Meaningful Complexity:
Balancing Utility and Usability
Improving a highly complex and technical feature, for automating chatbot conversions (RPA), through workflow and information architecture optimization to improve usability and navigation while aligning technical feasibility.
Lead Product Designer
Strategy, Research, Design, Testing
6 months
Usability and utility are equally important and together determine whether something is useful
-- Jakob Nielsen
  • The company: Zeals' chatbot platform enhances the traditional consumer experience by providing personalized customer service through chatbots on different messaging platforms.
  • The user: Zeals' Communication Designers (CD) design these personalized chatbots as an extension of their clients' marketing or advertising strategy through our enterprise platform.
  • The feature: During conversations, when customers are ready to make a purchase, the chatbot asks for necessary information and relays it to the client's webform using a process called Robotic Process Automation (RPA).
The redesign of the feature plays a crucial role in attracting new clients by providing a seamless and convenient avenue for customer engagement and data collection.
What do we want to achieve?
Key Goals:
  • Increase number of new clients: Increasing the number of RPA implementation means more conversions. However, only a few CDs are familiar with the RPA setup process due to its complexity.
  • Improve time to implementation: Reducing implementation time will enable providing RPA solutions to more clients. Currently, the set-up process takes an average of 15 to 20 business days, slowing down adoption.
  • Accommodate new database: A new database is being deployed to accommodate more complex chatbots. Ensuring that RPA is technically compatible with the new database is crucial to avoid business disruption.
What do users think?
To get to the root of the problems that hinder the product from achieving its goals, we try to understand the pain points and challenges that communication designers and developers face when setting up RPA, and to identify opportunities for improving the feature.
The old system was a bottleneck for improving speed due to excessively complex customizability.
Setting up RPA is highly technical and involves unnecessary steps that are not very useful for Communication Designers (but may be useful for developers).
Navigation between different disconnected products is a major contributor to the feature’s complexity.
Communication Designers struggle navigating RPA's interface, which forces them to move between disconnected parts of the platform, causing delays and inefficiencies.
Requesting developers to create "in-chat webforms" further delays the implementation process.
Having an "in-chat webform" is crucial for attracting new clients. However, the whole process is time-consuming, and the creation of “in-chat webform” requires a significant amount of steps and coordination with developers.
How might we...
After gaining insights and learning more about users' needs, we then focus on core problems and brainstorm potential solutions.
How might we improve speed of setup through reducing complexity to provide more implementations for clients?
Every application has an inherent amount of complexity that cannot be removed or hidden. Instead, it must be dealt with, either in product development or in user interaction.
-- Larry Tesler
According to Larry Tesler, there's a limit on how simple a product can be. Oversimplifying features of a product that is inherently complex risks losing its usefulness. Our goal is to reduce complexities without significant risks.
To simplify the setup process, I optimized the user flow by removing redundant steps discovered from the CDs workflow. This was done by identifying the core steps/features, and determining which could be removed or handled as background services through collaboration with developers.
How might we improve the setup flow through optimization of navigation and information architecture?
Having a cohesive user flow allowed all the setup steps to be performed within a single, central feature called the RPA Builder.
Brainstorm sketches and mockups
Final RPA Builder dashboard
To ensure users are well-informed of their progress and completed settings, it's important to provide clear feedback, even if it requires displaying a significant amount of information on the screen.
One of the most time consuming tasks was finding the error within the form automation settings. Displaying the overall status of the settings in one glance helps shave off hours (or even days) of trying to debug/finding the error.
How might we accommodate the new database, while adding new features to improve implementation speed?
I removed the need for Communication Designers to request developers to create in-chat webforms, by allowing them to create no-code webforms within RPA Builder.
Periodic design reviews with the entire team ensure alignment, prepare engineers for implementation and testing, and provide security by clarifying the scope and effort needed.
Final Designs
How can we further improve?
Continuous discovery and iteration helped the team identify and address usability issues in the new feature, leading to iterative improvements to the product, and to the whole flow.
Key Insights:
  • Technical and vague copy caused delays on users' decision making
  • Visual differences in related UI elements caused confusion in users' navigation
  • Templates go against the concept of personalizing chatbot designs
In an effort to continuously improve our product, I established a feedback process after each release that includes quantitative (System Usability Scale) and qualitative (interviews, workshops) measurements to assess effectiveness. The process was proposed to managers and product leads and implemented several months after the initial RPA feature release.
  • Continuous improvement through SUS (70.83), interviews, workshops
  • Demo day conducted to address concerns from feedback
  • Received an award for the project: For implementing continuous iteration processes, and standard discovery practices in the development team.
  • Implementation speed improved at an average of 33%
  • No. of Communication Designers who are able to use the system continuously increasing, while providing support through training, and documentation
SUS Scores
Interview Results
Value Achievement Award
RPA Demo Day in office
Users struggle to articulate their needs, and may express preferences that differ from their true motivations
Generative research, through interviews and observations, uncovered users' underlying needs beyond their expressed preferences. This deeper understanding of the core features and pain points of the old system led to a more impactful redesign that truly resonates with users.
Deep understanding of the system's technical capabilities greatly helped in the redesign
Integrating technical considerations into the design process allowed the team to create a user-centric solution that met the users' needs while also aligning with the development team's capabilities and infrastructure.
Improving copywriting played a critical role in enhancing user experience
The use of technical terms and jargon often created barriers for users, hindering their ability to understand and navigate the product effectively. By ensuring that the language used was aligned with users' mental models, the team was able to empower users to confidently interact with the feature.
A balance of simplicity, and supplemental learning is key
We were able to help users get familiarized with the complex feature through a right balance of simplicity that matched their mental models, and a right amount of supplemental learning and contextual clues. This helped even newcomers use the new feature, and get familiarized in the shortest possible time.