Conjoint Analysis

pricing

product-design

Conjoint Analysis measures how consumers trade off product attributes — features, price, brand — to understand preference, predict demand, and simulate market scenarios

Author

TVE Analytics Team

Published

March 27, 2026

Overview

Conjoint Analysis is a survey-based method that measures how consumers make trade-offs between product attributes. Instead of asking what matters most in isolation, conjoint presents realistic product configurations and forces respondents to choose — revealing what they actually value when trade-offs are involved.

The core idea: products are bundles of attributes, and consumers evaluate those bundles holistically. Conjoint decomposes that holistic evaluation into the contribution of each attribute and level, producing utility scores that quantify how much each feature, price point, or brand contributes to preference.

TVE primarily uses Choice-Based Conjoint (CBC), the industry standard for most commercial applications. In CBC, respondents see a series of choice tasks — each presenting 3–5 product concepts defined by different combinations of attributes — and select which one they would choose (or “none of these”).

Note

How does conjoint relate to VWD / Gabor-Granger? Van Westendorp and Gabor-Granger evaluate price in isolation — they work well when the product concept is fixed and the only question is “what should we charge?” Conjoint treats price as one attribute among many, allowing you to understand pricing in the context of feature trade-offs, brand effects, and competitive dynamics. Use conjoint when the pricing question cannot be separated from the product design question.

Types of Conjoint

Type	Description	When to Use
Choice-Based Conjoint (CBC)	Respondents choose their preferred option from a set of product concepts. The standard approach for most commercial applications.	Most pricing, product design, and competitive simulation questions
Adaptive Choice-Based Conjoint (ACBC)	An adaptive version that narrows in on each respondent’s preferences through a BYO (build-your-own) stage, screening, and choice tasks.	Complex products with many attributes where standard CBC would require too many tasks
Menu-Based Conjoint	Respondents build a product by selecting from menus of optional features at stated prices — modelling add-on and bundling decisions.	Configurable products, subscription tiers, optional extras

TVE selects the appropriate conjoint variant based on the number of attributes, the product structure, and how the results will be used.

How Choice-Based Conjoint Works

Define attributes and levels: The analytics team works with the project team to define the product attributes (e.g., brand, features, price) and the levels within each attribute (e.g., Brand A / Brand B / Brand C; £499 / £599 / £699).
Generate choice tasks: An experimental design creates balanced, efficient combinations of attribute levels. Each choice task presents 3–5 product concepts plus a “none of these” option.
Respondents complete choice tasks: Each respondent sees 10–15 choice tasks and selects their preferred option (or “none”) in each. The trade-offs they make reveal their underlying preferences.
Estimate utilities: Hierarchical Bayes (HB) estimation produces individual-level utility scores for every attribute level — quantifying how much each feature, price point, or brand contributes to preference.
Simulate market scenarios: The utility scores feed into a market simulator that can test “what if” scenarios — e.g., “What happens to our share if we add this feature and increase price by £50?”

Key Assumptions

For conjoint to produce valid results:

Respondents can evaluate realistic product concepts: The attributes and levels must reflect meaningful purchase decisions the respondent can engage with.
Attributes are independent: Each attribute should be evaluable independently of other attributes. Avoid attributes that logically depend on each other.
Levels within each attribute are mutually exclusive: A product can only have one level per attribute (e.g., one price, one brand).
The “none” option is realistic: Respondents should genuinely be able to walk away — the category is not a forced purchase.
The number of attributes is manageable: Too many attributes (typically more than 7–8) make tasks cognitively overwhelming. Use ACBC or attribute reduction strategies for complex products.

Business Questions This Answers

Conjoint is ideal when clients ask: “How do consumers trade off different product attributes — and what configuration maximizes appeal, share, or revenue?”

Conjoint is not limited to pricing. It can be used for any decision where consumers evaluate bundles of attributes — including product concept design (e.g., what combination of game elements makes the most appealing game?), service configuration, and competitive positioning.

It helps answer strategic questions such as:

Product design: Which combination of features maximizes preference? What is the optimal product configuration?
Pricing with trade-offs: How much more would consumers pay for a specific feature? What is the price premium for our brand vs. competitors?
Competitive simulation: What happens to our share if a competitor launches a new product or drops their price?
Portfolio optimization: Which products should we offer to maximize total portfolio share without excessive cannibalization?
Feature valuation: What is the incremental value of adding a specific feature to the product?
Willingness to pay: What is the maximum price consumers will accept for a given product configuration?

It is particularly valuable when:

Price cannot be evaluated independently of features or brand
You need to simulate competitive market dynamics
The product has multiple configurable attributes
Stakeholders need a market simulator tool for ongoing decision-making

When NOT to Use

Conjoint may not be a good fit when:

The product concept is fixed and only price varies: If the product is defined and you only need to find the right price, use VWD / Gabor-Granger — it is faster, cheaper, and more focused.
You only need a ranking of individual items: If the question is “what matters most?” without trade-offs against price or brand, use MaxDiff instead.
The product has too many attributes: More than 7–8 attributes in standard CBC becomes cognitively burdensome. Consider ACBC, attribute reduction via PCA, or splitting into separate exercises.
Respondents lack category familiarity: If the product is radically new with no reference framework, respondents cannot make meaningful trade-offs.
Sample size is too small: Conjoint requires larger samples than simpler methods — at least 300 respondents for reliable estimates.
Budget or timeline is very constrained: Conjoint requires more design, programming, and analysis time than simpler pricing methods.

Data Requirements

Data Type:
Survey data from a conjoint exercise. Respondents complete 10–15 choice tasks, each presenting 3–5 product concepts with varying attribute levels.
Attributes and Levels:
Typically 4–7 attributes with 2–5 levels each. More attributes require more choice tasks and larger samples.
Sample Size:
Minimum 300 respondents for standard CBC. 500+ preferred for stable individual-level utilities and subgroup analysis. Each subgroup needs at least 200 respondents.
Experimental Design:
The analytics team generates a statistically efficient experimental design that ensures all attribute levels appear with balanced frequency and minimal confounding.
Product Stimulus:
Clear, concise attribute descriptions that respondents can evaluate quickly. Avoid long text descriptions — use labels, icons, or short phrases.

Time Allocation

Stage	Hours
Kick-off & attribute design	4
Experimental design & programming	4
Analysis - overall	12
Analysis - per subgroup	3
Market simulator build	4
Total (overall + 2 subgroups)	30 hours

Note: Add 3 hours per additional subgroup. ACBC or menu-based conjoint may require additional hours for design and analysis. Market simulator is optional but highly recommended.

Key Milestones (Analytics Perspective)

Analytics Briefing: Initial briefing to understand business objectives, product structure, competitive context, and how results will be used
Attribute & Level Design: Collaborative session to define attributes, levels, and any prohibitions (impossible combinations). This is the most critical input from the project team.
Experimental Design Review: Analytics team generates the experimental design and provides example choice tasks for review
Questionnaire Review: Full survey review including conjoint exercise, screener, and profiling questions
Interim Data Provided: Preliminary data for quality checks — response times, straight-lining, choice consistency
Analytics Run Final Outputs: Utility estimation, importance scores, preference simulations, and market simulator delivered
Simulator Handover (if applicable): Interactive market simulator delivered with documentation for ongoing use
Debrief Attended by Analytics (if necessary): Analytics team available for findings presentation and Q&A session

Questionnaire

Sample Size Calculations

Overall analysis:
Minimum 300 respondents. 500+ preferred for stable Hierarchical Bayes estimates.
Subgroup analysis:
At least 200 respondents per subgroup for reliable segment-level utilities and simulations.
Rule of thumb:
More attributes and levels require larger samples. A common formula: at least 300 × (largest number of levels in any attribute) / (number of concepts per task) respondents.
ACBC:
Can achieve reliable estimates with somewhat smaller samples (200–300) due to its adaptive design, but 400+ is preferred.

Choice Task Structure

Each choice task presents respondents with a set of product concepts and asks them to choose. The structure follows a standard format:

Example — EV Home Charger (Pod Point Pricing Strategy):

This project used 11 attributes to understand how EV owners trade off charger features, services, and price:

Attribute	Levels
Unit Design	Design 1, Design 2, Design 3 (shown as images)
Installation	Separate (arrange via electrician) / Included with purchase
Warranty Length	3 years / 5 years / 7 years
Warranty Coverage	Charging unit only / Unit and installation
Price	£749 / £999 / £1,049 / £1,099 / £1,299
Charging Rate	3.6kW (12–16 hours) / 7kW (6–8 hours)
Smart Charging Tariff Compatible	No / Yes
Energy Tariff with Unit	None / Octopus Energy / EDF Energy / Pod Point Energy / Ovo Energy
Tethered / Untethered	Untethered / Tethered
Solar Integration	No / Yes
Review Site Endorsement	None / Which? Trusted Trader / WhatCar / AutoTrader / Trustpilot

Which of these EV charger packages would you most prefer to purchase?

┌──────────────────┬──────────────────┬──────────────────┐
│    Package A     │    Package B     │    Package C     │
│                  │                  │                  │
│  [Design 1]     │  [Design 3]     │  [Design 2]     │
│  Installation:  │  Installation:  │  Installation:  │
│    Included     │    Separate     │    Included     │
│  Warranty: 5yr  │  Warranty: 3yr  │  Warranty: 7yr  │
│  Coverage: Unit │  Coverage: Unit │  Coverage: Unit  │
│    only         │    + install    │    only         │
│  Rate: 7kW     │  Rate: 3.6kW   │  Rate: 7kW     │
│  Smart tariff:  │  Smart tariff:  │  Smart tariff:  │
│    Yes          │    No           │    Yes          │
│  Tariff: EDF   │  Tariff: None   │  Tariff: Octopus│
│  Tethered      │  Untethered     │  Tethered       │
│  Solar: No     │  Solar: Yes     │  Solar: No      │
│  Endorsed by:  │  No endorsement │  Endorsed by:   │
│    Trustpilot   │                 │    Which?       │
│                  │                  │                  │
│     £1,049      │      £749       │     £1,299      │
├──────────────────┼──────────────────┼──────────────────┤
│       ○          │       ○          │       ○          │
└──────────────────┴──────────────────┴──────────────────┘
                                      ○ None of these

Respondent sees 10–15 tasks like this, each with different attribute combinations. Designs and brand logos were shown as images for realism.

Attribute & Level Design Guidelines

The quality of conjoint results depends heavily on how attributes and levels are defined. This is the most important input from the project team.

Non-Negotiable Elements:

✅ Mutually exclusive levels within each attribute — a product can only have one level per attribute
✅ Independent attributes — the value of one attribute should not depend on the level of another
✅ Realistic combinations — all product concepts shown must be plausible. Use prohibitions to prevent impossible combinations (e.g., a budget brand at a premium price)
✅ Price as an attribute — include price as an attribute with 4–6 realistic price points spanning the expected range
✅ “None of these” option — respondents must be able to opt out if no concept appeals to them
✅ Concise level descriptions — short labels or phrases, not paragraphs. Respondents need to evaluate quickly.
✅ Balanced number of levels — aim for similar numbers of levels across attributes (e.g., all 3–4 levels) to avoid level-number bias

Optional Enhancements:

Add a maximum acceptable price question before the conjoint exercise to calibrate purchase intent
Include a fixed competitive product in each task as a benchmark (similar to Gabor-Granger with competitive context)
Run a holdout task — a choice task not used in estimation, used to validate the model’s predictive accuracy
Add a follow-up purchase intent question after each choice (“How likely would you be to actually buy your chosen option?”)
Use shelf-like visual displays for FMCG categories to increase realism

Attribute Definition Quality:

Poor Attribute Definition	Improved Attribute Definition
“Good quality” (subjective)	“Build material: Aluminum / Plastic / Carbon fiber” (concrete)
“Fast delivery” (vague)	“Delivery time: Next day / 3–5 days / 7–10 days” (specific)
“Premium features” (bundled)	Separate into individual feature attributes
“10 levels of storage” (too many)	Reduce to 3–4 representative levels
“Warranty: Yes / No” (unbalanced)	“Warranty: 1 year / 2 years / 3 years” (graduated)

Attribute Importance

The first output is a summary of how much each attribute contributes to the purchase decision overall. This shows the relative importance of price vs. brand vs. features — answering “what matters most when people choose?”

Importance is calculated from the range of utility scores within each attribute. Wider ranges mean the attribute has more influence on choice.

Utility Scores (Part-Worths)

Individual-level utility scores for every attribute level. These quantify the preference contribution of each level — e.g., how much preference increases when moving from 512GB to 1TB storage, or from Brand A to Brand B.

Utilities are the building blocks for all downstream analyses: preference shares, willingness to pay, and market simulation.

Market Simulator

The most powerful output. An interactive tool that allows stakeholders to:

Define product configurations — set attribute levels for your product and competitors
Predict preference shares — estimate what percentage of the market would choose each product
Test “what if” scenarios — “What happens to our share if we add Feature X and increase price by £100?”
Optimize products — find the configuration that maximizes share, revenue, or margin
Sensitivity analysis — see how share changes as you vary one attribute while holding others constant

Additional Output Types

Conjoint analysis typically includes:

Willingness to pay estimates: The price premium consumers would pay for a specific feature or brand
Price sensitivity curves: How share changes as price increases, holding other attributes constant
Preference share predictions: Market share estimates for defined competitive scenarios
Cannibalization analysis: How adding a new product to your portfolio affects existing products
Segment-level utilities: How different customer segments value attributes differently
Reach/frequency of ideal products: What proportion of the market finds each configuration acceptable

Previous Project Examples

Project 1: Sony PlayStation — Elements of Gaming (April 2020)

Conjoint used to understand what makes a game appealing — not pricing, but product concept optimization
16 attributes covering game design elements: Genre, Sub-genre (literary genre), Era, Setting, How You Play (objective), Main Way to Play, Pace/Speed, Tone, Art Style, Protagonist, Customization, Personal Impact, Skill Level, New vs Continued IP, Single Player, Multiplayer
Respondents saw game concepts as bundles of these elements and chose which game they would most want to play
Used to build a “game generator” simulator — allowing PlayStation teams to test which combinations of elements produce the most appealing game concepts
Attributes and levels were developed through desk research, stakeholder interviews, and virtual ideation sessions
Project Folder

Project 2: Pod Point — Pricing Strategy (August 2024)

Choice-Based Conjoint with 11 attributes covering unit design, installation, warranty, charging rate, smart tariff compatibility, energy tariff partnerships, solar integration, review endorsements, and price
Visual stimuli used for unit designs and brand logos to increase realism
Market simulator built in Excel for ongoing pricing and product configuration decisions
Ran alongside Van Westendorp and Gabor-Granger for complementary pricing insights
Project Folder

Academic Papers and Textbooks

Core Statistical Foundations

Green, P. E., & Srinivasan, V. (1978).
Conjoint Analysis in Consumer Research: Issues and Outlook.
Journal of Consumer Research.
– The foundational paper establishing conjoint analysis in marketing research.
Louviere, J. J., Hensher, D. A., & Swait, J. D. (2000).
Stated Choice Methods: Analysis and Applications.
– Comprehensive textbook covering choice-based conjoint theory, experimental design, and estimation.

Methodological Guidance

Orme, B. K. (2020).
Getting Started with Conjoint Analysis (4th ed.).
– The most practical guide to conjoint analysis for applied researchers. Covers CBC, ACBC, MaxDiff, and market simulation.
Sawtooth Software (2023).
The CBC Technical Paper.
– Technical documentation for the industry-standard conjoint software, covering experimental design, HB estimation, and simulation methods.

Market Research Applications

Rao, V. R. (2014).
Applied Conjoint Analysis.
– Covers practical applications of conjoint in product design, pricing, and competitive strategy, with business case examples.

Ready to use conjoint analysis in your project? Contact the analytics team to discuss your requirements and next steps.

Email: Analytics@dtadvisorygroup.com

What to prepare for our discussion:

The product or service being studied — what is the purchase decision?
Candidate attributes and levels — what product features, brands, and price points should be tested?
Competitive context — which competitors should be included, and what are their current configurations?
Whether a market simulator is needed for ongoing decision-making
Target audience definition and expected sample size
Desired subgroup analyses (markets, segments, demographics)
Decision timeline and budget parameters

--- title: "Conjoint Analysis" author: "TVE Analytics Team" date: last-modified categories: [pricing, product-design] description: "Conjoint Analysis measures how consumers trade off product attributes — features, price, brand — to understand preference, predict demand, and simulate market scenarios" --- :::::: panel-tabset ## What It Is ------------------------------------------------------------------------ ### Overview Conjoint Analysis is a survey-based method that measures how consumers make trade-offs between product attributes. Instead of asking what matters most in isolation, conjoint presents realistic product configurations and forces respondents to choose — revealing what they actually value when trade-offs are involved. The core idea: **products are bundles of attributes**, and consumers evaluate those bundles holistically. Conjoint decomposes that holistic evaluation into the contribution of each attribute and level, producing utility scores that quantify how much each feature, price point, or brand contributes to preference. TVE primarily uses **Choice-Based Conjoint (CBC)**, the industry standard for most commercial applications. In CBC, respondents see a series of choice tasks — each presenting 3–5 product concepts defined by different combinations of attributes — and select which one they would choose (or "none of these"). ::: callout-note **How does conjoint relate to VWD / Gabor-Granger?** Van Westendorp and Gabor-Granger evaluate price in isolation — they work well when the product concept is fixed and the only question is "what should we charge?" Conjoint treats price as one attribute among many, allowing you to understand pricing in the context of feature trade-offs, brand effects, and competitive dynamics. Use conjoint when the pricing question cannot be separated from the product design question. ::: ------------------------------------------------------------------------ ### Types of Conjoint | Type | Description | When to Use | |------------------|-------------------------------|------------------------| | **Choice-Based Conjoint (CBC)** | Respondents choose their preferred option from a set of product concepts. The standard approach for most commercial applications. | Most pricing, product design, and competitive simulation questions | | **Adaptive Choice-Based Conjoint (ACBC)** | An adaptive version that narrows in on each respondent's preferences through a BYO (build-your-own) stage, screening, and choice tasks. | Complex products with many attributes where standard CBC would require too many tasks | | **Menu-Based Conjoint** | Respondents build a product by selecting from menus of optional features at stated prices — modelling add-on and bundling decisions. | Configurable products, subscription tiers, optional extras | TVE selects the appropriate conjoint variant based on the number of attributes, the product structure, and how the results will be used. ------------------------------------------------------------------------ ### How Choice-Based Conjoint Works 1. **Define attributes and levels:** The analytics team works with the project team to define the product attributes (e.g., brand, features, price) and the levels within each attribute (e.g., Brand A / Brand B / Brand C; £499 / £599 / £699). 2. **Generate choice tasks:** An experimental design creates balanced, efficient combinations of attribute levels. Each choice task presents 3–5 product concepts plus a "none of these" option. 3. **Respondents complete choice tasks:** Each respondent sees 10–15 choice tasks and selects their preferred option (or "none") in each. The trade-offs they make reveal their underlying preferences. 4. **Estimate utilities:** Hierarchical Bayes (HB) estimation produces individual-level utility scores for every attribute level — quantifying how much each feature, price point, or brand contributes to preference. 5. **Simulate market scenarios:** The utility scores feed into a market simulator that can test "what if" scenarios — e.g., "What happens to our share if we add this feature and increase price by £50?" ------------------------------------------------------------------------ ### Key Assumptions For conjoint to produce valid results: - **Respondents can evaluate realistic product concepts:** The attributes and levels must reflect meaningful purchase decisions the respondent can engage with. - **Attributes are independent:** Each attribute should be evaluable independently of other attributes. Avoid attributes that logically depend on each other. - **Levels within each attribute are mutually exclusive:** A product can only have one level per attribute (e.g., one price, one brand). - **The "none" option is realistic:** Respondents should genuinely be able to walk away — the category is not a forced purchase. - **The number of attributes is manageable:** Too many attributes (typically more than 7–8) make tasks cognitively overwhelming. Use ACBC or attribute reduction strategies for complex products. ------------------------------------------------------------------------ ## When to Use ### Business Questions This Answers Conjoint is ideal when clients ask: **"How do consumers trade off different product attributes — and what configuration maximizes appeal, share, or revenue?"** Conjoint is not limited to pricing. It can be used for any decision where consumers evaluate bundles of attributes — including product concept design (e.g., what combination of game elements makes the most appealing game?), service configuration, and competitive positioning. It helps answer strategic questions such as: - **Product design:** Which combination of features maximizes preference? What is the optimal product configuration? - **Pricing with trade-offs:** How much more would consumers pay for a specific feature? What is the price premium for our brand vs. competitors? - **Competitive simulation:** What happens to our share if a competitor launches a new product or drops their price? - **Portfolio optimization:** Which products should we offer to maximize total portfolio share without excessive cannibalization? - **Feature valuation:** What is the incremental value of adding a specific feature to the product? - **Willingness to pay:** What is the maximum price consumers will accept for a given product configuration? It is particularly valuable when: - Price cannot be evaluated independently of features or brand - You need to simulate competitive market dynamics - The product has multiple configurable attributes - Stakeholders need a market simulator tool for ongoing decision-making ------------------------------------------------------------------------ ### When NOT to Use Conjoint may not be a good fit when: - **The product concept is fixed and only price varies:** If the product is defined and you only need to find the right price, use **VWD / Gabor-Granger** — it is faster, cheaper, and more focused. - **You only need a ranking of individual items:** If the question is "what matters most?" without trade-offs against price or brand, use **MaxDiff** instead. - **The product has too many attributes:** More than 7–8 attributes in standard CBC becomes cognitively burdensome. Consider ACBC, attribute reduction via PCA, or splitting into separate exercises. - **Respondents lack category familiarity:** If the product is radically new with no reference framework, respondents cannot make meaningful trade-offs. - **Sample size is too small:** Conjoint requires larger samples than simpler methods — at least 300 respondents for reliable estimates. - **Budget or timeline is very constrained:** Conjoint requires more design, programming, and analysis time than simpler pricing methods. ------------------------------------------------------------------------ ### Data Requirements - **Data Type:**\ Survey data from a conjoint exercise. Respondents complete 10–15 choice tasks, each presenting 3–5 product concepts with varying attribute levels. - **Attributes and Levels:**\ Typically 4–7 attributes with 2–5 levels each. More attributes require more choice tasks and larger samples. - **Sample Size:**\ Minimum **300 respondents** for standard CBC. **500+** preferred for stable individual-level utilities and subgroup analysis. Each subgroup needs at least **200 respondents**. - **Experimental Design:**\ The analytics team generates a statistically efficient experimental design that ensures all attribute levels appear with balanced frequency and minimal confounding. - **Product Stimulus:**\ Clear, concise attribute descriptions that respondents can evaluate quickly. Avoid long text descriptions — use labels, icons, or short phrases. ------------------------------------------------------------------------ ## Project Management Info ### Time Allocation | Stage | Hours | |-----------------------------------|--------------| | Kick-off & attribute design | 4 | | Experimental design & programming | 4 | | Analysis - overall | 12 | | Analysis - per subgroup | 3 | | Market simulator build | 4 | | **Total (overall + 2 subgroups)** | **30 hours** | **Note:** Add 3 hours per additional subgroup. ACBC or menu-based conjoint may require additional hours for design and analysis. Market simulator is optional but highly recommended. ------------------------------------------------------------------------ ### Key Milestones (Analytics Perspective) 1. **Analytics Briefing:** Initial briefing to understand business objectives, product structure, competitive context, and how results will be used 2. **Attribute & Level Design:** Collaborative session to define attributes, levels, and any prohibitions (impossible combinations). This is the most critical input from the project team. 3. **Experimental Design Review:** Analytics team generates the experimental design and provides example choice tasks for review 4. **Questionnaire Review:** Full survey review including conjoint exercise, screener, and profiling questions 5. **Interim Data Provided:** Preliminary data for quality checks — response times, straight-lining, choice consistency 6. **Analytics Run Final Outputs:** Utility estimation, importance scores, preference simulations, and market simulator delivered 7. **Simulator Handover (if applicable):** Interactive market simulator delivered with documentation for ongoing use 8. **Debrief Attended by Analytics (if necessary):** Analytics team available for findings presentation and Q&A session ------------------------------------------------------------------------ ## Data Collection Guide ### Questionnaire #### Sample Size Calculations - **Overall analysis:**\ Minimum **300 respondents**. **500+** preferred for stable Hierarchical Bayes estimates. - **Subgroup analysis:**\ At least **200 respondents per subgroup** for reliable segment-level utilities and simulations. - **Rule of thumb:**\ More attributes and levels require larger samples. A common formula: at least **300 × (largest number of levels in any attribute) / (number of concepts per task)** respondents. - **ACBC:**\ Can achieve reliable estimates with somewhat smaller samples (200–300) due to its adaptive design, but 400+ is preferred. ------------------------------------------------------------------------ #### Choice Task Structure Each choice task presents respondents with a set of product concepts and asks them to choose. The structure follows a standard format: ::: example-box **Example — EV Home Charger (Pod Point Pricing Strategy):** This project used 11 attributes to understand how EV owners trade off charger features, services, and price: | Attribute | Levels | |-----------|--------| | **Unit Design** | Design 1, Design 2, Design 3 (shown as images) | | **Installation** | Separate (arrange via electrician) / Included with purchase | | **Warranty Length** | 3 years / 5 years / 7 years | | **Warranty Coverage** | Charging unit only / Unit and installation | | **Price** | £749 / £999 / £1,049 / £1,099 / £1,299 | | **Charging Rate** | 3.6kW (12–16 hours) / 7kW (6–8 hours) | | **Smart Charging Tariff Compatible** | No / Yes | | **Energy Tariff with Unit** | None / Octopus Energy / EDF Energy / Pod Point Energy / Ovo Energy | | **Tethered / Untethered** | Untethered / Tethered | | **Solar Integration** | No / Yes | | **Review Site Endorsement** | None / Which? Trusted Trader / WhatCar / AutoTrader / Trustpilot | ``` Which of these EV charger packages would you most prefer to purchase? ┌──────────────────┬──────────────────┬──────────────────┐ │ Package A │ Package B │ Package C │ │ │ │ │ │ [Design 1] │ [Design 3] │ [Design 2] │ │ Installation: │ Installation: │ Installation: │ │ Included │ Separate │ Included │ │ Warranty: 5yr │ Warranty: 3yr │ Warranty: 7yr │ │ Coverage: Unit │ Coverage: Unit │ Coverage: Unit │ │ only │ + install │ only │ │ Rate: 7kW │ Rate: 3.6kW │ Rate: 7kW │ │ Smart tariff: │ Smart tariff: │ Smart tariff: │ │ Yes │ No │ Yes │ │ Tariff: EDF │ Tariff: None │ Tariff: Octopus│ │ Tethered │ Untethered │ Tethered │ │ Solar: No │ Solar: Yes │ Solar: No │ │ Endorsed by: │ No endorsement │ Endorsed by: │ │ Trustpilot │ │ Which? │ │ │ │ │ │ £1,049 │ £749 │ £1,299 │ ├──────────────────┼──────────────────┼──────────────────┤ │ ○ │ ○ │ ○ │ └──────────────────┴──────────────────┴──────────────────┘ ○ None of these ``` **Respondent sees 10–15 tasks like this, each with different attribute combinations. Designs and brand logos were shown as images for realism.** ::: ------------------------------------------------------------------------ #### Attribute & Level Design Guidelines The quality of conjoint results depends heavily on how attributes and levels are defined. This is the most important input from the project team. **Non-Negotiable Elements:** - ✅ **Mutually exclusive levels** within each attribute — a product can only have one level per attribute - ✅ **Independent attributes** — the value of one attribute should not depend on the level of another - ✅ **Realistic combinations** — all product concepts shown must be plausible. Use prohibitions to prevent impossible combinations (e.g., a budget brand at a premium price) - ✅ **Price as an attribute** — include price as an attribute with 4–6 realistic price points spanning the expected range - ✅ **"None of these" option** — respondents must be able to opt out if no concept appeals to them - ✅ **Concise level descriptions** — short labels or phrases, not paragraphs. Respondents need to evaluate quickly. - ✅ **Balanced number of levels** — aim for similar numbers of levels across attributes (e.g., all 3–4 levels) to avoid level-number bias ------------------------------------------------------------------------ **Optional Enhancements:** - Add a **maximum acceptable price** question before the conjoint exercise to calibrate purchase intent - Include a **fixed competitive product** in each task as a benchmark (similar to Gabor-Granger with competitive context) - Run a **holdout task** — a choice task not used in estimation, used to validate the model's predictive accuracy - Add a **follow-up purchase intent question** after each choice ("How likely would you be to actually buy your chosen option?") - Use **shelf-like visual displays** for FMCG categories to increase realism ------------------------------------------------------------------------ **Attribute Definition Quality:** | Poor Attribute Definition | Improved Attribute Definition | |----------------------------------|--------------------------------------| | "Good quality" (subjective) | "Build material: Aluminum / Plastic / Carbon fiber" (concrete) | | "Fast delivery" (vague) | "Delivery time: Next day / 3–5 days / 7–10 days" (specific) | | "Premium features" (bundled) | Separate into individual feature attributes | | "10 levels of storage" (too many) | Reduce to 3–4 representative levels | | "Warranty: Yes / No" (unbalanced) | "Warranty: 1 year / 2 years / 3 years" (graduated) | ------------------------------------------------------------------------ ## Output Examples ### Attribute Importance The first output is a summary of how much each attribute contributes to the purchase decision overall. This shows the relative importance of price vs. brand vs. features — answering "what matters most when people choose?" Importance is calculated from the range of utility scores within each attribute. Wider ranges mean the attribute has more influence on choice. ------------------------------------------------------------------------ ### Utility Scores (Part-Worths) Individual-level utility scores for every attribute level. These quantify the preference contribution of each level — e.g., how much preference increases when moving from 512GB to 1TB storage, or from Brand A to Brand B. Utilities are the building blocks for all downstream analyses: preference shares, willingness to pay, and market simulation. ------------------------------------------------------------------------ ### Market Simulator The most powerful output. An interactive tool that allows stakeholders to: - **Define product configurations** — set attribute levels for your product and competitors - **Predict preference shares** — estimate what percentage of the market would choose each product - **Test "what if" scenarios** — "What happens to our share if we add Feature X and increase price by £100?" - **Optimize products** — find the configuration that maximizes share, revenue, or margin - **Sensitivity analysis** — see how share changes as you vary one attribute while holding others constant ------------------------------------------------------------------------ ### Additional Output Types Conjoint analysis typically includes: - **Willingness to pay estimates:** The price premium consumers would pay for a specific feature or brand - **Price sensitivity curves:** How share changes as price increases, holding other attributes constant - **Preference share predictions:** Market share estimates for defined competitive scenarios - **Cannibalization analysis:** How adding a new product to your portfolio affects existing products - **Segment-level utilities:** How different customer segments value attributes differently - **Reach/frequency of ideal products:** What proportion of the market finds each configuration acceptable ------------------------------------------------------------------------ ## Resources ### Previous Project Examples **Project 1: Sony PlayStation — Elements of Gaming (April 2020)** - Conjoint used to understand **what makes a game appealing** — not pricing, but product concept optimization - 16 attributes covering game design elements: Genre, Sub-genre (literary genre), Era, Setting, How You Play (objective), Main Way to Play, Pace/Speed, Tone, Art Style, Protagonist, Customization, Personal Impact, Skill Level, New vs Continued IP, Single Player, Multiplayer - Respondents saw game concepts as bundles of these elements and chose which game they would most want to play - Used to build a "game generator" simulator — allowing PlayStation teams to test which combinations of elements produce the most appealing game concepts - Attributes and levels were developed through desk research, stakeholder interviews, and virtual ideation sessions - [Project Folder]('Z:\Shared\TVE%20Data\1.%20Client%20Projects\Sony\PlayStation\60.%203324%20-%20Elements%20of%20Gaming%202020%20-%20April%202020') **Project 2: Pod Point — Pricing Strategy (August 2024)** - Choice-Based Conjoint with 11 attributes covering unit design, installation, warranty, charging rate, smart tariff compatibility, energy tariff partnerships, solar integration, review endorsements, and price - Visual stimuli used for unit designs and brand logos to increase realism - Market simulator built in Excel for ongoing pricing and product configuration decisions - Ran alongside Van Westendorp and Gabor-Granger for complementary pricing insights - [Project Folder]('Z:\Shared\TVE%20Data\1.%20Client%20Projects\Pod\4415%20-%20Pricing%20Strategy%20-%20August%202024') ------------------------------------------------------------------------ ### Academic Papers and Textbooks #### Core Statistical Foundations - **Green, P. E., & Srinivasan, V. (1978).**\ *Conjoint Analysis in Consumer Research: Issues and Outlook.*\ *Journal of Consumer Research.*\ -- The foundational paper establishing conjoint analysis in marketing research. - **Louviere, J. J., Hensher, D. A., & Swait, J. D. (2000).**\ *Stated Choice Methods: Analysis and Applications.*\ -- Comprehensive textbook covering choice-based conjoint theory, experimental design, and estimation. ------------------------------------------------------------------------ #### Methodological Guidance - **Orme, B. K. (2020).**\ *Getting Started with Conjoint Analysis (4th ed.).*\ -- The most practical guide to conjoint analysis for applied researchers. Covers CBC, ACBC, MaxDiff, and market simulation. - **Sawtooth Software (2023).**\ *The CBC Technical Paper.*\ -- Technical documentation for the industry-standard conjoint software, covering experimental design, HB estimation, and simulation methods. ------------------------------------------------------------------------ #### Market Research Applications - **Rao, V. R. (2014).**\ *Applied Conjoint Analysis.*\ -- Covers practical applications of conjoint in product design, pricing, and competitive strategy, with business case examples. ------------------------------------------------------------------------ ## Get Started Ready to use conjoint analysis in your project? [Contact the analytics team](mailto:Analytics@dtadvisorygroup.com) to discuss your requirements and next steps. **Email:** Analytics\@dtadvisorygroup.com **What to prepare for our discussion:** - The product or service being studied — what is the purchase decision? - Candidate attributes and levels — what product features, brands, and price points should be tested? - Competitive context — which competitors should be included, and what are their current configurations? - Whether a market simulator is needed for ongoing decision-making - Target audience definition and expected sample size - Desired subgroup analyses (markets, segments, demographics) - Decision timeline and budget parameters ::::::