I started with supporting UI/UX design on front-of-house customer experiences and transitioned to product design for back-of-house features, relating to policy administration and claims handling. The majority of my work focused on achieving parity with a 3rd-party system, reducing repetitive workflows and optimising usability.
To understand the existing system and identify tactical gaps and opportunities, discovery methods like user interviews, shadowing, surveys, UI/UX audits, content and data model mapping were used to assess complexity and challenges of the system, highlight user pain points and recurring UI/UX patterns.
Over time, insights gained through tactical design projects assisted in validating hypotheses and providing steer for iterative improvements.
Customer and payment information is repeated across policies
Policy
Pet A + Claims
Customer
Payment
Policy
Pet B + Claims
Customer
Payment
Serving increasing user types, domains, and intent
•
Initially developed as a basic policy administrative system for customer support agents and claims handlers, the platform struggled to support a growing range of user types, such as complaints handlers, and auditors.
•
Each user type brought unique and often overlapping workflows, resulting in a complex network of non-linear experiences.
•
Rapid expansion of the business also required quicker delivery of numerous features across multiple domains and squads in a short timeframe. On top of that, solutions demanded significant product design effort and resource.
Examples of journeys
Customer Support Agents
Processing sales over the phone
Managing account details
Assessing account and policy history
Claim Handlers
Registering claims
Calculating payouts
Assessing policy and claim history
Scaling for growth and instability
•
Business objectives grew as the business expanded, and so were domains, regions, squads with their micro-cultures and workflows.
•
Market instability in the post-pandemic era further contributed to shifting priorities, making it increasingly challenging to maintain a consistent design vision.
•
The system needed to be able to adapt rapidly without sacrificing clarity or momentum of both users and collaborators.
Migrating to a new VUE framework
•
Due to limitations of the legacy custom front-end framework, the business began migrating to an open-sourced framework to enable faster feature implementation and interaction capabilities.
•
As each framework has their own usability quirks, the migration risked significant UI/UX changes that could disrupt processes of users and collaborators.
Designer-collaborator ratio and capacity constraints
•
As I was the sole designer working on back-of-house tooling, supporting multiple domains often created bottlenecks.
•
Dual-track agile development and prioritisation improved output speed but also introduced other challenges in aligning outcomes consistently across domains.
Mental switching
•
Designing support systems presented a challenge in ensuring solutions were holistic, which required frequent mental switching between perspectives of internal users, collaborators and customers. The mental switching required to address both frequent users (e.g. back-of-house systems users) and infrequent users (e.g. customers) can unintentionally transfer learnability-focused biases into designs where usability is more critical.
•
Balancing these perspectives was crucial in determining usability and learnability gaps.
Anticipatory design reduces task time
Hypothesis
Proactively surfacing relevant information based on user intent helps users complete tasks more efficiently.
Prediction
Users complete tasks faster and more confidently, leading to shorter phone and offline assessments, and reduced claim cycle times.
Predictable structures speed up comprehension
Hypothesis
Clear information architecture and consistent interaction patterns support quicker understanding and navigation.
Prediction
Users resolve queries faster by confidently navigating through predictable layouts and flows.
Modular systems enable efficient scaling
Hypothesis
Modular design supports scalable growth and easier feature integration without major redesigns.
Prediction
Collaborators adapt more easily to product changes, reducing delivery friction and enabling plug-and-play enhancements.
Clear design intent supports autonomous decisions
Hypothesis
Clear, well-defined design intent can empower collaborators to make autonomous, design-informed decisions.
Prediction
Greater design autonomy reduces dependency on design oversight, improving efficiency across teams.
Familiar patterns increase user confidence
Hypothesis
Designs that align with common mental models improve onboarding and task fluency.
Prediction
Both users and collaborators feel more confident and self-reliant, leading to fewer errors and smoother workflows.
Design intent varies by user role and frequency
Hypothesis
Frequent users prioritise usability and speed, while infrequent users need intuitive guidance and clear learnability.
Prediction
By aligning design intent to user frequency, the product supports both expert efficiency and safe, confident use by occasional users.
Spotlighted interaction design principles
For further reading on interaction design principles mentioned, check out First Principles of Interaction Design by Bruce Tognazzini.
Interaction structures
Interaction structures were further defined in two dimensions: Horizontal and Vertical.
Horizontal interaction
Horizontal interaction represented the flow of task types. From left to right — Search → Assessment → Transaction.
Vertical interaction
Vertical structures mapped interface elevation to categories of actions. From bottom to top — Assessment ↑ Single-task ↑ Support.
Information structures
Content models were used to communicate core entities and their hierarchical relationships (e.g., Customer, Subscription, Pet, Coverage), serving as both design and alignment references across engineering domains.
They played a key role in evolving the platform from legacy limitations, such as single-policy accounts to multi-policy accounts.
Here are two examples to demonstrate how they were used, from simple to more granular and complex relationships.
Example A: Simple example of an information structure for a customer.
Entity relationships communicated:
A customer
An insurance cover
A pet
Example B: In-depth example of a customer's information structure for a Customer Support Agent.
Entity relationships communicated:
A customer
A subscription
A pet
A condition
A product
A coverage
Entity containers and navigation
Entity containers were used as a way to simplify UI/UX decisions and provide a consistent and reusable UI structure for assessing and navigating information.
Entity actions were consistently placed in the top-right corner of containers to ensure familiarity and predictability for both users and collaborators.
Impact
Spotlighted ID principles applied here
Anatomy (Animated demo)
Example of nested containers using a fixed information and action structure.
Containers are split into two sections: 'Header' and 'Body':
Navigation
This is an example of an entity container of a customer with two pets. Entity actions serve as entry points for users to quickly navigate and filter content of another page.
Entry points
Queries
View all claims ➜
View timeline of all events ➜
View documents ➜
View all claims of Pet B ➜
View timeline of Pet B ➜
View documents related to Pet B ➜
Page content
Claims tab
Pre-filtered claims
Timeline tab
Pre-filtered timeline
Documents tab
Pre-filtered documents
Information icebergs
Information icebergs was a method of simplifying complex information and visual clutter by prioritising meaningful information first in a hierarchy structure. This provides the user more control of information density, progressively disclosing each level without overwhelming.
Spotlighted ID principles applied here
Anticipation, Efficiency of the user, Fitt's Law, Simplicity
Example: Payment breakdown (Interactive demo)
Click on each line item to show/hide more information
Anticipatory inputs
Anticipatory inputs reinforced a user-first approach by recommending repetitive inputs or pre-filling where possible.
By anticipating and providing smarter defaults, form input and data inconsistency errors were minimised, reducing time spent on claim registration and assessments.
Spotlighted ID principles applied here
Anticipation, Consistency, Defaults
Example: Search input (Animated demo)
It was understood that customers would likely have no more than two vets, so it was worth recommending previous vet entries while providing the ability to search for other vets.
In this example, when a user initially clicks on the Vet search input, a dropdown menu with previously-added vet options drops down to facilitate quicker data entry.
Cascading changes
By cascading changes, it reduced user effort and registration inconsistencies by providing additional control to data updates when needed within a single journey.
This approach leveraged the principle of anticipation by simplifying workflows through offering meaningful options at the right time.
Reduction in repetition of form filling using previously-saved inputs, and more importantly, further additional assessments if data was inconsistent. Users were also able to stay in a single workflow when handling registrations over the phone, without exiting and performing another input workflow.
Spotlighted ID principles applied here
Anticipation, Consistency, Efficiency of the user
Cascading example (Animated demo)
The postcodes of the customer and their pet are usually the same but there can be instances where they differ, which also meant there was more than one entry point in editing postcodes.
In this example, when the user edits a customer or pet's postcode, they are prompted to determine if the postcode applies to the other entity too.
Reflecting on this experience as a sole designer supporting back-of-house features across multiple domains, my biggest takeaway was the value of uncovering resilient and adaptable design strategies. In the initial years, a significant amount of time was spent addressing UX debt caused by tactical workarounds driven by short-term business objectives.
Working in a fast-paced, agile environment often meant features were released before foundational alignment could be established. This led to unintended consequences like users and teams inventing their own workarounds, which introduced inconsistencies across the experience. Embedding foundational structures earlier could have reduced UX side effects.
Approaches were constantly ‘work-in-progress’ and adapting to new complexities. Insights learnt from tactical projects were used to validate hypotheses. Not everything could be solved immediately, and some ideas needed time to mature.
Is it adaptable or flexible to change? Is it scalable?
What are it's dependencies and can/should they be decoupled?
It was also important to maintain UI/UX consistency and reduce design oversight by fostering a culture of shared ownership, empowering collaborators to make confident, design-informed decisions.
Designing with predictability became a key guiding principle for users and collaborators. Reducing reliance on design oversight meant fostering shared ownership with collaborators, giving them the confidence and tools to make design-informed decisions on their own.
For high-frequency users, usability is the key. And for the rest, learnability is key. For collaborators, it's important to understand both types of users. To help them manage this mental switch easily, I leaned on and referenced the mental framework from Kate Kaplan’s about designing for complex systems.
This case study documents an iterative long-term vision. It’s not a polished artefact but a living body of work built over five years. Thanks for reading this far.