Reimagine the Way to Read Research Report
- Company
- Flashpoint.AI
- Duration
- Feb 2026 (2 weeks)
- Team
- Solo
- Role
- Design Engineer — Audit, UI/UX Design & FE Dev
Flashpoint.AI runs market research end to end. The report is the last mile — the artifact a client takes to a decision. It was also a weak point in Flashpoint.AI's product. I audited it, redesigned it, and shipped the rebuild in two weeks.
Some content shown in this project — screens, data, and events — is simulated for demonstration purposes. It does not reflect real product output or actual events, and shouldn't be taken as such.
The Last Mile Was the Weakest Link
One of Flashpoint.AI's main user groups is people who commissioned research because they need the insights and decide something right away. Speed is one of the platform's core promises. A report that takes hours to parse breaks that promise at the most important step.
“The problem: unclear, disorganized, and hard to read.”
Design Director, on the original report
Less Noise, More Decision-Ready
I redesigned the report from a flat list the reader had to work through into a structure built for fast insights and decisions, with titled insights you can skim, evidence one click away for every finding, and recommended actions closing out the page.
Built for quick insights
A list of insights, each with a summary title for quick skim and the full finding underneath, linked back to its source.
Insights in chart view
Visualize insights with a chart from its original source.
Ready when you are
Recommend action for your next steps, with each suggestion backed by the research evidence — ready to become the reader's next proposal.
What the Audit Was Looking For
I opened the report cold, with a decision to make, and audited it against a simple bar: immediate insight you can act on, backed by real data from the Survey and GenR&D tools, not just confident phrasing, and understood in a quick glance. The original report missed on every count.
Audit 01 — Information Hierarchy
First glance of the overview didn't answer "what did we learn"
The first audit focuses on whether you can access insights right away. Looking only at the first page, four findings point to the same failure: inefficient information hierarchy and reading flow.
Audit 02 — The Model's Output
Not everything is worth reading
LLM-generated content could get overwhelming and redundant, so at the second part of the audit I identified which parts of the model's output were worth surfacing and which should stay hidden.
What the Redesign Had to Satisfy
Speed to insight
User goal
Understand what the research found, right away.
Business goal
Cut time-to-decision so clients act in the same sitting.
Product goal
Surface and group findings by relevance to identify the piece you need faster.
Trust in the report
User goal
Know where every finding and number comes from.
Business goal
Reduce disputed findings and support tickets.
Product goal
Attach traceable evidence to every finding, so the numbers and conclusions appear with a visible source.
Insights to Action
User goal
Walk away with a next step, not just a list of findings.
Business goal
Make this the report clients rely on and return to.
Product goal
Convert findings into prioritized, actionable recommendations, not a flat list of observations.
The Final Report Experience
I redesigned the report from a document that summarizes research into a decision-making tool. Information is structured to be easier to consume, evidence is available whenever needed, and recommendations support teams turn research into action.
Five Versions to Find the Structure
I ideated in Figma with dummy copy for quick explorations and make sure the structure works with "terrible" writing. Only once the skeleton held, I move to code.
- 1
Flat bullets, jargon columns — the problem restated
- 2
Hierarchy, grouped actions, confidence gauges
- 3
Provenance and topic drill-downs
- 4
Titled insights, sample framing in the header
- 5
Chart view and comparative analysis
Interested in working together?
Let's talk about your project.
