Behavioral prototyping — How can we test complex behavior of a product without really implementing it

This week in “Mastering Prototyping Techniques”, we further embraced a new technique of prototyping — behavioral prototyping. This technique aims to allow designers to conduct usability testing on the complex behaviors of a product without implementing it digitally. To mimic the behaviors of the product, one person in the design team would pretend they are the product and reacts to the interaction the user has with the product.

Per usual, we implemented a prototype using this technique to demonstrate our understanding and learning.

Behavioral Prototype

Amazon Alexa

To demonstrate our understanding, we chose to create a behavioral prototype of a voice-controlled cooking assistant system. To allow ourselves to focus on designing a behavioral prototype instead of the product, we imagined our cooking assistant system to be part of Amazon Alexa, thus the common interaction with the cooking assistant system would be the same as interacting with Amazon Alexa.

Design — use scenario

We started by thinking about the scenario our users would have with the cooking assistant system. We started by a group brainstorming on the use cases and we sketched out the use cases in a storyboard form along the way. After discussion, we decided to define our scenario to be “the user is hungry and would like to ask Alexa for direction on making food”. With the scenario clearly defined, we were then able to define our usability research questions as well as the direction of the behavioral prototype.

one of the sketched storyboard

Design — usability testing & approach

With the scenario defined, we proceeded to define the usability questions. After group brainstorming, we defined the usability questions to be following:

  1. How can users know what kind of commands/options are available?
  2. Can users navigate back and forth through the steps of the recipe?
  3. Can users successfully follow along the instructions?

We further defined our methodology of this usability testing to be largely qualitative — we collected data from observing user’s behavior interacting with the system, their thoughts (from thinking aloud) and their verbal feedback. As any standard usability testing, we defined one of our team members to be the moderator and anther one to be the note taker.

Design — wizarding the behavior

To rapidly prototype the behavior of the voice-controlled Alexa cooking assistant system, we decided to use two iphones to mimic this interaction. We connected two phones via phone call and we turned on the speakers of both phone. With the screen of both iphones locked, we concealed the fact that the phones were in a phone call. Then, we decided to have one of our team members take one of the iphones and hide in the dark to interact with the users on the other iphone using their voice. The behavior of a voice-controlled cooking system was then rapidly mimicked! Of course, we created a script to make sure the voice interaction of the behavior prototype was consistent.

We then carried out the usability testing with one potential user, you can watched part of the usability testing session here:


After the usability test, the note from the test session was thoroughly read by the team. Then, using the notes, we held another team brainstorming session on what went well for the prototype and what needed improvement.

What went well

The participant was able to quickly learn how to use the system from the interaction with the system. The language used by the system was clear enough for the participant to make decision. The participant was also able to follow along the cooking instructions with little difficulties.

What needed improvement

The participant reported that although he could follow along the instruction through talking to the system, he would like some sorts of visual cues to assist him in completing the task. Further, he claimed that the instructions could sometimes be too long. To address this, we decided to divide the long instructions into separate sub-steps.


Overall, the design of the behavioral prototype successfully allowed us to test on the complex behavior of a design idea. The design of the Alexa cooking system was somewhat effective as the participant was able to follow along for the most part. In this era of rapid technological development, I can definitely see behavioral prototyping become a powerful tool for designers to validate their design ideas for complex system. This could save cost and time of the design drastically.