Conduct A/B testing.

The Problem

I want to test two different variables with my audience and compare their reactions.

The Plan

A/B testing, or split testing, is a classic method of market research and employee research for testing the control and the variable of an experiment. Create a Flex Conversation and randomize or rotate the variables to ensure participants see the questions in different order.

• Program a Flex Discussion Guide Using Sections
• Set up Display Logic
• Analyze Your Data
• Alternate Method: Two Live Conversations

Program a Flex Discussion Guide Using Sections

First, create your Flex Conversation, then get started programming your Discussion Guide. By programming your Discussion Guide in the platform, you can easily incorporate Sections which are a necessary aspect of this process. Sections allow you to organize your discussion guide, as well as program randomization and rotation. Be sure to program all the content you want to be randomized within these segments, and name them something that will help you remember such as “control” and “variable.”

Set up Display Logic

From the Display Logic page, you can easily select your sections and whether you’d like them to be rotated or randomized.

• Rotate: Sections will display in a rotating order for each participant, based on when they start the conversation. For example, the first participant will see 1, 2, 3, the second participant will see 2, 3, 1, etc.Random
• Randomize: Sections will display in a random order for each participant. This option typically has the highest number of possible order permutations.

Analyze Data

You can monitor your data throughout the Conversation, then once the Conversation has completed download our Participant Data CSV to see which participant saw which concept first. You can then use that to cut the data in our CSV Exports and compare.

Alternate Method: Two Live Conversations

Conduct two separate Remesh Live sessions with separate participant groups. The control and variable would be evaluated independently, reducing the risk of bias in the conversations. After both sessions are completed, you would analyze the separate datasets from these conversations for comparative analysis and could also merge these datasets as well to compare the data from each session.

Note that in order to analyze the conversations afterwards, the questions need to be consistent between the two sessions, which might be difficult to manage. Additionally, running two separate sessions requires additional effort to recruit participants and moderate for each of the conversations. Finally, analyzing the two separate sessions requires additional effort to compare and contrast the data collected.

Next Steps

Ready to use Remesh to conduct A/B testing? If you already have a Remesh subscription, simply create your Conversation to get started! Feel free to get in touch if you have any questions.

If you do not have an existing Remesh subscription, please reach out and our team will be happy to assist!