00

W1 W2 W3 W4 W5 W6 W7 W8 W9

W10 W11 W12

3.00 4.00 5.00 6.00 7.00 8.00 9.00

W1 W2 W3 W4 W5 W6 W7 W8 W9

W10 W11 W12

3.00 4.00 5.00 6.00 7.00 8.00 9.00

W1 W2 W3 W4 W5 W6 W7 W8 W9

W10 W11 W12

Cluster 1 – 28% Cluster 2 – 23%

Cluster 3 –32% Cluster 4 – 17%

39

Cabernet Sauvignon Study: Results

•  Findings demonstrated that although the method was not robust, the approach gave useful and ac]onable results

•  A research program was ini]ated to develop a systema]c approach to building designs using sensory informa]on to ensure contrast

40

Is there a Preference? •  To state a true preference a consumer must be

able to see a real difference •  Otherwise it’s just a guess •  Consequently we must present the consumer

with truly different samples

Sensory Design

41

Let’s consider a sensory space

Can we find logical contrasts to test?

Factor Scores plot : dimension 1 versus 2

-2.58 2.58

-2.58

2.58

Tobacco

Smoke

Asparagus

Coffee

FloralVanilla

Green

WoodyPepper

EucalyptusFruityRaisin

Leather

Sweet

Sour

BitterAstringent

W1

W2

W3

W4W5

W6W7

W8

W9

W10

W11W12


-2.58 2.58

-2.58

2.58

TobaccoSmoke

AsparagusCoffeeFloralVanilla

GreenWoodyPepper

Eucalyptus

FruityRaisin

Leather

Sweet

Sour

BitterAstringent

W1

W2W3

W4

W5

W6

W7

W8

W9W10

W11

W12

42

Quadrangles and Triangles Factor Scores plot : dimension 1 versus 2

-2.58 2.58

-2.58

2.58

Tobacco

Smoke

Asparagus

Coffee

FloralVanilla

Green

WoodyPepper

EucalyptusFruityRaisin

Leather

Sweet

Sour

BitterAstringent

W1

W2

W3

W4W5

W6W7

W8

W9

W10

W11W12


-2.58 2.58

-2.58

2.58

TobaccoSmoke

AsparagusCoffeeFloralVanilla

GreenWoodyPepper

Eucalyptus

FruityRaisin

Leather

Sweet

Sour

BitterAstringent

W1

W2W3

W4

W5

W6

W7

W8

W9W10

W11

W12

43

Whole Grain Bread Study

In a 2012 study of whole grain breads, 570 consumers 16 samples using an improved SID of 16:6, with nested designs of 16:3 and 16:4

44

GPA of 16 Whole Grain Breads 55 Sensory Attributes

-1.10 1.10

-0.50

0.50

Variance Accounted for: DIM1 – 66% DIM2 – 9% DIM3 – 7% DIM4 – 4%

45

3 4 5 6 7 8 9

Overall Liking All 570 Category Consumers

46

7.1

6.8

6.6

6.5

6.4

6.2

5.9

5.9

5.8

5.8

5.7

5.5

5.5

5.2

5.0

7.3

J LOVE IT!

L HATE IT!

46

2 3 4 5 6 7 8 9

Cluster 1 Overall Liking 25.8 % 147 Consumers

47

7.4

6.7

6.7

6.6

6.5

6.1

5.8

6.0

4.9

5.6

4.9

4.4

4.5

4.4

3.2

7.8

J LOVE IT!

L HATE IT!

47

3 4 5 6 7 8 9


48

7.9

7.8

7.7

7.5

7.3

7.0

6.6

6.9

6.3

6.5

6.0

5.9

5.9

5.7

5.6

8.0

J LOVE IT!

L HATE IT!

48

3 4 5 6 7 8 9


49

6.9

6.2

6.1

5.9

5.7

5.5

5.3

5.5

4.6

4.9

4.6

4.5

4.5

4.5

3.7

7.6

J LOVE IT!

L HATE IT!

49

•  Improves the efficiency of large studies •  Can be used for any product category •  Improves the quality of data collected •  Delivers actionable consumer clusters •  Saves resources

Conclusions about Sensory Informed Design

50

Combining highly efficient methods can reduce costs

without compromise

51

How is it that we never have enough ]me to do the job right,

but always enough ]me to do it over?

52

1.  Provide an effec]ve strategy for category assessments

2.  Reduce large numbers of possible test products

3.  Understand the product sensory space

Objec]ves

53

4.  Design highly efficient consumer studies

5.  Combined methods

6.  Deliver reliable and robust outcomes

Objec]ves

54

1.  Projec]ve Mapping (Napping)

2.  Calibrated Descrip]ve Analysis (FCM)

3.  Sensory Informed Design

The Methods

55

4.  E-‐M Imputa]on of Missing Data

5.  Cluster Analysis on Consumer Liking

6.  Correla]on of Sensory and Consumer Data

The Methods

56

Projec]ve Mapping Whole grain breads 50 products to a set of 16

57

GPA of 16 Whole Grain Breads 55 Sensory AVributes

-1.10 1.10

-0.50

0.50

Variance Accounted for: DIM1 – 66% DIM2 – 9% DIM3 – 7% DIM4 – 4%

58

Sensory Informed Balanced Incomplete Block Design (SID)

Sample sets that •  maximize sensory contrast •  ensure consumer liking •  results reflect consumer sensory preference

59

SID Procedure

60

SID Consumer Studies

Whole Grain Bread Nested design 16 present 3 16 present 4 570 consumers 6 products for liking

61

Conclusions

•  Projec]ve Mapping

–  efficient

–  selects a representa]ve sensory product set.

•  FCM trained DA panel

–  less than half the ]me

–  greater precision.

62

Conclusions

•  SID consumer research

–  Eliminates fa]gue or boredom

–  provides a solid basis for consumer segmenta]on.

•  EM Imputa]on

–  realis]c values for the missing data

–  integrates with clustering to iden]fy liking segments.

63

Pusng it all together

Projec]ve Mapping

Calibrated Descrip]ve Analysis

SID-‐based Consumer Research

Clustering and

Correla]on

64

Asian Noodles Descrip:ve Study

March 2014

Canadian Interna:onal Grains Ins:tute

65

Objec:ve

•  To compare the sensory profiles of two types of instant noodle product, Deep Fried and Steam Dried, against benchmark products that are popular in the target market, China.

•  To use the sensory differences to guide product improvement in the noodle prototypes being developed at CIGI.

66

67

Deep Fried Instant Noodles

Table 1. Deep Fried Test and Benchmark Products Product Name Category Best Before Date UPC

Deep Fried Instant Noodles 5% Low Protein Pea Flour Prototype N/A N/A Deep Fried Instant Noodles 5% High Protein Pea Flour Prototype N/A N/A Deep Fried Instant Noodles 15% Low Protein Pea Flour Prototype N/A N/A Deep Fried Instant Noodles 15% High Protein Pea Flour Prototype N/A N/A Master Kong Instant Noodle Benchmark May 29, 2014 6900873059010 Doll Instant Noodle Benchmark November 5, 2014 079551730445

Products

Test products were provided by Canadian Interna5onal Grains Ins5tute. Benchmark products were procured by Compusense from local Asian grocery stores. Tes5ng was conducted at the Compusense Research Facility in Guelph, ON the week of March 17, 2014

68

Deep Fried Benchmark #1: Master Kong (康师傅)

•  It is a product of China. •  Based on our research this is a long exis]ng and popular brand of instant noodles that owns the most of market share in China.

•  The product represents generic deep fried instant noodle.

•  We have already sourced and purchased enough for this project.

69

Deep Fried Benchmark #2: Doll (公仔面)

•  It is a product of Hong Kong.

•  The product represents generic deep fried instant noodle.

•  This is certainly a popular brand in Hong Kong and Southern China.

70

Product Means and ANOVA

ATributes p value LSD Fried 15% High Fried 5% High Fried 15% Low Fried 5% Low Master Kong Doll

Overall Colour 0.00 3.5 64.0 a 48.1 c 58.3 b 47.5 c 57.9 b 57.4 b

Surface Shine 0.00 2.8 55.2 b 55.5 b 51.9 c 57.2 b 64.1 a 61.6 a

Overall Aroma 0.00 1.5 30.5 b 29.3 b 29.2 b 29.5 b 32.1 a 29.5 b

Starch Aroma 0.25 -‐ 17.5 17.0 17.6 17.2 18.3 17.3

Green Aroma 0.03 1.3 1.4 ab 2.1 a 2.6 a 1.8 ab 2.7 a 0.7 b

Overall Wheat Aroma 0.36 -‐ 18.4 17.1 17.3 17.8 17.5 17.9

Oil Aroma 0.99 -‐ 11.3 11.1 10.8 10.9 10.9 11.1

Surface S]ckiness (hands) 0.00 2.3 28.6 cd 29.7 c 27.2 d 29.7 c 38.3 a 33.1 b

Elas]city 0.01 3.7 35.3 b 35.2 b 33.3 b 36.5 ab 34.6 b 39.9 a

Surface Roughness 0.50 -‐ 22.9 22.4 21.7 21.7 21.4 22.7

Surface S]ckiness (mouth) 0.00 2.6 30.5 cd 32.6 bc 28.6 d 32.0 c 35.8 a 34.8 ab

Firmness 0.00 1.7 42.2 a 40.3 b 41.4 ab 40.4 b 33.8 d 38.1 c

Chewiness 0.00 2.1 34.0 a 32.3 ab 32.6 ab 32.2 ab 24.9 c 30.6 b

Cohesiveness 0.00 1.6 23.8 a 21.9 b 22.7 ab 22.5 ab 19.7 c 21.4 b

Residue 0.02 1.4 19.2 ab 18.3 bc 19.1 ab 18.3 bc 17.3 c 19.7 a

Significant at p=0.05 Highest Score Lowest Score Scale: Unstructured Scale from 0 to 100

Table 2(a). Deep Fried Category Product Means and ANOVA

71

Product Means and ANOVA

ATributes p value LSD Fried 15% High Fried 5% High Fried 15% Low Fried 5% Low Master Kong Doll

Overall Flavour 0.02 1.6 28.2 b 28.0 b 28.2 b 28.4 b 30.4 a 29.5 ab

Starch Flavour 0.64 -‐ 17.0 17.1 16.4 17.5 17.1 16.9

Green Flavour 0.01 0.9 1.1 b 1.1 b 1.6 ab 1.1 b 2.4 a 1.0 b

Overall Wheat Flavour 0.12 -‐ 18.2 17.1 17.7 17.8 16.6 18.1

Oil Flavour 0.09 -‐ 13.3 13.4 13.2 13.4 13.6 14.8

Sal]ness 0.00 1.5 20.3 b 19.2 b 20.0 b 20.3 b 21.8 a 22.0 a

Sourness 0.57 -‐ 6.8 6.9 7.1 7.0 7.5 7.2

Sweetness 0.01 1.2 15.3 b 15.2 b 15.4 b 15.4 b 16.8 a 16.7 a

BiVerness 0.53 -‐ 8.3 8.4 9.0 8.9 9.1 9.0

Overall AUertaste 0.46 -‐ 20.7 19.9 21.0 20.4 20.8 20.6

Starch AUertaste 0.29 -‐ 12.8 13.6 13.4 13.4 13.9 12.8

Green AUertaste 0.01 0.7 0.7 b 1.0 b 0.8 b 0.9 b 1.8 a 0.6 b

Wheat AUertaste 0.43 -‐ 13.5 13.1 13.1 12.6 12.6 13.2

Sal]ness AUertaste 0.00 1.2 12.8 c 13.0 c 13.5 bc 12.8 c 14.5 ab 14.7 a

Significant at p=0.05 Highest Score Lowest Score Scale: Unstructured Scale from 0 to 100

Table 2(b). Deep Fried Category Product Means and ANOVA (con%nued)

72

Principal Component Analysis (Covariance) of Fried Instant Noodles (PC1 versus PC2)

Fried 5% Low

Master Kong Doll

Overall Colour

Surface Shine

Firmness Chewiness

Cohesiveness

-‐25

25

-‐25 25 Surface S]ckiness (mouth)

Fried 15% High

Surface S]ckiness (hand)

Fried 5% High

Fried 15% Low

Variance Accounted for: PC1 – 54% PC2 – 35% PC3 – 8% PC4 – 2%

PC1

PC2

73

Differen:al Profiles of Significant ATributes

The charts are created by subtrac]ng the profile of one product versus another product of interest to highlight the aVributes that are different.

74

Difference from the Doll Benchmark

-‐10.0 -‐8.0 -‐6.0 -‐4.0 -‐2.0 0.0 2.0 4.0 6.0 8.0 10.0

Overall Colour

Surface Shine

Overall Aroma

Green Aroma

Surface S]ckiness (hands)

Elas]city

Surface S]ckiness (mouth)

Firmness

Chewiness

Cohesiveness

Residue

Overall Flavour

Green Flavour

Sal]ness

Sweetness

Green AUertaste

Sal]ness AUertaste

Fried 5% Low

Fried 5% High

Figure 1. 5% Low and 5% High Mean Differences from the Doll Benchmark Product

Greater Than Doll Benchmark

Less Than Doll Benchmark

75

Difference from the Doll Benchmark Figure 2. 15% Low and 15% High Mean Differences from the Doll Benchmark Product

-‐10.0 -‐8.0 -‐6.0 -‐4.0 -‐2.0 0.0 2.0 4.0 6.0 8.0 10.0

Overall Colour

Surface Shine

Overall Aroma

Green Aroma


Elas]city


Firmness

Chewiness

Cohesiveness

Residue

Overall Flavour

Green Flavour

Sal]ness

Sweetness

Green AUertaste

Sal]ness AUertaste

Fried 15% Low

Fried 15% High

Greater Than Doll Benchmark

Less Than Doll Benchmark

76

-‐12.0 -‐8.0 -‐4.0 0.0 4.0 8.0 12.0

Overall Colour

Surface Shine

Overall Aroma

Green Aroma


Elas]city


Firmness

Chewiness

Cohesiveness

Residue

Overall Flavour

Green Flavour

Sal]ness

Sweetness

Green AUertaste

Sal]ness AUertaste

Fried 5% Low

Fried 5% High

Difference from the Master Kong Benchmark Figure 3. 5% Low and 5% High Mean Differences from the Master Kong Benchmark Product

Greater Than Master Kong Benchmark

Less Than Master Kong Benchmark

77

-‐15.0 -‐10.0 -‐5.0 0.0 5.0 10.0 15.0

Overall Colour

Surface Shine

Overall Aroma

Green Aroma


Elas]city


Firmness

Chewiness

Cohesiveness

Residue

Overall Flavour

Green Flavour

Sal]ness

Sweetness

Green AUertaste

Sal]ness AUertaste

Fried 15% Low

Fried 15% High

Greater Than Master Kong Benchmark

Less Than Master Kong Benchmark

Difference from the Master Kong Benchmark Figure 4. 15% Low and 15% High Mean Differences from the Master Kong Benchmark Product

78

•  Consumer segmenta]on is essen]al to understand liking

•  Large consumer tests are expensive

•  Large consumer tests take a lot of ]me and resources

•  Integrated methods are efficient

Consumer Category Understanding

79

Consumer-‐Driven Product Development

WORKS

80

•  Analy]cal sensory product profiles

•  Accuracy and precision

•  Reliable measurement

•  Advanced sta]s]cs

•  Publish innova]ve methods

Science-‐based and Research Focused

81

compusense.com +1 519 836 9993 – Worldwide

1 800 367 6666 – Toll free in North America [email protected]

Thank you!

consumer driven product development for export markets

Data & Analytics