Driving aged care reform with a new five star rating

Driving aged care reform with a new five star rating

Driving aged care reform with a new five star rating

Building on the existing rating system, I led the design and front-end build of a new five star rating system that would provide consumers with greater visibility of data for over 2,700 Residential Aged Care providers in Australia.

Building on the existing rating system, I led the design and front-end build of a new five star rating system that would provide consumers with greater visibility of data for over 2,700 Residential Aged Care providers in Australia.

Building on the existing rating system, I led the design and front-end build of a new five star rating system that would provide consumers with greater visibility of data for over 2,700 Residential Aged Care providers in Australia.

Scroll to read case study

Scroll to read case study

During Covid-19 citywide shutdowns, we were given the opportunity to deliver a new star rating system for Residential Aged Care providers with a 12 week deadline.

6 weeks to prepare the designs.
6 weeks of test, refine and build.

Leveraging precedence and subject matter experts, with insight into industry challenges, allowed us to produce designs at pace ready for user testing.

This project stage was crucial to ensure a successful experience for the go-live.

The client flagged one critical risk.

Legal action by aged care organisations, if they received low star ratings due to our algorithm design, ultimately resulting in revenue loss.

As a response, our team strategically engaged with all user groups to give a voice in co-designing the algorithm.

The designed experience has the power to instigate an industry-wide reform.

Defining what success looks like

Defining success

Project owner

successful system implementation

The star rating system is:
designed and prototype built; tested with both residents and providers; adopted and supported by the industry.

Aged care Provider

System adoption

The star rating system is:
fair in its representation of my provider; sensitive to the context and current cohort; clear on how to improve on the star rating.

data Policy owners

increase in data veracity

The star rating system is:
based on existing data;
encourages providers to report data;
improves overall data quality.

Future Aged care RESIDENT

useful to compare

The star rating system is:
easy to understand;
allows me to compare providers fairly;
informs me with relevant knowledge.

Crafting the user testing cohort

7

Aged Care Provider
organisation leaders

22

people seeking options of
residential aged care

User testing cohort

7

Aged Care Providers

People who provide leadership
within an aged care organisation  

6

Individuals

People exploring aged care options
for themselves 

7

Support Network

People exploring aged care options
for their loved one

5

Newly Decided

People who have secured aged care
in the last 6 months

4

Care Navigators

People who provide assistance
for those who require aged care support

Building the scope for user testing

NEW sitemap

Key screens to be designed:
Search, Provider page, Compare, Calculations

Previous design

Existing banner with existing dot rating

NEW tab navigation

new rating:
From four dots to five stars

new provider page

Overall star rating with four subcategory star ratings

before

The original Service Compliance Rating is now one of the four subcategories in the new star rating system.

After

A provider’s profile now displays a rating overview with the ability to dive deeper into each individual subcategory.

Key final wireframes

search provider

star ratings calculations

provider profile

compare providers

provider subcategory rating

agile Design methodology

Hypothesize → Design → Test → Refine

Hypothesize → Design →
Test → Refine

Guiding principles

How do we design a trustworthy
government-based five star rating system?

fair

Base the rating on multiple facets of quality, using existing qualitative and quantitative data.

transparent

Educate users around how to objectively interpret, evaluate and make use of the available quality of care data.

ACTIONABLE

Provide technical explanations, where appropriate, as to how the ratings are derived and areas that can be improved.
Emphasise that the star ratings does not replace the need to personally contact and visit care providers.

Provide technical explanations, where appropriate, as to how the ratings are derived and areas that can be improved. Emphasise that the star ratings does not replace the need to personally contact and visit care providers.

LIMITATIONS WITH DATA

dealing with data integrity

We discovered that many consumer’s lacked trust with existing data. To build trust, we solicited inputs from experts familiar with the data's veracity and reliability to allocate fair subcategory weightings on how much they would contribute to the overall star rating.

A weighting methodology builds a contingency for inflated data and future proofs the system as it can be adjusted alongside data maturity.

Project challenges

No access to the
existing design system

We built our own Figma styles library through inspecting the live website. It was important that the prototype designed was consistent and familiar for the user testing sessions. As this was a rapid project, we built components as required, rather than starting off with fully functional design system.

Project challenges

No access to the
existing design system

We built our own Figma styles library through inspecting the live website. It was important that the prototype designed was consistent and familiar for the user testing sessions. As this was a rapid project, we built components as required, rather than starting off with fully functional design system.

EXISTING TYPEFACE

Calibre by Klim Type Foundry

Calibre by Klim Type Foundry

ABCDEFGHIJKLMLNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
1234567890!@#$%^&*()

This is a Heading 1

This is a Heading 1

60pt

60pt

64pt

64pt

Medium

Medium

This is a Heading 2

This is a Heading 2

40pt

40pt

44pt

44pt

Medium

Medium

This is a Heading 3

This is a Heading 3

32pt

32pt

40pt

40pt

Medium

Medium

This is a Heading 3

This is a Heading 3

24pt

24pt

32pt

32pt

Medium

Medium

This is a Body

This is a Body

20pt

20pt

28pt

28pt

Regular

Regular

This is a Description

This is a Description

16pt

16pt

22pt

22pt

Regular

Regular

existing 8 pixel grid spacing

Existing Colour themes

Navigation theme

Primary
Purple

Primary
Purple

#42175F

#42175F

Primary
Black

Primary
Black

#282828

#282828

Black–80

Black–80

#5a5a5a

#5a5a5a

Black–70

Black–70

#6F6F70

#6F6F70

Black–60

Black–60

#959595

#959595

Black–40

Black–40

#acacac

#acacac

Black–20

Black–20

#e1e1e1

#e1e1e1

Black–10

Black–10

#f6f6f6

#f6f6f6

Home page theme

Primary

Blue

Primary

Blue

#FFB600

#FFB600

Secondary
Blue

Secondary
Blue

#1E4387

#1E4387

Blue–60

Blue–60

#305498

#305498

Blue–40

Blue–40

#849ECF

#849ECF

Blue–20

Blue–20

#CBD7EF

#CBD7EF

Blue–10

Blue–10

#F2F6FF

#F2F6FF

Provider profile theme

Primary

Green

Primary

Green

#012d18

#012d18

Secondary

Green

Secondary

Green

#03703c

#03703c

Green–60

Green–60

#317555

#317555

Green–40

Green–40

#67a98a

#67a98a

Green–20

Green–20

#bfdbce

#bfdbce

Green–10

Green–10

#eef6f3

#eef6f3

BEFORE

COMPARISON BY Location

We visualised the provider distribution on a stacked bar graph, highlighting how the provider scored compared with their city’s and national averages.

When we received the data sets, we discovered that the attribute of location was not available. On top of that, users commented that the diagram made the low rated providers look like terrible choices.

Despite scrapping the diagram early into testing, we continued to hear the need to compare providers in a setting that was relevant.

User participant

“What if the only provider in my
community was a 3 star rating?”

“What if the only provider in my community was a 3 star rating?”

AFTER

COMPARING BY LOCALITY CLASS

After speaking to many users, we heard the importance of being able to compare like-for-like providers. We discovered the Modified Monash Model classification system, that classifies based on locality types. By filtering the diagram with this system, we can empower consumers to be equipped with the relevant knowledge to make the best decision in their circumstance.

Weekly interim findings report

BEFORE

User sentiment (Item GF 7.0)

Users want relevant labels and description of what each rating means on the Search Results page (e.g. 5 = Excellent etc.)

AFTER

An introduction to star ratings

Users want relevant labels and description of what each rating means on the Search Results page (e.g. 5 = Excellent etc.)

label 1.0

Confidence

Our first set of labels described the level of confidence we had in the data set. This is because we discovered a lot of providers were missing some subcategory data.

Users were confused as to what 'confidence' referred to.

User participant

“Does high confidence mean
the provider is good or bad?”

“Does high confidence mean the provider is good or bad?”

As a result we changed the rating labels and dealt with the issue of missing data sets by implementing an overall star rating cap based on what is missing.

label 2.0

Expectation

For the next iteration of labels, we pivoted to more user-friendly language. However, it was perceived as too casual and interpreted as consumer reviews rather than government requirements.

label 3.0

Standards

The next iteration of labels, we used the language of standards to plainly depict their compliance. However, users considered this very strong language and suggested that they would not settle for anything less than 4 star rating, even though for some consumers that is their only option.

FINAL EXECUTION

Acceptable

The set of labels we landed on was somewhere in-between. It doesn’t oversell 5 stars nor scares users away from 2 stars. It was important that we used the language of acceptable for 3 stars to re-emphasise the baseline that was created.

iteration 1.0

Case-mix adjusted values

Our initial design displayed case-mix adjusted values. However, most users criticised how the Care Minute value reported never reflected the actual perceivable amount of care received as users thought it was displaying true minutes.

iteration 2.0

% tracked to care minute target

Instead of showing the case-mix adjusted value, we opted to display the percentage tracked to the provider’s target. The challenge with this iteration is that every provider has a different target. This iteration was the least receptive to users as it lacked transparency and it was confusing how they could not see any indication of minutes calculated.

ITERATION 1.0

Final ITERATION 3.0

case-mix adjusted values with target label

One thing we learnt through the iterations was both the importance of displaying the right metric and how the metric was perceived. The final design reverted back to showing case-mix adjusted values, supplemented with explanations in the labels and body text to support the user with the correct interpretation of the figures.

CALCULATIONS

How is the care minute rating calculated?

On the calculations page, we used the % tracked to the target as the metric to emphasise that this is how the star ratings are calculated.

Designing for edge cases

As part of designing for industry-wide adoption, edge cases were also designed and tested to ensure that all providers could be properly and fairly represented.

words from a user testing participant

words from a user testing participant

“Pass onto the team my congratulations for the work they have completed and more importantly for me, their respect and demonstration they have listened.


I am so very, very pleased and do feel very confident, as a consumer, I will be able to make an informed decision for future aged care services.”

“Pass onto the team my congratulations for the work they have completed and more importantly for me, their respect and demonstration they have listened.

I am so very, very pleased and do feel very confident, as a consumer,
I will be able to make an informed decision for future aged care services.”

What the team said about me

Anna Wilkinson

UX testinglead

Nathan has fantastic technical skills. He is efficient, organised and can quickly interpret complex information and present it in a way that is clear and simple for end users. Nathan has presented extremely well multiple times to our clients and has built a trusted relationship with them over time.

Chris Ong

UX/UI stream lead

Nathan consistently demonstrates technical, communication and leadership skills far above his grade. He developed high quality user interface designs, presented his design work to senior executives at the Federal Department of Health, and led other members of the development team, taking initiative where necessary and required little to no oversight in most situations.

Nick Warren

Project lead partner

Nathan was engaged, effective, brought new ideas to the team and began his leadership journey in working with our vacationer. Really pleased with the way you worked and we’ve had great feedback on the video too!

James Martin

Salesforce developer lead

Nathan was excellent throughout our project. He owned solutions and was instrumental in ensuring the projects success. He showed great initiative in working with the rest of the team, ensuring everyone worked together to deliver.

Next steps

Project takeaways

Although we successfully achieved the deliverables and business outcomes set out by our client in this phase, here are some tasks I would undertake with additional budget and time.

re-test the design
with participants

We only tested with each participant once. Because of the project timeframe, we had to rapidly iterate the design which consequently resulted in users of different testing weeks seeing different designs. I would love to go back to all participants and show them the iterations and further test whether some of their hypothesis or concerns were resolved by the opinions or ideas they shared.

For more rigorous testing, I would try engaging with a greater variety of participants to build personas such that the star rating representation can provide more value: people of different cultures, people with English as a second language, people of disability, people from rural areas, people with only access to one care provider. etc.

re-TEST the experience
once connected to live data

The prototype was only connected to a small set of
de-identified data. This meant we were unsure as to what the actual distribution of providers on star ratings would be.

The prototype was only connected to a small set of de-identified data. This meant we were unsure as to what the actual distribution of providers on star ratings would be.

With live data we could re-test the star rating implementation for providers with incomplete data sets and uncover other missed edge cases to round out the experience.

With live data we could re-test the star rating implementation for providers with incomplete data sets and uncover other missed edge cases to round out the experience.

The prototype was only connected to a small set of de-identified data. This meant we were unsure as to what the actual distribution of providers on star ratings would be.

With live data we could re-test the star rating implementation for providers with incomplete data sets and uncover other missed edge cases to round out the experience.

building a product roadmap
of supporting interfaces

Although this project was initiated as part of a greater industry-wide aged care reform for Australia, we only designed and tested the interface for end users.

Now that the project has gone live, it would be productive to revisit the project and see what other supplementary interfaces could leverage this system. One example being an organisation portal that could show all the providers an organisation owns, the provider star ratings, the ability to manage and track data reporting timelines, a view on how providers compare with others in the area, recommendations and steps to improve star rating. etc.