OVERVIEW

Zensors is a startup based out of Carnegie Mellon University's Future Interfaces Group that is working on developing a flexible and intelligent computer vision product that anyone can use. I was brought on to the Zensors project as part of a design team that could produce a robust front-end that would be ready to ship as a product in 8 months.

Problem:

How can I support a startup's front-end development process with research at all stages of the product development cycle?

Solution:

  • Lead a two-month exploratory user research study to identify product-market fit
  • Develop personas to communicate research findings
  • Integrate evaluative user testing into two-week development sprints over a two month development process
  • Advocate for a forward-looking mental model research study that aims to understand the fundamental motivations of users

Project Details:

Masters of Human-Computer Interaction capstone thesis project, 8 months duration, 5 team members

Methods

Interviews

Heuristic Evaluation

Expert Evaluation

Think-Aloud Protocol

Cognitive Task Analysis

Thematic Analysis

Responsibilities

User Research

Business Development

UX Design

Project Management

Deliverables

30+ User Interviews

50+ User Tests

28 Question Survey

3 Personas

4 Research Proposals

3 Findings Reports

10+ Research Protocols

Our capstone project has three key research phases: exploratory research, user testing for MVP, and mental model research. You can read more in detail about each research phase below or read about design and development from our process website here.

PROCESS

Phase 1: Bringing value through research to Zensors

I anchored our fast-paced design and development work in insights from users. My main responsibilities include devising research strategy, writing protocol, conducting user studies, analyzing data, and presenting findings to clients.

The first major task our clients asked for help with was finding the right market vertical for their product to launch in. To accomplish this, I came up with a plan for leveraging contextual and remote interviews as a primary method for the exploratory research phase (February - May 2018) of our capstone project.

You can learn more about our research strategy in the presentation below. I also developed protocol surrounding our research interviews, assisted with recruitment methods, and lead data analysis for our three month exploratory research phase.

Hold on a minute. You might be wondering: Why qualitative methods for such a large-scale exploration of potential markets? Why not a survey instead?

A qualitative approach suit our purposes for several reasons. First, our team needed to personally understand the concerns and needs of users from a variety of different verticals in order to narrow down on the ones with the most potential. Second, large-scale methods like surveys would be incredibly hard to do correctly in the short time span we had to identify a vertical and start designing. Third, a lack of scalability in our methodology wasn't a worry since we were comfortable with a high amount of false negatives in order to find one true positive for our market.

Based on this strategy, we interviewed 30+ professionals across 11 different market verticals over one month. These verticals were decided through discussion with our client and personal research on industries that benefited analyzing physical spaces. Reddit, craigslist, Facebook, LinkedIn, personal connections — we cast a wide net to recruit participants. We synthesized our user interviews using thematic analysis and heat mapping to provide a convincing argument to our clients that Zensors could make a strong impact in coworking and baking industries.

We presented our initial findings to our clients in a formal presentation and made a strong case for pursuing coworking and baking that had the possibility of being a strong fit for Zensors.

Our next step was to discover more about coworking spaces and baking. While most of our interviews were conducted remotely, we also had the opportunity to do contextual interviews including visits to coworking offices and bakeries. We quickly realized that industrial baking wasn't as promising as it originally sounded, and shifted the remainder of our exploratory research phase to verifying that coworking would be a strong vertical for Zensors.

Here are some photos of our visits to coworking offices to speak with employees about their needs. Contextual interviews also contained a post-it exercise in which we asked the participant to walk around their facilities and tag objects or locations they would want to obtain data about.

After three weeks of in-depth research on coworking, we ended up with a mountain of interview data collected from over 10 different companies. Based on this data, I wrote three personas that embodied the broad categories of potential users that we saw.

These personas helped inform our design process and became a core part of a presentation to the Future Interfaces Group in which we presented the results of our market research, user research, and design strategy from February 2018 to May 2018.

Phase 2: Integrating user testing and MVP development

With exploratory research complete in May, our clients wanted us to develop a fully functioning front-end UI for deployment as part of a customer-ready MVP targeting July deployment. This was a major challenge considering we had no functioning code and just a few mid-fidelity wireframes at the start of our evaluative research phase (May - July 2018).

From a research perspective, I had to shift from research taking lead on the project to playing a support role for design and development. To help my team shift gears, I worked to integrate user testing into a packed design and development schedule. Since we were using two week sprints to guide development, I devised a research plan that best allowed us to iterate quickly based on feedback from user testing.

With only one month to validate and iterate on our designs, we decided to focus on user testing for a MVP. Integrated within each sprint was time for preparation, recruitment, synthesis, and design. Often, each phase would need to be tightly controlled and passed on to another team member as soon as possible. We also decided to offset our design and user testing so that it would be possible to conduct user testing, synthesize, and iterate within a single two-week period.

We utilized a number of budget usability testing methods due to the entrepreneurial nature of our project. As a start-up focused on developing a product, costly user testing methods were simply out of the question for Zensors. We chose to use expert and heuristic evaluation for internal user testing with HCI professionals and students, and conducted external think-aloud testing using coworking or facilities management professionals as participants.

An early prototype built in InVision.

Above you can see our team's InVision prototype, which is semi-interactive and allowed us to conduct user testing. I worked closely with our design leads to ensure that the testing protocol and prototypes match. Below, you can see a screenshot of a remote think-aloud test conducted on Google Meets. We received permission to record video from participants.

A screenshot from one of our remote user tests.

By the end of our one month MVP development phase, we completed 12 heuristic or expert evaluations with HCI practitioners and 13 think-alouds with professionals working in facilities management. For each sprint, we analyzed our data thematically to identify common issues users had, ranked these issues, and built a design backlog that would be used for the next iteration of the interface.

I'm currently working on presenting our MVP usability testing research to our clients. Our conclusions from user testing will be uploaded as the project continues!

Evaluative Research: Studying mental models

With our MVP development nearing completion, I'm recently piloted an ambitious mental model study focused on visualizing naive models of Zensors as a system, correlating mental models with performance in user testing, and integrating these models into our design process. With this mental model work, we would be able to suggest a future direction for Zensors design and product strategy based on user behavior. For example, identifying mental models that inhibit successful use of the product interface could help inform the design of an onboarding process for Zensors in the future.

This study was recently completed, and draws from a number of research theories and methods in cognitive psychology and learning sciences. I'll be uploading our results once we wrap up our capstone project in August!