How to set up a conjoint analysis study
Setting up a conjoint study with Conjointly is a quick and easy process. This checklist guide will help you get the most out of our platform for your next survey.
The first inputs that you need for a conjoint study are the attributes and levels of your product that you are looking to test. Consider price as a potential attribute, particularly if you are looking to find metrics such as price elasticity of demand or marginal willingness to pay. If you wish to test your products within a competitive context, consider including attributes that belong to competitors, such as their branding or claims.
Next consider if you want to include additional questions within your survey, as these can give valuable insights and be used in methods such as segmentation.
Finally, prepare any materials you wish to use to customise your survey. These can include images, logos, and videos to customise the experiment, as well as any extra information you would like the respondent to see to have a better understanding of your products.
Describe your experiment
Choose a name that distinguishes your experiment from others you are conducting to avoid confusion. ️
Consider your experiment’s purpose (e.g. finding optimal price), what kind of product you are researching, and who your study is for when writing your description.
Specify attributes
Attributes are ‘dimensions’ of your product, such as price, colour, shape, size, brand, or location.
Specify attributes which you believe to be most important to your customers’ buying decisions.
Include any attributes whose importance you are unsure of but would like to check.
Add a maximum of seven attributes to keep survey design user-friendly.
Specify levels for attributes
Levels are the ‘values’ that each attribute can take. For example, the attribute ‘colour’ can have levels ‘blue’, ‘red’, ‘transparent’.
Have at least two levels per attribute.
Describe your attributes and levels in clear language that consumers will understand.
Add respondents
Discrete choice experimentation typically requires between 100 and 1,000 participants. Conjointly offers several options to get participants for your experiments.
- Bring your own respondents: Use a list of your customers or leads, or source respondents from elsewhere or integrate with your other survey tool.
- Predefined panel: Use a specific pre-defined panel of respondents (including those who are pre-defined for your company). Data collection is managed by the Conjointly fieldwork team.
- Self-serve sample A: Automatically and urgently invite respondents from global panel networks. Target them by location, demographics, and profiling data. Set your own price per complete and manage your own quotas.
- Managed general sample: Buy responses from our panel networks. Target them by location, demographics, and profiling questions. Data collection is managed by the Conjointly fieldwork team.
- Email campaign: Send survey invitations and reminders via Conjointly to your own contacts.
Customise your experiment or add profiling questions
Conjointly lets you change question text, add a redirect link, and perform other customisations as optional features.
Ask non-conjoint questions to use for customer segment profiling.
Ask other questions types such as multiple-choice, long text, short text, number, constant-sum.
Review, approve, launch
Conjointly manages the response collection process once you have launched your experiment. We will notify you when your survey report is ready.
Review your experiment settings.
Agree to Terms & Conditions.
Pay for your experiment securely.
Launch.
FAQs
How many attributes can I test in a conjoint survey?
The general rule of thumb is to include no more than 7 attributes in a conjoint study because including more than 7 attributes will impose a substantial cognitive load on respondents, especially when they are accessing the survey via a mobile device. If your attributes or levels description is lengthy, we recommend either make the description more concise or include fewer attributes in the study.
If you do have a need to test a dozen or more attributes we suggest one of three options:
First, run a screening survey to see what attributes respondents rank high (you can do a multiple-choice question or a Likert scale question type). Once you get the winning attributes, you can use these in a conjoint study.
Alternatively, you can employ partial profile conjoint analysis, which can handle a large number of attributes because not all attributes are shown to participants at the same time. Please contact us to arrange this type of study.
Another way is to use MaxDiff with just one attribute “Features” to narrow down the number of attributes to the top seven most important ones. Once you have identified these attributes, you can set up a conjoint using only the top attributes identified earlier.
How do I change the order of attributes in my experiment?
If you’d like to present your conjoint questions in a particular order, you can arrange the order of attributes during your experiment set-up.
To enable this:
Go to your experiment list
Click the Advanced Settings tab and select the “Conjoint survey options” module.
Select “Manually specify fixed order of attributes” from the “Order of attributes” dropdown menu.
You can now move and rearrange the attributes by clicking the button next to the attributes.
How to add description to conjoint attributes / levels?
You can add a description to your attributes by clicking on the button as shown below:
You can also select how the description will be displayed, either displayed at the bottom of the screen, when the respondent hovers their mouse over the attribute, or both.
You can see an example of the survey interface with an attribute description displayed as a tooltip and at the bottom of the screen:
Next steps
- Learn more about how to avoid common mistakes in conjoint studies.
- See more example conjoint reports in your experiments.