1 Two emerging opportunities for consumer research Innovation of Ideas Graphics Designs Howard R. Moskowitz Moskowitz Jacobs Inc. White Plains, NY USA.

11

Two emerging opportunities for consumer research

Innovation of IdeasGraphics Designs

Howard R. MoskowitzMoskowitz Jacobs Inc.White Plains, NY [email protected]

22

Opportunity # 1: Innovation

Bringing respondents into the design of features

33

A General Point Of View

Innovation may be accomplished by recombining old features into new mixtures

If so … then can we create a system to make this easy?

What do we do? What do we learn?

44

One way is to develop ‘creative consumers’

Find consumers who are articulate– Work with them– Get ideas– Hope that the system generates innovation

Or work with them and ‘creatives’– Use consumer inputs– But rely on ‘creatives’ to create/innovate

55

What does this imply

Trust in the ‘expert’ Outsource creation / innovation to

someone– Put your hopes in that person

If it were my money– I’d rather spend it on a machine that creates– Yes.. There is no soul … but the machine may

end up being more efficient

66

So let’s design our machine

Precompiled databases of featuresA combinatorial machine

A consumer who respondsAn analytic strategy to recombine

77

Why we did what we did Innovation is like the ‘weather’

– Everyone talks about it– There are metrics and books

But how ..how do we actually do it?– Inspiration?– Deus ex machina

88

The features of the machine Assembled Raw materials: A set of defined

databases prepared for conjoint analysis– Different product features – Already populated with actionable elements….the key step

New Idea Seeding: Within each database.. 3-4 more questions to open up the respondent mind– Problem scenarios demanding a specific solution

Combinatorics machine to create, test, use ideas– With specific, pre-set analysis strategies

Chance favors the prepared mind (Pasteur)– With high level optimization to create new, strong ideas

99

Getting to the heart of innovation

DatabasesDatabasesDatabasesDatabases

FieldworkFieldworkFieldworkFieldwork

AnalysisAnalysisAnalysisAnalysis

ResultsResultsResultsResults

1010

We’re going to talk about foods

But think of consumer electronics or any other area

1111

Part 1 – Precompiled database of raw materials (idealets)

Rationale: Do the client’s homework for them

If the client can’t or won’t think .. Then the researcher should think for the client

1212

Database means assembling raw ideas

Corporate culture believes in single & simple problems

However lots of innovation comes from broadening scope & using analogies– Need multiple studies– Independent of close-in needs of a manufacturer– If the work is done…people MAY actually use it

1313

Precompiled database of ideasCreate 70 databases of raw concept elements ahead of time

1414

Common structure, actionable ideasCreate/standardize six silos of ideas…..

With the elements ‘real’ and ‘appropriate’

– Appearance/Texture: What does it look like, feel like?

– Primary ingredients: What does it contain ?

– Secondary ingredients: What healthful features does it offer?

– Taste/Flavor: What does it taste like?

– Packaging: How is it stored or packaged?

– Merchandising: Where is it merchandised in the store, or how is it used?

1515

Make all the elements availableIf users see the full range ..they’ll work out of the box

1616

Problem-solutionSeeding the next generation ideas

We wanted to have the respondents help us ideate– Give them open ended problems– Ask for solutions– Do this with 3-4 questions per database

Rationale– Making things specific focuses their minds

1717

Examples of problem-solution

1818

Part 2 -- Fieldwork

If we’re going to have an invention machine…

Make it easy to inventMake data acquisition a ‘snap’

After all .. We’re half way there already with precompiled data bases

1919

Field ExecutionInternet-based conjoint

Easy to set up– Template driven

Easy to execute– Send to respondents– Get data– Automatic analysis

So far … so good– Precompiled ideas +– Simple data acquisition

2020

Example of concept + rating scale

2121

Part 3 - Analysis

Using the data to drive innovation by identifying what works & what doesn’t,

and where there are ‘synergisms’

2222

ModelingIdentifying what works

We set up the elements ..& varied them

Now let’s relate elements to ratings– Additive Constant = baseline interest without any elements– Utilities = additive or conditional probability of interest if

element is present in concept

Looking at– Total panel & key subgroups (standard stuff)– ‘Latent’ segments .. Groups with like minds (based on

patterns of utilities)

2323

Deconstruct concepts to componentsTotal panel, subgroups, even so-called latent segments

Go after Segment 2 – the Health Seekers

2424

2525

Part 4 - Results

Create new ideas

Combine ideas from different datasets according to evolution rules

2626

Use optimizer Select objectives

Mix & match components … making job easier

Objective..who / what we’re creating

Results … new combination

2727

Evolve, innovate by ‘rule based’ genetic algorithms

Merge winning ideas from different categories into a new product idea / even new category

Cookies Donut Chocolate Candy New Features Category #1 - Appearance/Texture

Bite size for a quick indulgence

Bite-size honey glazed donut holes with swirls of vanilla icing for a quick indulgence

Dense, velvety chocolate with a heavy texture for a decadent taste Bite size for a quick indulgence

Soft and chewy…just like homemade

A classic cake donut made with buttermilk…moist and rich

Creamy milk chocolate with a soft, chewy center for a satisfying experience

A classic cake donut made with buttermilk…moist and rich

Crisp and crunchy…perfect for dunking

A moist dark chocolate cake donut…for the real chocolate lover

Creamy chocolate with a crunchy, nougat center

Creamy chocolate with a crunchy, nougat center

Category #2 - Ingredients (Primary)Sweetened with natural fructose for a healthy indulgence

Made with canola oil which helps lower blood cholesterol levels

Made with the finest Swiss chocolate

Sweetened with natural fructose for a healthy indulgence

Made with only the freshest ingredients…eggs, milk, butter

For a healthy source of protein…made with unpasteurized egg whites

Made with the finest Belgian chocolate…for the discerning chocolatier

For a healthy source of protein…made with unpasteurized egg whites

Made with unprocessed whole grain flour…keeping all the goodness in

Sweetened with natural fructose for a healthy indulgence

Carob as the main ingredient…a healthy alternative

Carob as the main ingredient…a healthy alternative

2828

The bottom lineBeyond metrics around innovation & into its heart

Innovation requires hard work and some thinking– Homework up front removes some barriers– Setting up an innovation bed of elements, field,

analytics & optimization further helps Showing a path might move people out of

their ‘dogmatic slumbers’

2929

Opportunity #2: Graphics

Now that we have innovation treated as science ..

What about research in a more ‘artistic area’

Say… magazine cover design!!!!

3030

Business, Art, Science The magazine industry

How can we sell more at the ‘stand’?– Better content– Better covers– This isn’t the subscription part of the business

How do we attract advertisers?– Right now it’s a share of wallet issue– Limited money– Now what do we offer them?

3131

Let’s return to the combinatorics ideaBut this time …completely graphical (techno-art?)

Template (structure) Features and elements (components) Combine features by experimental design Test the combinations among consumers Model the results Interpret the data

3232

We saw before ..No problem with partial text concepts

Mind fills in the blanks

5 Element concept

4 Element concept

3 Element concept

2 Element concept

3333

But..what about graphics?Does the mind fill in the blanks?

Full concept 5 Elements

4 Elements Concept

2 Elements Concept

3 Elements Concept

3434

Experimental designABS = absent (incomplete picture)

3535

Let’s try it out

Applying the same ‘scientific’ approach to design

Well … at least the beginnings of design

3636

A1 A2 A3

Category 1 – Magazine Cover Background

Category 2 – Logo

B1 B2 B3

Category 3 – Head/Subhead

C1 C2 C3

3737

Example of setting up a visual stimulus according to a

systematized design

Example from ‘tea’Same approach done with

magazine covers

3838

3939

Let’s try again…

4040

Now the package looks right Replace each visual with encompassing

rectangle Result – categories placeholders layout

(‘template’) The concrete example produces the schematic (not the other way around!!)

Create Template

Category ECategory E

Category CCategory C

Category DCategory D

Category BCategory B

CategoryCategory AA(background)(background)

Category DCategory D

CategoryCategory BB

Category CCategory C

4141

Template is a ‘multilayered cake’

Each layer == category

“Holes” == transparencyEach layer == category

4242

The respondent experience

4343

Who participated

Email invitation sent out from a list provider

523 respondents participated & completed (11% response rate)– Over a period of 36 hours

A sense of who they are comes from the classification pages at the end of the interview

4444

4545

4646

4747

4848

Respondents are consistent Individual data .. good fit of ‘model

0.5 0.6 0.7 0.8 0.9 1.0

Individual R-Square For Model

Poor Good

4949

The two types of information

First response = interest – Number = conditional probability of a respondent

being interested in the visual package if the feature is present

– Constant = baseline interest, without visuals Second response = time

– Number of tenths of seconds required for respondent to process information

– Computer measures response time, allocates time to the components … discovering what holds the eye

5050

A1 A2 A3

Category 1 – Magazine Cover Background

Category 2 – Logo

B1 B2 B3

Category 3 – Head/Subhead

C1 C2 C3

To refresh your memory ..elements

5151

What the different features contribute

-2 C2

-1 C3

2 C1

Head/Subhead

-1 B3

0 B2

2 B1

Logo

-4 A2

3 A3

8 A1

  Background/Cover 

18 Constant

TOT  

5252

How about segments

We looked at naturally occurring groups of consumers– Segmentation based on pattern of hot buttons (response

to element features) Three segments

– Seg 1 (43%) = elaborate old houses– Seg 2 (23%) = clean, blue, New England colonial– Seg 3 (34%) = hates New England type colonial

Our research shows visual segmentation often not as clear nor compelling as text-base segments– Whole new area of insights

5353

What the different features contribute

1-4-2-2 C2

2-6-2-1 C3

3112 C1

-25-3-1 B3

13-30 B2

38-22 B1

Logo

-1880-4 A2

0-7103 A3

33158 A1

     Background/Cover 

22181318 Constant

S3S2S1TOT  

Head/Subhead

5454

What about pair-wise interactions? How do we find interactions..if we don’t know

where they are?– Synergisms, suppressions– Virtually all conjoint methods have to ‘build in’ the

interactions to find them What about graphics interactions and the ‘art’

of design … can we discover them?– We’ll use the same empirical discover approach

that we did with cookies

5555

Interactions… Discovery system**

Use main effects design– With ‘zero conditions’

Permute design – Each respondent has different combinations

Put all data together from all respondents– Across all data you can estimate both linear and

significant pairwise interactions terms– Discover & estimate magnitude

**Patent pending**Patent pending

5656

Three permuted designs - example

C2C4C3

C3C2C2

C4C1C1

Category 3 – Head/Subhead

B1B2B4

B2B4B3

B3B3B2

Caegory 2 - Logo

A2A1A3House3

A3A2A2House2

A1A3A1House1

Category 1 – Background/Cover

 Design 3Design2Design 1

5757

  Linear Interaction

A1B3 ? -3

A2C2 ? 2

A3C2 ? -1

B1C2 ? -3

Only four significant interactions emergedNot big ones … and mostly negativeIn other studies we discover interactions > +15 or < -15

5858

In larger scale studies with hundreds of interactions .. Only 3% – 5% are significant

Strictly empirical ..which synergize, which suppress

SuppressionSynergismNo Interaction

5959

Response time as a new parameter

Question….. How much time is spent looking at a specific feature of the cover?– Approach … measure response time in 10ths

of seconds– Allocate response time to each element– Additive constant = ‘dead time’ that cannot be

allocated Use .. Engineer attention to the cover!!

– Whole new area for consumer research

606030 C3

11 C2

20 C1

-6 B3

-3 B2

3 B1

19 A3

25 A2

17 A1

19 CONSTANT

Response Time (10ths Seconds)

6161

What’s the bottom line?

6262

Systematics generate five knowledge benefits

1. Better data:

– why settle for guessing or focus groups when you get solid quantitative answers?

– better respondents experience

2. Clearer results: looking at the results gives you immediate insight and direction

3. Multi-media: whether concepts, packages … you get to test many stimuli

6363

Five key knowledge benefits (cont.)

4. Segmentation: you get to see new segments, and what turns them on

5. Synergies and suppressions: identify what works together, what doesn’t

6464

Systematics generate three process benefits

1. Democratizes the insights business: You don’t have to have years of experience to get clear direction from the data

2. Faster: Overnight vs. a few days (weeks)

3. Cheaper: About the price of a focus group fully analyzed (with segmentation)

6565

Contact Info:

Moskowitz Jacobs inc.White Plains, NY

Bert KriegerDanny MoskowitzDr. Howard R. [email protected](914) 421-7400

MJI’s website: www.mji-designlab.comIdeaMap.Net internet tool: www.ideamap.netIdeaMap.Net open-source innovation: www.innovaidonline.net