The power of conjoint analysis

Keys to evaluating choice are: • The product/service/proposition attributes • Levels within these attributes In the aforementioned example there are four key attributes, namely: • Price

The Power of Conjoint Analysis

Insight into preference, choice and trade-off

Download free books at

David Murray

The Power of Conjoint Analysis

Insight into preference, choice and trade-off

The Power of Conjoint Analysis : Insight into preference, choice and trade-off

© 2012 David Murray & bookboon.com

ISBN 978-87-403-0214-1

Contents

Public Sector Case Study: Exercising Choice for Specialist Healthcare Treatment 35

Download free eBooks at bookboon.com

Click on the ad to read more

360°

Discover the truth at www.deloitte.ca/careers

© Deloitte & Touche LLP and affiliated entities.

360°

Discover the truth at www.deloitte.ca/careers

© Deloitte & Touche LLP and affiliated entities.

360°

thinking

Discover the truth at www.deloitte.ca/careers

© Deloitte & Touche LLP and affiliated entities.

360°

Discover the truth at www.deloitte.ca/careers

1 Introduction

The Role of Statistics in Market Research

Wikipedia defines statistics as ‘The study of the collection, organisation, analysis and interpretation of data’ Wordwebonline goes on to add the fact that data can be represented in numerical format, which will inform the basis of this particular insight

Statistics is a branch of applied mathematics concerned with the collection, interpretation and to an extent, the manipulation of quantitative data These data can be compiled from a number of sources:

• Secondary data – these are data that already exist in hard or soft documents where the analyst has the appropriate authority to extract and interpret Examples of these documents, freely available in the public domain, are Census Statistics, Regional Trends, Social Trends, Euromonitors and data available upon subscription from Mintel, the Target Group Index or National Shoppers’ Survey

• Primary data – these are data necessitating fieldwork to derive meaningful information in the form of Market Research Surveys where data can be compiled through observation or communication using:

- Self-completion questionnaires

- Face-to-face interviews door-to-door or on the street

- Telephone interviews

- Online questionnaires completed upon an e-mail request or by an invite on a website

The completed questionnaires will contain response codes to enable the analyst to input or data capture responses in numerical format For example a response to a dichotomous will create three codes:

Fig 1 Dichotomous coding frame

or a response to a semantic scale will create either three codes:

Fig 2 Three scale semantic coding frame

Fig 3 Five scale semantic coding frame

Qualitative responses can also be formatted into a coding structure to permit quantitative data analysis, for example in a customer satisfaction survey, the question asks:

“What one thing would improve the service Company X provides its customers?”

A randomly selected ten questionnaires from the hundred carried out reported the following answers:

• “Answer the telephone within five rings”

• “Staff to be polite and courteous”

• “Quicker response to my telephone call”

• “Replenish stock levels”

• “Shelves run out of stock very quickly”

• “More staff to be available to answer queries”

• “When I phone up, not to be passed from one staff member to another”

• “More accurate invoices”

• “Phone me back quicker with an answer”

• “Invoice not clear”

Within these open-ended responses there are some key words:

If codes were to be affixed to these key words or statements, we can now quantify open-ended responses

to this very important question and analyse the data:

Fig 4 Hard coding open-ended analysis

Please note one respondent referred to two service aspects, (telephone and staff available)

These codes have been set up based on a random selection of open-ended responses from every nth, or

in this case, 10th questionnaire These codes together with their interpretation, or coding frame, can now

be used to code all one hundred responses with a default ‘others’ for any responses which do not apply

to the descriptions within the coding frame

You will have noticed a simple analysis of responses in this table - this is what is known as Univariate Analysis Univariate analysis is carried out using a single variable, in this case percentage response to

the question, “What one thing would improve the service Company X provides its customers?”

The analysis depicted is a frequency analysis of the distribution of responses So we can conclude that 40%, or 4 in every 10 responses made a comment pertaining to the company’s telephone service The standard output of Univariate Analysis is a frequency distribution table and for a more visual impact,

Fig 5 Integrating hard coded open-ended questions into the coding frame

(Some respondents will have given more than one answer).

Let us now have a look at the gender distribution of these responses:

Fig 6 Gender distribution of hard coded responses

We are now in a position to add more detail and insight into the analysis by carrying out Bivariate Analysis Bivariate Analysis determines the relationship between these two variables, improvement and gender Bivariate Analysis can be helpful in testing simple hypotheses of association, for example:

• Males are more likely to query invoices

• Females are more likely to complain about items being out of stock

Let us now look at testing these hypotheses:

• One can see that males are more likely than females to focus on invoice accuracy

Furthermore, let us assume that the overall sample of one hundred respondents was split fifty males and fifty females We can now add value to this analysis and state that of all

males who took part in the survey, 18% or almost 1 in 5, are more likely to query an invoice whereas only 4% or 1 in 25 females were likely to query invoice accuracy

Therefore males are 4½ times more likely to query an invoice The hypothesis ‘Males are more likely to query invoices’ has been tested and proved accurate

Fig 8 Comparative profile and penetration of hard coded responses

Click on the ad to read more

Increase your impact with MSM Executive Education

For more information, visit www.msm.nl or contact us at +31 43 38 70 808

or via admissions@msm.nl

the globally networked management school

For more information, visit www.msm.nl or contact us at +31 43 38 70 808 or via admissions@msm.nl

For almost 60 years Maastricht School of Management has been enhancing the management capacity

of professionals and organizations around the world through state-of-the-art management education.

Our broad range of Open Enrollment Executive Programs offers you a unique interactive, stimulating and multicultural learning experience.

Be prepared for tomorrow’s management challenges and apply today

Executive Education-170x115-B2.indd 1 18-08-11 15:13

By applying the same procedure as in the previous analyses, three in every four complainants about of-stock items are female Furthermore, four in every ten females voiced their opinion about this problem compared with only one in eight males In terms of the hypotheses, we can now accept the hypothesis that females are more likely to complain about items being out-of-stock than males

out-Let us now assume that the questionnaire contained five point semantic scales regarding their satisfaction with the company’s service The questionnaire asks the respondents to either give a score of between 1 and 5, with 5 being very satisfied and 1 being not at all satisfied, or just a tick box under the appropriate satisfaction heading for each of these service aspects:

Aspect of service provided Very Satisfied Satisfied Neither nor Dissatisfied Very Dissatisfied

5 When I ring the telephone is answered quickly 1 2 3 4 5

service provided

Fig 9 Five scale very satisfied to very dissatisfied semantic scale pertaining to key business drivers

Based on the answers given by the one hundred respondents, we can now correlate the overall satisfaction,

(all things considered, I am happy with the service), scores with each of the seven aspects of the service

delivery Correlation analysis is the statistical relationship, in this instance, between two variables, the dependent variable (all things considered), and each of the independent variables (e.g opening hours).Output is expressed as the ‘coefficient of correlation’ or the strength of the association between the dependent and independent variables The coefficient is always 1.0 or less with 1.0 equating to perfect correlation The closer the coefficient is to 1.0, the better the correlation

Should three independent variables result in similar coefficients of correlation of 0.7 or greater, then we need to test variations of these or combinations of these three variables using a multivariate statistical technique called Regression Analysis

Multivariate Analysis relates to the relationship between a number of variables, usually three or more, and can be applied for not just explaining current the current relationship between variables but can also be used for modelling purposes to predict possible outcomes or ‘What if’ scenarios by measuring the impact on a dependent variable by changing the values of independent variables

Here is a list of multivariate analysis techniques to explain multidimensional data relationships:

• Regression Analysis – the focus is on the relationship between a dependent variable, e.g overall satisfaction, and one or more independent variables e.g polite staff, telephone

answered quickly and market leaders Having established the coefficient of correlation at a level of say 0.67, where only two-thirds of the relationship is explained, one can vary any

of the independent variables to measure the impact on the coefficient of correlation so

as to increase the outcome from 0.67 to 0.8 or beyond There are a number of regression techniques available, but coverage of these is beyond the scope of this book

Click on the ad to read more

GOT-THE-ENERGY-TO-LEAD.COM

We believe that energy suppliers should be renewable, too We are therefore looking for enthusiastic

new colleagues with plenty of ideas who want to join RWE in changing the world Visit us online to find

out what we are offering and how we are working together to ensure the energy of the future.

• Factor Analysis – this is a method of data reduction by grouping together those variables which are closely correlated This grouping together of closely associated variables forms new variables called factors or principle components For example, in a survey on ‘Motorists’ Opinions’, the following scale variables from a range of fifty original variables, made up Factor 1, (Moy 2001)

- Good petrol consumption

- Low maintenance costs

- Keeps its value

- Good value for money

We might interpret and label factor 1 as the ‘Economy’ factor

• Cluster Analysis – assigns a set of objects into groups called clusters, so that the objects in the same cluster are more similar to each other than those in other clusters

• Correspondence Analysis – provides a means of displaying or summarising a set of data in two-dimensional graphic form

• Discriminate Analysis – identifying exploitable subgroups of consumers to design products with benefits The essence of this is to match goods and services to consumer requirements The method of maximising identification and discrimination of key variables with a data set

is Automatic Interaction Detector, a sub technique of Cluster Analysis

• Conjoint Analysis – or trade-off analysis is a collection of standard statistical techniques that provides objective insights into consumer preferences This multivariate statistical analysis technique and its related models now form the basis of this book

2 An Outline of Conjoint Analysis

It has been said that the term ‘Conjoint’ was derived from two words: ‘considered jointly’ I cannot confirm that this is true but it does illustrate the fundamental idea behind Conjoint Analysis

Since its introduction in 1971, conjoint analysis has been one of the fastest growing marketing research techniques, particularly in the USA A large number of blue-chip companies have adopted the technique, namely The Boston Consulting Group, Hewlet Packard, Levi Strauss, McKinsey and Company, Proctor and Gamble, Smith, Kline and French, Xerox and South-Western Bell This appears to be due to the actionability of its results and the predictive power it has displayed (Wilderstrom, 1994)

In conjoint analysis, consumers are asked to react to a number of hypothetical concepts or service descriptions When consumers are offered a wide range of choices, they would ideally like to have everything at the easiest level of access, but will inevitably have to ask themselves:

“What am I willing to give up?” As we can’t have everything, trade-offs are inevitable

Let’s look at a simple scenario – booking a week’s holiday in Spain Upon evaluating a holiday brochure, the customer notices holidays are available in February, Easter, May and August There are four different types of accommodation – camping in a tent, an apartment, hotel or a villa Amenities available include en-suite bathroom facilities, located near to the beach, located near to a golf course or located near to a bullring The prices listed in the brochure are holidays from £200 - £1,500 with two intermediary price points of £500 and £1,000 Ideally, as human nature prevails, we would love to select a villa by the beach

in August for £200, but that offer will never arise So, depending on one’s personal budget, the time of year one can go on holiday subject to work constraints and personal holiday pursuits, one has to make trade-offs:

“OK, I can’t afford a villa, but maybe an apartment for £500 but at least it’s near the beach and August

Keys to evaluating choice are:

• The product/service/proposition attributes

• Levels within these attributes

In the aforementioned example there are four key attributes, namely:

• Price per week

• Time of year

• Accommodation type

• Amenities

Within each of these attributes there are four levels each namely:

Price per week

• En-suite bathroom facilities

• Located near to the beach

• Located near to a golf course

• Located near to the bullring

Within this holiday offer or proposition, there are 4x4x4x4=256 permutations, clearly too much for

a consumer to evaluate To simplify choice procedures, a subset of these, called a fractional factorial design (Kuhfeld 1994), is presented to the consumer The selection of permutations is randomised and the application software (SPSS from IBM) extracts the selection automatically In this instance, the number of permutations randomly selected, are 16 i.e (4+4+4+4) This is considered sufficient to model the remaining 240 permutations under the assumption that the sample of respondents interviewed, is statistically robust

Sample Size Considerations:

We would recommend a sample size of at least 385 respondents to derive statistically valid conclusions The reason for this is that, at this particular level, the margin of sampling error starts

to decrease at a diminishing rate The statistically validity based on a sample size of say 200, may

be questionable because the margin of sampling error is too high to measure statistically significant differences between two data observations For example, based on a 50%/50% response to a Yes/No question, the margin of sample error for 200 respondents would be ±7%, meaning that the amount

of dispersion around the 50% proportion, results in a range of 43% - 57% answering either ‘Yes’ or

‘No’ to the question As can be seen from this graph overleaf and data table, there is a level where the margin of sampling error starts to decrease at a diminishing rate We are again assuming a 50%/50% response:

Click on the ad to read more

With us you can

shape the future

Every single day

For more information go to:

www.eon-career.com

Your energy shapes the future.

Fig 10 Margin of sampling errors for given sample sizes

The margin of sampling error is derived from the formula:

ඥሺ’ሺ’ െ ͳሻሻȀ

߲݊ ൌ [

Where ߲݊ ൌ margin of sampling error

P = proportion expressed as a decimal

This means that as the sample size increases over 385 correspondents, the reduction in the margin of sampling error starts to plateau for every incremental increase So, we hereby recommend a sample of at least 385 respondents Obviously the greater the sample size, the greater the statistical validity, but there

is the consideration of increased fieldwork costs to offset against this increased validity

Let’s now return to our example of holidaymakers’ evaluating a range of options When formulating a questionnaire to accommodate this evaluation, one must ensure that the travel company can offer every single level, because if the results indicate preference for purchasing a holiday package comprising an element the company cannot offer, the exercise becomes purely academic

SPSS has now made the randomised selection of levels within attributes to enable the modelling of all possible outcomes, four examples of which could be:

Example 1:

Time of Year: AugustAccommodation: HotelAmenities: Located near a bullring

Example 2:

Time of Year: MayAccommodation: VillaAmenities: Located near a golf-course

Example 3:

Time of Year: FebruaryAccommodation: ApartmentAmenities: Bathroom facilities

Fig 12 Randomised selection of levels within attributes

All sixteen options are then printed onto laminated A4 cards for respondents to evaluate

The questionnaire must be very explicit and easy to interpret in its outline of what the respondent should

do with the cards An ideal script could be:

“I am now going to pass you some cards All cards have four aspects of a holiday offer in Spain printed on

them which relate to:

• Price of the holiday

• The time of year the holiday is offered

Click on the ad to read more

www.job.oticon.dk

It doesn’t matter how many cards are in each pile.”

The respondent has now created his/her pile of holiday offers – one good/attractive pile, one poor/unattractive piles

The interviewer now hands the respondent the good/attractive pile and asks the respondent to sort this pile into the order of preference with the overall most preferred at the top and the least preferred at the bottom The respondent now ranks from most preferred to least preferred The interviewer now records the card numbers on the grid on the questionnaire with the most preferred in position 1, the second most preferred in position 2 and so on The interviewer must take care and record the card number sequence

in the exact same order that the respondent has sorted the cards At this stage the grid should have the following hypothetical appearance, assuming eight cards have appeared in the good/attractive pile:

Fig 13 Ranking of most preferred options – holiday example

The interviewer continues:

“Now please do the same for the second pile, (poor/unattractive options), with the most preferred at the top and the least preferred at the bottom”

Again the interviewer must ensure the recorded sequence coincides with the order in which the respondent has sorted the cards The grid on the questionnaire should now have the following hypothetical appearance:

Fig 14 Ranking of all options – holiday example

At the quality control stage, pre data entry, the data analyst should ensure there are:

• No card numbers are duplicated

• All cards have been rated

• There are no omissions in the grid

This particular process is not just limited to PAPI (paper assisted) questionnaires They can also be applied to online questionnaires The analytical procedure remains the same; it’s just the method of presenting the options to online respondents that differs One tried and tested method we have used recently is to present each group of options in turn, on a screen, and ask the respondent to give a score

of between 1 and 100, the higher the score the more desirable the group of options is The reason for requesting a score of between 1 and 100 is the likelihood of a respondent giving the same score twice

is relatively remote compared with a score of between 1 and 20 In the event of an identical score being awarded to a previous range of options, the programme will prompt with a message requesting that a different score be awarded

To use our holiday example, we now have scores for each of the 16 options evaluated:

Fig 15 Online ranking of all options – holiday example

The rating of these options shows that there is a high preference for some of these options, (options 1,

2, 4, 7, 11, 14), an indifferent level of preference, (options 3, 10, 15 and low levels, (options 5, 6, 8, 9,

12, 13, 16)

What we can now do is sort these scores from highest to lowest to enable us to have a ranking of the 16 options presented to the respondent as is demonstrated in the table below:

Fig 16 Online ranking of all options sorted from most preferred to least preferred

We now have a ranking which resembled that derived from the PAPI engagement method Both databases can now be integrated to increase the sample size and thereby add value to the analysis

We now have a full dataset of preference rankings for all 16 options to enable us to build a rating based model based on evaluating the interrelationships between each of the 16 service levels using linear regression analysis A full understanding of conjoint requires years of experience in mathematical statistics, computer programming and spreadsheets Full understanding however is not required to conduct and interpret conjoint analysis It is only essential to understand the two key outputs of conjoint analysis:

Click on the ad to read more

• Utility values – these are a measure of quantitative values attached to each level within each attribute Utility values, in essence, are a ‘common currency’ translating attractiveness and desirability of relationships between levels into a positive score or, conversely, the lack of

‘fit’ between other relationships between level sets into a quantitative negative value Utility values are obtained from an ordinary linear regression analysis using the rank data as the dependent variable and the profile designs/cards as the independent variables

• These utility values, analogous to the coefficients of correlation derived from the regression analyses, are called part-worths and can be used to evaluate the relative importance of the attributes to which they belong

The results show which combinations of levels generate the highest levels of attractiveness or desirability and which particular attributes most influence preference and the relative importance

of each attribute

Utility values are initially calculated at respondent levels The regression is repeated for each respondent’s data The data analysis, once completed, can be averaged over all the respondents to show the average utility value for each level within each attribute Please note the importance and utility scores are purely for illustrative purposes – this also applies to the case studies in the next chapter

With these data, we can now map out the importance scores for each attribute and the utility scores for each level

So let us populate our holiday model with some hypothetical importance and utility scores:

Fig 18 Importance scores sorted from most to least preferred

Statistically, in terms of the sample size, price is the most important attribute There is no statistically significant difference between Time of Year and Accommodation Type, so both are equally important, though not as important as price Amenities is the least important attribute So one can conclude from this that Price, Time of Year and Accommodation Type are the three key drivers in the decision making process.SPSS has also computed the individual utility scores for each level within each attribute What we need

to establish is which combinations of levels derive the highest utility scores to enable the executives of the holiday company to formulate their offer(s) to the marketplace

Remember, earlier in this chapter, we stated that the four levels within these four attributes yields 256 permutations – (4x4x4x4) We are now in a position to simulate the highest possible levels of utilities based on all the 256 combinations or permutations of the four levels within the four attributes

We would strongly advise an analyst (or end user of these data) to set up a spreadsheet and map out all

256 permutations This sounds a daunting task but there is a logic behind ensuring all 256 are mutually exclusively covered without any duplications Here is some guidance:

1 In row 1 of a spreasheet, type in your headings – PRICE, TIME, ACCOMMODATION, AMENITY, UTILITYp, UTILITYt, UTILITYacc, UTILITYam

2 Enter the lowest price level £200 into Cell A2 and its utility score 2.17 into Cell E2

3 Copy these Cells down to A65 and E65 respectively

4 In Cell A66 enter £500 with its utility score 4.10 into Cell E66

5 Copy these cells down to A129 and E129 respectively

6 Repeat these processes for £1,000 (utility score 3.48) and for £1,500 (utility score 1.18) by copying and pasting or dragging these values down 64 Cells

7 Enter ‘February’ in cell B2 and its utility score -3.04 into cell F2

8 Drag or copy these down 15 cells in turn into cells B17 and F17 respectively

9 Enter ‘Easter’ into cell B18 and its utility score 0.64 into cell F18

10 Drag or copy these down 15 cells into cells B38 and F38 respectively

11 Type ‘May’ into cell B34 and its utility score 2.13 into cell F34

12 Drag or copy these down 15 cells into cells B49 and F49 respectively

13 Type ‘August’ into cell B50 and its utility score 1.8 into cell F50

14 Drag or copy these down 15 Cells into cells B65 and F65 respectively

Click on the ad to read more

It all starts at Boot Camp It’s 48 hours

that will stimulate your mind and

enhance your career prospects You’ll

spend time with other students, top

Accenture Consultants and special

guests An inspirational two days

packed with intellectual challenges and activities designed to let you discover what it really means to be a high performer in business We can’t tell you everything about Boot Camp, but expect a fast-paced, exhilarating

and intense learning experience

It could be your toughest test yet, which is exactly what will make it your biggest opportunity.

Find out more and apply online.

Choose Accenture for a career where the variety of opportunities and challenges allows you to make a difference every day A place where you can develop your potential and grow professionally, working

alongside talented colleagues The only place where you can learn from our unrivalled experience, while helping our global clients achieve high performance If this is your idea of a typical working day, then

Accenture is the place to be

Turning a challenge into a learning curve.

Just another day at the office for a high performer.

Accenture Boot Camp – your toughest test yet

Visit accenture.com/bootcamp

15 Now copy and paste B2:B65 and F2:F65three times commencing with cells B66 and F66 respectively till you have completely filled all 256 rows

16 Type ‘Tent’ into Cell C2 and its utility score -2.46 into Cell G2

17 Drag or copy these down 3 Cells to C5 and G5 respectively

18 Type ‘Apartment’ into cell C6 and its utility score 1.85 into cell G6

19 Drag or copy these down 3 cells into cells C9 and G9 respectively

20 Type ‘Hotel’ into cell C10 and its utility score 3.73 into cell G10

21 Drag or copy these down 3 Cells into cells C13 and G13 respectively

22 Type ‘Villa’ into Cell c14 and its utility score 1.64 into cell G14

23 Drag or copy these down 3 Cells into cells C17 and G17

24 Now copy and paste C2:C17 and G2:G17 into cells C18 and G18 respectively till you have completely filled all 256 rows

25 Type ‘Bathroom’ into cell D2 and its utility score 5.23 into cell H2

26 Type ‘Beach’ into cell D3 and its utility score 4.61 into cell H3

27 Type ‘Golf’ into cell D4 and its utility score -0.87 into cell H4

28 Type ‘Bullring’ into cell D5 and its utility score -2.82 into cell H5

29 Now copy and paste D2:D5 and H2:G5 into cells D6 and H6 respectively till you have completely filled all 256 rows

30 Enter a new column heading in Cell I1 and name it ‘Total Utility’ Now simply add across the utility scores in Cells E2:H2 and drag down this entry till you have reached Cell I257

31 Now sort the spreadsheet from highest to lowest Total Utility

Of the 256 aggregated utility scores, now let us have a look at what the top ten combined utilities and the bottom ten are telling us:

Fig 19 Top ten aggregated utility scores for holiday model

Conclusions:

• Without a doubt, the most preferred accommodation type is a hotel

• En-suite bathroom facilities appear to have a higher perceived value and appeal than their hotel having shared facilities but near a beach

• To ensure a hotel with en-suite bathroom facilities in May, holiday makers may trade off this attractive offer by paying a premium for it, (as per option 3) compared with options 2 and 4

• Easter breaks can be offered but there is the expectation that it will demand a cheaper price and with en-suite bathroom facilities

The main dilemma Marketing Management has is which price point to charge as demand for a hotel with en-suite bathroom facilities in May or August are price inelastic This is where it is very important

to converse with Finance Management to evaluate the margins, particularly if £500 only represents

a breakeven price or low profit margin under the hypothesis that a low price will generate bookings Based on the combined utility score, this travel company can now offer this package at £1,000 without impacting too much on sales

What we can learn from the bottom ten combined utilities?

Click on the ad to read more

Option Description Total Utility

Fig 20 Bottom ten aggregated utility scores for holiday model Conclusions:

• To be located, irrespective of accommodation type, near a bullring is definitely not appealing

• Nor is a tent near a golf course even at £200 in February

• The holiday company should not offer camping accommodation

Despite this being a simple hypothetical example to illustrate how Conjoint Analysis works, it does demonstrate how effectively choice can be quantified and evaluated and how the product offering can

be positioned It also gives a number of strategic options based on the resultant inelasticity of demand between the two price points of £500 and £1,000 The trade-off between doubling one’s sales revenue and a slight drop in uptake (based on combined utilities) becomes a ‘no brainer’

3 Case Studies

We have now seen how a travel company, with the benefit of conjoint analysis, can:

• Position their offers based on the combinations with the highest levels of utility

• Price the propositions accordingly to the extent a premium price can be charged without much adverse impact on sales

• Ascertain what customers would expect for their money

• Promote preferred times of year

• Dispense with those aspects the company was assessing, which resulted in a low or negative utility

This was a hypothetical example with utilities produced to yield good results, whose resultant courses

of action would be easy to comprehend We are now going to consider practical case studies, one from the private sector and one from the public sector, which will not only demonstrate the actionability of this technique, but will also exemplify that outcomes are not as clear cut as the explanatory example in the previous chapter This will illustrate one drawback that the tool may have and that is human nature Human nature states that one would desire the best outcome at the cheapest price This is where trade-offs play a very important role in interpreting the results effectively The analyst must always be looking at the most profitable outcomes for his/her organisation, and ascertain the optimum mix of other attributes which will result in positive utilities being derived at profitable price levels This will become apparent

in the private sector case study

At this stage, one popular myth I would like to discard, is that conjoint analysis is not a tool to enable pricing decisions per se It is just one of many outcomes It is a snapshot of choice not just in markets where price is a key determinant in the purchasing decision, but also in the public sector where price has no role to play The latter case study will demonstrate how this technique Private Sector Case Study: The Bundling of Digital Services: will enable the shaping of, for example health care provision where patients can exercise their choice based on a range of environmental, social and political considerations But first let us consider a case study from the private sector:

Private Sector Case Study: The Bundling of Digital Based Services:

Background to study: In the early 1990s the telecoms market was deregulated in the UK permitting perfect competition in a marketplace where the monopoly powers of British Telecom had ruled for decades The entry, particularly from the USA and Canada, of telephony suppliers with their vision of creating an ‘information super highway,’ led to one of the biggest civil engineering projects in the UK since the Channel Tunnel Regulators would award franchises to the highest bidder permitting them to lay ducting as per the Oftel requirement to provide 95% of households with a service within that franchise