CBC core settings

Introduction

The goal of CBC design is to collect enough information from each respondent to accurately estimate their preferences without overwhelming them. Discover automatically recommends settings for the number of tasks, concepts, and versions in your exercise.

These recommendations strike a balance between precision and respondent effort, reduce order bias, and ensure that attribute levels are combined and distributed evenly across respondents.

If you want to change these defaults, turn on the Override recommendations toggle:

  • With the toggle off, input fields are disabled, and Discover’s recommended settings are applied.
  • With the toggle on, you can enter your own values. Switching it back off reverts to the defaults.

Tasks

Number of CBC Tasks

This setting controls the number of choice tasks each respondent will answer.

Sometimes, Discover may recommend a relatively high task count. In general, we don’t recommend lowering this number, since fewer tasks reduce the information available for estimating accurate preference scores.

If the recommended number of tasks feels too high for your audience, consider:

  • Reducing attributes or levels. This reduces the number of tasks required while maintaining high accuracy high. We suggest limiting exercises to about 10 attributes or fewer to minimize fatigue.
  • Using larger sample sizes. With hundreds of respondents, random error tends to cancel out. In these cases, you may be able to use fewer tasks while still achieving reliable aggregate results.

Concepts per task

Number of Concepts Per Task setting is highlighted.

This setting controls the number of concepts respondents compare in each task.

  • Discover bases its recommendation on the largest number of levels in any attribute.
  • Statistically, more concepts per task improve precision.
  • Practically, more concepts increase cognitive burden and may be harder to display on smaller screens.

Guidelines:

  • Show at least three concepts per task (plus the “None” option) in most studies.
  • In sensitive contexts (e.g., healthcare decisions), two concepts may be more manageable.

Troubleshooting

As you set up a CBC, certain design choices may be flagged with warnings or errors in the survey audit. These messages highlight potential issues with data quality or respondent experience. Here’s what each one means and how to address it.

Given your attributes and levels and if you need precise individual-level utility estimation, we recommend [#] CBC tasks.

The designer may recommend a large number of tasks if your attribute/level list is long. This warning is shown because that task count is needed to produce accurate utility scores for each respondent. If the number feels too high for respondents, consider reducing attributes or levels.

Tip: If you have ample sample size and are willing to accept more noise at the individual level than is typical for conjoint analysis, you may proceed with a sparse design featuring fewer tasks than recommended.

Requiring more than 30 CBC tasks may be difficult for respondents to adequately complete. It is typical in practice to have 15 or fewer tasks. Consider reducing tasks or the number of attributes and levels.

CBC designs with more than 30 tasks risk respondent fatigue and lower-quality data. While you can override this, it’s best to simplify attributes/levels to keep task counts reasonable. Proceed with caution if you exceed 30.

Number of concepts must be [#] (the total number of possible product combinations) or less.

This error occurs when the number of concepts you requested exceeds the number of unique product profiles that are possible. Showing duplicate concepts in a task can confuse respondents, so Discover enforces this limit.

Number of concepts per task must be 2 or more.

At least two concepts (in addition to the “None” option) are required; otherwise, respondents would have no meaningful choice to make.

Showing more than 16 concepts per task may be difficult for respondents to adequately compare all the given attribute levels. Consider reducing the total.

Comparing many concepts simultaneously can be taxing and may compromise data quality. Large sets are also more difficult to display on smaller screens. Keep concepts manageable to improve usability and achieve better results.