Overview Problem The Work Process Decisions Impact
Noah Koshy
Designer · Sydney
Case № 05 Industry

MRI Benchmark

Scouted by a co-founder of MRI Benchmark, I redesigned the startup's flagship assessment in 4 months instead of 6, then built a custom GPT and a Replit web platform to help its business clients act on their results.

RoleCreative Designer
Team4 people (Sydney)
Timeline
I ownedMRI Benchmark Platform Assessment & Re-Design, & AI Tool Workflow Integration
The sixty-second version
Problem
The startup's flagship assessment was a predominantly text-heavy survey, leading to drop-off and formulaic or insincere answers from the business clients completing it.
What I did
I redesigned the assessment around multimedia elements to reduce the cognitive load on clients, then built a custom GPT and a Replit web platform to help them understand and act on their results.
Result
I finished the assessment redesign, initially a 6-month task, in 4 months, which led to my role being extended and my scope expanding well beyond the original brief.
A co-founder scouted me
after working alongside him as a student at the Apple Foundation Program @UTS
4 months
to finish the assessment redesign, initially a 6-month task
2 recommendations
received on LinkedIn for my work in this role
↓ More details below if you want the reasoning.
§ 01

The Problem

A predominantly text-heavy assessment, leading to drop-off and formulaic or insincere answers from business clients.
"You will need to own the design process, as you will primarily be the sole designer on this work."
What I couldn't change
01 The assessment still had to do its original job, gathering results and presenting them intuitively, so that the business could then act on them.
02 Weekly stand-ups with executive leadership, which set the pace of work and the expectations of the role. The short decision cycles of a startup, presenting my thinking and findings at a faster pace than I was previously used to.
03 My role involved a level of technical consideration, due to the multi-hat skill requirement of a startup environment.
§ 02

The Work

Three pieces of work from the eight months, from the assessment through to the tools built around it. As none of this work can be shown directly, each piece has been described here instead.
My role was initially to focus on making the experience of the MRI Benchmark Assessment a more engaging and interactive process, introducing multimedia elements to support this, rather than the predominantly text-heavy survey used previously.
The redesign prioritised reducing the cognitive load on users, as well as utilising multimedia elements and implementing quality-of-life improvements to enhance the experience as a whole.
This improved experience sought to address drop-off and formulaic or insincere answers, thereby contributing to the improved effectiveness and usability of the results gathered.
Assessment Redesign
For the business clients completing the assessment.
Completed in 4 months, against an initial 6-month timeline.
My next task was creating a companion solution, to support the understanding and benefit that business clients would get through the survey results, as well as help them understand and take next action steps.
The custom GPT built would take the results generated into a report format as an input, thoroughly analysing it and drawing the most prominent findings from it.
It would also propose next steps that the business could take, based on the areas assessed as being valuable for improvement or their current strengths.
Custom GPT
Built through ChatGPT, for each client's report.
Replit was used to create a web platform solution, which would automatically integrate the custom GPT developed, taking the survey report data and generating summarised and synthesised findings from it.
Additionally, unlike the custom GPT which was limited to text output, the solution built in Replit generated graphs and other visual tools to support the information derived.
Web Platform
Built in Replit, integrating the custom GPT.
§ 03

How I got there

The approach behind the work, from start to finish.
Stage 01 · Understand
The first step in redesigning the assessment was to thoroughly understand its purpose and the benefit it brought to business clients, as well as any pain points that were faced as part of it.
Stage 02 · Research & Test
In addition to conducting user research, I completed usability testing by taking the assessment myself to form my own opinions and perspective.
Stage 03 · Present
I would present my findings, recommendations and work to executive leadership in weekly stand-up meetings.
Dead end
The custom GPT on its own. It was limited to text output.
Kept
Reducing the cognitive load on the people completing it.
…then the scope kept growing.
Keep scrolling — the wall moves sideways
§ 04 — Decisions & trade-offs

A better experience for the people completing the assessment would mean better results for the businesses using it.

Three decisions, from testing the assessment through to building the platform.
01
/ 03
Decision 01

Taking the assessment myself

Chose
Completed the assessment myself, alongside conducting user research.
Gave up
Some objectivity, as my own experience was only one perspective.
Why
To analyse the current assessment, identifying areas for innovation.
Decision 02

Introducing multimedia elements

Chose
Introduced multimedia elements such as videos, animations, or audio, along with quality-of-life improvements, in place of the predominantly text-heavy survey.
Gave up
The simplicity of a text-only survey, which is quicker to build and update.
Why
To improve the effectiveness and usability of the results gathered.
Decision 03

Building a web platform in Replit

Chose
Built a web platform in Replit that allowed integrating the previously created custom GPT.
Gave up
Keeping it to the custom GPT alone, which was quicker to set up.
Why
It helped business clients to understand their results and take next action steps, rather than only reading a report. It also introduced branching paths, to begin tailoring outputs to individual business circumstances and requirements.
§ 05

Impact & what I'd change

A broad range of experience in a startup environment.
Looking back

My time in this role led to new opportunities to test tools, develop skills and learn and grow further. I also received two LinkedIn recommendations, due to the high standard of work that I produced. If I did it again, I would measure drop-off before and after the redesign, to validate the approach more thoroughly.

— Noah
Startup · Creative Designer Scouted by a Co-Founder & Contract Role Extended Figma, ChatGPT, Replit
Next case
Case № 06 →
Telstra - Business Experience Specialist
All projects →
© Noah Koshy, 2022-2026.
MRI Benchmark · Scrapbook-styled Portfolio