LOGO Research - CONJOINT DCM 20210630

Discovering Consumers' Ideal Product Webinar 30.06.2021 10:30 – 12:00.

Discovering Consumers' Ideal Product Discrete Choice Modelling (DCM) a systematic analysis of decisions.

Research Director 20+ years in market research Stoyan Ninov With a solid background in marketing research, Stoyan is well aware of the tools and techniques to tackle your research issue. His vast experience in resolving different kinds of research problematics for different types of businesses will help you to find the optimal approach for your particular case, whatever it is. Stoyan will propose you the optimal methodology, will lead the LOGO Research team to use the right tools and to apply the proper analyses, delivering you the answers you need in the most effective manner. Email: [email protected] Who? Managing Director 20+ years in market research Vladislav Kolev During the last 20 years, Vladislav has a proven record of successfully supporting his clients on their way to success. A hands-on manager and LOGO Research corporate leader with a strong, critical and highly analytical mind, Vladislav will lead your team through any marketing problematics and will challenge any piece of information to find the answers you need. Vladislav puts no limits to partnership, which makes him a preferred supplier for marketers and companies seeking added value beyond bare data and preferring long-term partnership based on mutual confidence. Email: [email protected].

1. What is DCM 2. Areas of implementation 3. Design 4. How to read the results? 5. Market Simulation 6. “What if” scenarios 7. Useful additions 8. Conclusions 9. More about LOGO Research Table of Content.

What is DCM? Conjoint is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it..

Product / Service Attributes 6 Augmented Layer Brand Salience Digital Life Customer Experience Category Competitors Reputation Integrity Expected Layer Brand Price Package Performance Style Features Quality Core Layer Product / Service VK.

Usual research approach • Analyze preferrence towards brands (SBA/TOM, Aided BA) • Analyzed the better or best concept (testing) • Try to find out the most appropriate product / service communication (pre-test, post-test) • Play pricing games (PSM) • … 7 …by asking questions and evaluate each marketing aspect after another. VK.

Fight the complexity • Products are similar to people • They are complex 8 VK … just try to do it while the engine is working!.

what is Conjoint + DCM? • Discrete Choice Modelling (DCM) is type of Conjoint Analysis, the most preferred model for a conjoint questionnaire by far; * Developed by marketing professor Paul E. Green at the Wharton School of the University of Pennsylvania in 70’s of XX. 9 • “CONJOINT*” comes from the alternative products being defined by several attributes “considered jointly”. • Each of the alternative products in a conjoint analysis is defined by two or more attributes, which serve as variables in the study. • It estimates psychological tradeoffs that consumers make when evaluating several attributes together VK.

what is DCM? • DCM “translates” consumer behavior into empirical quantitative measures - better than any other research tools; 10 LONG-TERM MEMORY EXPLICIT (conscious awareness, declarative) Semantic Memory (Facts and General Knowledge) Episodic Memory (Personal Experience) IMPLICIT (unconscious awareness, nondeclarative) Procedural memory (Motor and cognitive skills) Priming Enhanced identification of objects or words 𝑥 + 𝑎 ! = % "#$ ! 𝑛 𝑘 𝑥"𝑎!%" VK.

what is DCM? • DCM uses interactive, more engaging and realistic data collection approach; 11 VK.

what is DCM? DCM allows for development of market simulation models to predict consumer behavior after product changes; 12 VK.

what is DCM? Traditional rating surveys and analyses do not have the ability to place "importance" or "value" on the different attributes a particular product or service is composed of. Discrete Choice Modeling can do this, which makes it the mostly used technique for modeling – evaluating and predicting based on consumer choices. estimates psychological tradeoffs realistic data collection approach “translates” consumer behavior development of market simulation models VK.

Typical questions to be addressed by DCM 14 VK What feature or attribute of a product is most influential in terms of market success? What are consumers focused on when making their purchase decisions? How attributes influence each other and the price? What is the elasticity of demand by attributes? How sensitive are consumers to price shifting or any other attribute changes? What trade-offs do consumers make? How does the modification of the product affect demand for it? What is the optimal combination of features?.

Areas of implementation.

Areas of implementation research methods 16 Consumer Decision Process Evaluation Product Development Development of strategies for market communication Product Optimization Market segmentation Market simulation Concept testing SN.

DCM conjoint areas of application estimate the chances of success for new products or services base for a potential estimate assistance at definition of target group (Positioning of product) calculation of price elasticities defining and optimizing the price and price- offs simulation of possible market developments development of new product concepts improving of the communication with the target group application areas in Marketing SN.

OTC Pack Optimization (size, label, shape, ingredients, price) Mortgage Loan preferences (period of time, interest, approval period, fees) Fast Food Retail Stores Menu preferences (menu combinations, price) Ideal Member of Parliament Profile (party affiliation, education, professional experience, age) DCM is widely applicable… mobile phone plans preferences (mobile carrier, data allowance, monthly fee) TV set preference (brand, display size, smart TV options, price) Price optimizations (brand, pack type, pack size, price) Car features preferences (engine capacity, type, price) Fitness Centers (range of services, payment terms, time period, monthly fee) Financial Services Consumer Goods Retail Public Consumer Services Consumer Electronics Telecom Automotive Health Care SN.

designing DCM.

DCM can tell you: What combination of attributes / product features will ensure highest preference for my products? What is the optimal price to ensure the highest possible revenue for my propositions? 20 SN.

Attribute / Product Category Shower Gel Health Insurance Software Package Brand Name Nivea MetLife Progress Brand Variant Crème Soft Accident Insurance Telerik UI Weight 500ml Employer Social Benefit Product License Bundle Package Small plastic tube Injuries, medical treatment, hospitalization, accidents Standard Support (Phone + Online Ticketing System) Content Almond Oil and Mild Scent Fractues, Severe Burns, etc. ASP .Net Ajax, Silverlight, Xamarin, UWP, WPF, Winforms Quantity Single package One year 1 year subscription Reliability Without alcohol Guaranteed acceptance policy Leading provider of app dev Marketing Claims Cleanses the skin without drying it out. Claims are generally processed within 10 business days Support, provided directly by the developers Price BGN 6,49 BGN 35 / monthly USD 1,299 per developer Attributes Product features, Marketing Instruments, All relevant components, Price 21 examples SN.

Attribute Levels Possible and analyzed items within a product attribute Attribute / Product Category Shower Gel Health Insurance Software Package Brand Name Nivea MetLife Progress Brand Variant Crème Soft Accident Insurance Telerik UI Weight 500ml Employer Social Benefit Product License Bundle Package Small plastic tube Injuries, medical treatment, hospitalization, accidents Standard Support (Phone + Online Ticketing System) Content Almond Oil and Mild Scenet Fractues, Severe Burns, etc. ASP .Net Ajax, Silverlight, Xamarin, UWP, WPF, Winforms Quantity Single package One year 1 year subscription Reliability Without alcohol Guaranteed acceptance policy Leading provider of app dev Marketing Claims Cleanses the skin without drying it out. Claims are generally processed within 10 business days Support, provided directly by the developers Price BGN 6,49 BGN 35 / monthly USD 1,299 per developer 22 examples 250ml 500ml 700ml 1000ml UNIQA Allianz DZI MetLife Grawe Bulstrad 1 year 2 years SN.

DCM design • The respondents are exposed to a series of choices. 23 SN • For each choice they are asked which product, if any, they are most likely to purchase. • Based on number of attributes and levels we define number of products (concepts) in one interaction (task), number of interactions and the sample size. • The number of products in one page could vary between 3 and 12 • Total Number of interactions is between 5 and 20 • The sample is normally between 200 and 600.

Specify the attributes’ levels Design questionnaire Identify the relevant attributes Agree on scope of project Define number of attributes, levels and sample size. In practice sample size varies from 150 to 1200. Minimum 200 respondents per group / segment are needed. Relevant for the management, expected to influence the preference, preferably not exhibit strong correlation. Full profile conjoint will not exceed 6 attributes! Levels have to be separated enough, realistic and not pushed as a winner or loser. Two levels per attribute is the minimum! More levels means bigger sample! Define which products to include, how many choice sets and alternatives to introduce (normally up to 20 bits, e.g. 3 alternatives and 6 attributes) Design market simulation Future analyses Estimate part-worths Collect data Each respondent receives a unique questionnaire. Randomization reduces the design efficiency but it reduces also the impact of order effects. Evaluate the attributes’ levels. Counting analysis, Logit (logistic regression), Latent Class analysis and HB (Hierarchical Bayesian) Used to define what product to offer to maximize shares given competitive environment, what is relevant price sensitivity and best portfolio (line optimization) Compare actual results with forecasts. Implement distribution data for future analyses. Steps VK.

Setting attributes in details • BRAND attribute with possibilities to load huge level number • Possibilities for Table or Shelf-Facing Display • PRICE attribute in equal price point on possible big price interval and individual brand based specifics • Check Fixed Choice Scenarios: used to measure test/retest reliability, model fit validation, test specific scenarios that are of interest or for warm-up tasks. • Possibilities to define prohibitions (price intervals or price points) on attribute / brand level VK.

How to read the results?.

Which is the most attractive proposition (combination of the product features)? Which features are most impactful (play the key role) for the customers choice? How can we optimize profitability, “playing” with any feature, (including price)? Questions to be answered by DCM VK.

UTILITY A utility is a measure of relative desirability or worth. The higher the utility, the more desirable the attribute level. Levels that have high utilities have a large positive impact on influencing respondents to choose products. Utility ZERO-CENTERED DIFFERENCES This normalization transforms the raw utilities for each respondent to a scale wherein the average difference between best and worst levels across attributes is equal to 100. Zero-Centered Differences REVENUES AND PROFITS Total Revenues are calculated as Share x Market Size in Units x Price. Profits are calculated by subtracting Total Cost from Total Revenues, where Total Cost is equal to Share x Market_Size_in_Units x Cost_per_Unit. Revenues and Profits SHARES OF PREFERENCE When two or more products are specified in the market simulator, we can estimate what percent of the respondents would prefer each product. Shares of preference represent the predicted shares for the products in the model given equal awareness and equal distribution. As awareness and distribution are not equal in the real world, and as other effects, such as inertia, may exist, conjoint results do not reflect actual market shares. Shares of Preference RELATIVE IMPORTANCE How much difference each attribute could make in the total utility of a product. That difference is the range in the attribute's utility values. We percentage those ranges, obtaining a set of attribute importance values that add to 100. Importances are ratio data. Relative Importance DCM terms VK.

actionable results VK It calculates a proportion of "wins" for each level, based on how many times a concept including that level is chosen, divided by the number of times a concept including that level appeared in the choice task. In addition it reports a Chi Square statistic for each main effect and joint effect indicating whether the proportions in that table differ significantly from one another. It is interpret as a measure of "Importance" for an attribute or assuming that the main-effect Chi Square test that is not significant indicates that the attribute had little impact on choice. Calculation of the choice data, main effects (importance of attributes) and joint effects (relative importance)..

type of analyses Counting analysis is a quick way to summarize the results of choice data. Calculation of the main effects and joint effects. It calculates a proportion of "wins" for each level, based on how many times a concept including that level is chosen, divided by the number of times a concept including that level appeared in the choice task. In addition it reports a Chi Square statistic for each main effect and joint effect indicating whether the proportions in that table differ significantly from one another. It is interpret as a measure of "Importance" for an attribute or assuming that the main-effect Chi Square test that is not significant indicates that the attribute had little impact on choice; Logit: The analysis is used to calculate the utilities. A utility is a measure of relative desirability or worth; Latent Class: Latent Class is a utility estimation method and assigns respondents into segments having similar preferences based on their choices in the questionnaire; HB (Hierarchical Bayes) - estimates reasonable individual part worths even with relatively little data from each respondent. VK.

Product Attributes: Brand, Pack Size, Price Example 31 Powered by: & Conjoint analytics SN • Brand: 6 levels • Pack Size: 3 levels (small: 250-255ml, middle: 400-500ml, big: 750ml) • Price: 3 Levels (low, middle, high).

Attributes & Levels 32 Low Tier Mid Tier High Tier Dove 250ml 4.49 5.49 5.99 Dove 500ml 7.99 8.99 9.49 Dove 750ml 8.99 9.99 10.99 Fa 250ml 2.99 3.69 4.49 Fa 400ml 4.49 4.99 5.99 Fa 750ml 5.99 6.99 7.99 Nivea 250ml 3.99 4.49 5.49 Nivea 500ml 5.99 6.99 7.99 Nivea 750ml 6.49 7.49 9.99 Cottage 250ml 3.79 4.79 5.79 Cottage 500ml - - - Cottage 750ml 6.99 7.99 8.99 Palmolive 250ml 3.49 4.19 4.99 Palmolive 500ml 4.99 5.99 6.99 Palmolive 750ml - - - All Nature 255ml 3.99 4.49 5.99 All Nature 500ml - - - All Nature 750ml - - - SN.

Design 6 product, 14 interactions 33 SN.

Importance of Attributes (%) Brand 58% Size 28% Price 14% 34 SN.

Utilities 35 -50,0 0,0 50,0 Dove Fa Nivea Cottage Palmolive All Nature 250/255 ml 400/500 ml 750 ml Low Medium High Utility Lower 95% CI Upper 95% CI Brands Pack Size Price VK.

64,5% 25,0% 6,0% 4,0% 45,8% 18,7% 20,0% 5,0% 3,2% 2,8% 4,0% 57,1% 30,3% 7,9% 4,7% 33,7% 23,5% 22,3% 7,9% 4,9% 3,0% 4,7% Brand A Brand B Brand C Brand D Brand A.1. Brand A.2. Brand B.1. Brand B.2. Brand C.1. Brand C.2. Brand D.1. market share rel preferance Category X Modern Trade Market Share vs Relative Preference Simulation on market prices 36 -7,4% 5,3% 1,9% 0,7% -12,2% 4,8% 2,3% 2,9% 1,7% 0,1% 0,7% preference - share VK.

Share of Preference 37 8,2% 20,4% 10,2% 24,3% 19,6% 10,1% 7,3% None Shares (%) ACTUAL PRICES (mid tier) 8,5% 20,7% 10,4% 25,1% 20,2% 10,5% 4,5% None Shares (%) PROMO PRICES (low tier) 7,8% 20,2% 9,8% 23,5% 18,0% 9,7% 10,9% None Shares (%) PROFIT PRICES (high tier) VK.

38 Base: All Respondents for the respective analysis BRAND x PRICE Sensitivity Analysis (Price Elasticity) Brand A Brand B Brand C Brand D Brand E Brand F Brand G Brand H 80.4% 28.4% 81.1% 39.7% 70.2% 19.4% 41.8% 43.7% Relative Preference at Current Price (4th in a row) * (expected rel. preference at lowest price - expected rel. Preference at highest price) / rel. preference at current price Expected Relative Preference in lower/higher price with current price of all other brands Relative Preference (%) Sensitivity rate*: 0.89 0.95 0.99 1.05 1.09 1.15 1.19 1.301.30 1.30 1.30 1.35 1.39 1.45 0.85 0.89 0.95 0.99 1.05 1.09 1.15 0.95 0.99 1.05 1.09 1.15 1.19 1.25 1.09 1.15 1.19 1.25 1.29 1.35 1.39 1.35 1.39 1.45 1.50 1.55 1.60 1.65 0.89 0.95 0.99 1.05 1.08 1.10 1.15 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% VK.

Market Simulation.

interactive reporting “оn call”: the client simulator • market simulations • “what if” market scenarios: à “play” with any attribute(s) à maximize your market share, your revenue or your profitability à respond to competitive actions SN.

Market Simulation Assumptions • We have interviewed the right people • Each person is in the market to buy • We’ve used a proper measurement technique • Respondents have answered reliably and truthfully • All attributes that affect buyer choices in the real world have been accounted for • Equal availability (distribution) • Long-range equilibrium (equal time on market) • Equal effectiveness of sales force • No out-of-stock conditions SN.

Simulation Scenarios Category X Modern Trade Maximize Share of Preference Maximize Revenue Channel Brand SKU Size Flavor Price Modern Trade Brand A Pack 1 630 Apple 1.09 1.15 1.29 1.39 1.59 Pack 1 630 Orange 1.09 1.15 1.29 1.39 1.59 Pack 1 630 Orange 1.09 1.15 1.29 1.39 1.59 Pack 1 630 Pear 1.09 1.15 1.29 1.39 1.59 Pack 2 700 Apple 1.39 1.45 1.59 1.69 1.75 Pack 2 700 Orange 1.39 1.45 1.59 1.69 1.75 SoP Effect 6.4% Rev Effect -11.2% Channel Brand SKU Size Flavor Price Modern Trade Brand A Pack 1 630 Apple 1.09 1.15 1.29 1.39 1.59 Pack 1 630 Orange 1.09 1.15 1.29 1.39 1.59 Pack 1 630 Orange 1.09 1.15 1.29 1.39 1.59 Pack 1 630 Pear 1.09 1.15 1.29 1.39 1.59 Pack 2 700 Apple 1.39 1.45 1.59 1.69 1.75 Pack 2 700 Orange 1.39 1.45 1.59 1.69 1.75 SoP Effect -5.0% Rev Effect 9.4% 24,6% -31,9% -0,6% -10,3% -10,0% -0,7% -9,7% 5,3% -28,9% -0,6% -10,4% -10,0% -9,8% Brand A 630 Pack 1 Brand A 700 Pack 2 Brand B 700 Pack 1 Brand B 630 Pack 2 Brand C 630 Pack 1 Brand C 500 Pack 2 Brand D 375 Pack 1 SoP Rev -5,4% 1,6% 0,9% 1,1% 1,0% 0,2% 0,5% 5,0% 22,7% 4,2% 15,0% 22,6% 10,6% Brand A 630 Pack 1 Brand A 700 Pack 2 Brand B 700 Pack 1 Brand B 630 Pack 2 Brand C 630 Pack 1 Brand C 500 Pack 2 Brand D 375 Pack 1 SoP Rev Percentage Difference Percentage Difference SN.

Predicting Revenue from Conjoint Results • Estimate actual awareness / distribution to be achieved. • Track results over time and compare predicted to actual. • Model a historical action and compare to the model results. • Cut any increase in share predicted by the model in half. SN.

“What if” Scenarios.

Market Modelling – My Scenario Market Size – 1mln units 45 8,2% 20,4% 10,2% 24,3% 19,6% 10,1% 7,3% None Shares (%) ACTUAL PRICES (mid tier) 7,8% 20,1% 9,2% 22,9% 18,9% 14,3% 6,8% 0 0,05 0,1 0,15 0,2 0,25 0,3 None Shares (%) My Scenario PRICES Promo Others @ actual prices Share Implication: +4.2% Revenue Implication: +17.2% +4.2% Palmolive 250ml 3.49 4.19 Palmolive 500ml 4.99 5.99 VK.

Conclusions.

! DCM conjoint advantages 47 •realistic •no difficult (irrelevant) questions •hard to manipulate •easy.

pros and cons disadvantages • designing conjoint studies can be complex • difficult to use for product positioning research because there is no procedure for converting perceptions about actual features to perceptions about a reduced set of underlying features • respondents are unable to articulate attitudes toward unknown/new categories • poorly designed studies may over-value emotional/preference variables and undervalue concrete variables • does not take into account the weight of purchase so it can give a not so exact reading of market share advantages • estimates psychological tradeoffs that consumers make when evaluating several attributes together • measures preferences at the individual level • uncovers real or hidden drivers which may not be apparent to the respondent themselves • realistic choice or shopping task • able to use physical objects • if appropriately designed, the ability to model interactions between attributes can be used to develop needs based segmentation VK SN.

аddress your challenge & turn it into benefit! [email protected] 3, Yanko Sakazov Blvd. Floor 1 1527 Sofia | Bulgaria www.logo-mrc.com.