Coles damage assessment 

My role

user experience design (lead)


Firemark Labs (IAG)


Coles Insurance Australia

The challenge

IAG customers were often under-supported in their efforts to equip themselves with enough knowledge to decide whether to undertake a claim process. Working together with Coles Insurance, Firemark labs tackles this challenge with AI. 

The outcome

Ongoing efforts are being made to integrate our AI-powered car damage assessment capability within Coles claim lodgment beta site. An automatic damage assessment enables their customers to make an inform choice should they make the claim or not.

Problem statement

"It was too much of a headache, so I decided not to claim"  


When Samantha's car was damaged after a minor hit-and-run in a carpark, she cautiously weighed in whether to lodge a claim. She wondered if her premiums might increase if she made that claim so she paid out-of- pocket for the repairs instead amidst the confusion. These were one of the key pain points we gathered from IAG's Customer Experience team's 18 months long research on mapping the short-tail claim experience for customers. From the research, we believe successful claim outcomes often depended on customers preparedness for the process and knowing the insurance company's requirements but often customers lack support and assurance from the company to achieve this.

Knowing that our AI lab can help, my team partnered with Coles Insurance, one of the many insurance providers underwritten by IAG, to develop a car damage assessment capability within the Coles claim lodgment website.

Screen Shot 2019-01-03 at 10.17.54 am

Participants of the research  conducted by IAG CX team


Our evolving customer journey map

Right: the user flow

Coles user flow

Our solution

Automatic car damage assessment provides a frictionless experience to reduce uncertainty and worry during a stressful time


In the event of a light car accident, customers just have to snap their car damages and upload the images, all done through the mobile site. The image classification AI within the app compares the customer’s photos with thousands of other anonymized crash photos to generate a cost estimate for their repair.  The customer can decide whether to lodge the claim or not without having to call the call centre. 

02_Damage Assessment_full@1.5x
15_Damage Assessment@1.5x
05_Damage Assessment_full@2x
17_Damage Assessment@1.5x
06_Damage Assessment_full@1.5x
20_Damage Assessment@1.5x
07_Take photos@1.5x

Coles design system


As a lab's POC, the damage assessment capability is built on top of Coles' simple claim lodgment site (closed Beta stage). I worked together with the Coles team to integrate the screen flows with the existing Coles' design pattern for continuity and to reduce cognitive load. The site is especially designed with focus on mobile devices and has undergone testing with customers. 

Screen Shot 2019-01-04 at 10.12.20 am
Screen Shot 2019-01-04 at 10.12.06 am


AI limitations and maturity expectations


As with any AI products, we have to be realistic about our image classification model's  current capabilities and limitations. The accuracy level of algorithms is dependent on the data training. Currently as an MVP, the damage assessment capability is trained to detect simple damages inflicted upon Toyota Carolla car make. More images and training are needed to achieve our north star product of a vehicle damage cost estimator app.

Our northstar: A standalone vehicle damage cost estimator app based on IAG price list for Australia market