Logo Help Center
Chat with us
Open a ticket
Sign in
  1. aytm Help Center
  2. Insight Guides
  3. Survey Best Practices

Articles in this section

  • Monadic Concept Testing with aytm (Video)
  • Best Practices: Translating surveys
  • Kano Continuous Methodology
  • Conjoint: Best Practices and Design Recommendations
  • Exporting side-by-side concept comparisons (Video)
  • Kano Discrete Analysis Methodology
  • Piping Responses from MaxDiff

Conjoint: Best Practices and Design Recommendations

Here are some suggestions and recommendations when using choice-based conjoint studies in your surveys on the aytm platform. 

 

Express or Segmentation? mceclip0.png

When programming, you will need to indicate whether you need Express or Segmentation Conjoint.

Express

This option offers a less robust analysis that treats respondents as one big data pool. It requires fewer choice trials and is suitable when you are only interested in a total-level view.

Segmentation

This option includes a more robust analysis methodology that utilizes Hierarchical Bayesian modeling to produce individual-level utility scores, allowing for deeper analysis of your results. This option requires more data for individual-level utilities and thus leads to more choice trials for respondents.

This option is ideal when you are interested in examining conjoint results among subgroups. It also produces auto-segments based on similar preferences, noting additional similarities between traits and answers.

 

Things to Keep in Mind

The Basics

Conjoint analysis requires at least two attributes, or features, with each having at least two levels, or options. Generally, a standard conjoint design has no more than five attributes each, with no more than six to seven levels.

Conjoint analysis assumes all attributes and levels are independent. Additional information can be displayed as part of the task that reflects a combinations of certain attributes or options, but should not be part of the underlying design or analysis.

For example, you should not test: 
(a) shelf price,
(b) cans per pack, and
(c) price per can each
as attributes within one conjoint.

Why? Because price per can is a direct result of shelf price and cans per pack.

In cases where two attributes are truly an impossible combination, such as a new Samsung smartphone running iOS, prohibitions may be put into the design, however, these should be used rarely and sparingly. In the majority of cases, it is not recommended to restrict the experimental design.  Instead, opt not to simulate combinations that would not be produced in the real world (e.g., best feature set at the lowest price point).

Attributes and Levels

It is more important to think in how many unique combinations of levels are possible than strictly number of attributes and levels. This can be calculated quickly by multiplying the number of levels across all attributes (e.g., three attributes with two levels each is 2X2X2=8 unique combinations).

  • The aytm platform suggests a limit of 50,000 possible combinations.

When assembling your conjoint design, carefully determine the number of levels and breadth of levels. The more levels an attribute has and/or the greater the disparity between levels, the more likely it is that that attribute will be important compared to others, and vice versa.

  • This is mostly commonly seen with price: testing a price range of $5 to $100 will be much more important in the analysis and results than testing a price range of $5 to $10. Everything is relative!

 

Design Recommendations

Include a “None” Option

We’ve all gone to a shelf, looked at our options, and walked away without a purchase. That’s what the none option represents and why it’s so important to include in the majority of conjoint exercises. When you include a none option, you can simulate scenarios with and without a none option. When you don’t include a none option, you cannot simulate the real world action of walking away without purchase.

Anchor Choices with Price

Successful conjoint exercises compel respondents to make tradeoffs, mostly commonly with what a product or offering will cost. Many features sound great to consumers until they’re faced with the price to pay for those features. Even when looking for early-stage directional data on the types of features consumers are interested in--regardless of cost--including some reference to price in the conjoint design or choice task itself  can help respondents root their responses in the reality that any product has a cost.

Example:
Imagine this product cost $XX. Please select the configuration most appealing to you.

 

When to Work With Professional Services

Here's a helpful breakdown of what you get when you use a choice-based conjoint with aytm, depending on whether you do it yourself or seek our Professional Services team.

  DIY Professional Services
Design/Creation

Simple Editor Set-Up

  • Specify Attributes and Levels
  • Set Profiles per Task, None Option
  • Aggregate or Individual Analysis Option

Automated Design Sheet

Design Consultation

Advanced Programming

Custom Design Sheet

Analysis/Reporting

Interactive Online Tool

  • Run Market Simulations
  • Interactive preference table

Automated Analysis Producing Utilities

Simulator Excel Output

Preference Likelihood Excel Output

Custom Simulator

  • Additional metrics (e.g., est. revenue)
  • Built-in data filters
  • Custom builds based on objectives!

Full Reporting Available

Consultative Review Included

 

Additional Resources

  • Overview: Conjoint Research Test
  • Conjoint Express Methodology
  • Conjoint Segmentation Methodology
  • Facebook
  • Twitter
  • LinkedIn
Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request
Return to top
  • Our platform
  • Solutions
  • Pricing
  • Our panels
  • Help center
  • Contact us
  • Blog
  • Privacy
  • TOU
  • About
  • Careers
  • Innovation lab
  • Demo
© 2022, Umongous, Inc. All rights reserved.

zendesk theme design by aytm c/o diziana