The tool nobody at the top opened
Every layer below the CXO was already served. He was the one nobody was designing for.
Practice — Executive product design · Research under access constraints · Native app design
Role — Led the design engagement — direction, and the four calls that follow
Team — ~12 — design, front-end, AI/ML, GenAI, data science — with the client’s US and India teams
Period — September 2024 – July 2025
Client — A global alco-beverage leader · ten verticals, fifty countries
Designed for the user nobody was designing for.
A reporting director spends three weeks assembling the competitor picture. It lands as a quarterly deck. By then the market has moved, and if a leader has a follow-up question it goes back down the same chain.
Twenty hours per report. Two to three weeks to land. Repeated every quarter, duplicated in every vertical. Fractal’s published figure for what the product returned is around 200 hours per cycle — a different measure from the twenty, and both correct.
The analysts were served. The layers below the top were served. The CXO never opened the tool, because it answered questions he was not asking.
The sponsor was Strategy & Technology. The first audience was the CEO’s office. And the constraint was the user himself: a CXO has minutes, not meetings — and once research was cut, almost no access at all.
The rules we designed by
The user is the one with the least time, not the most requests.
Data is the hero; the interface is not.
Built for them, not issued to them.
What we did
01 · Negotiated research back in.
The sponsor cut user research — “I can speak for my leaders.” I named the adoption risk and traded down to four immersion sessions on how leaders consume, not what they would request. It cost a month and bought the use-case map the whole product was validated against.
02 · Traded diagrams for screens.
Information-architecture reviews drew polite, unusable feedback; flowcharts meant nothing to the sponsor. We folded the IA into low-fi wireframes co-created live, then carried real quarterly data into med-fi rather than lorem ipsum. Instead of defending the diagram, and the process.
03 · Let a user need break the scope.
Immersions surfaced the adoption driver: push notifications. Leaders do not form app habits — the app has to call them. I backed a platform change mid-project, PWA to native iOS and Android. Instead of holding scope and protecting the timeline.
04 · Refused the house style.
Every internal tool wore the corporate primary palette. We used the secondary — muted, so data stays the hero; distinct, so it reads as built for leadership rather than issued to it. Two phased releases let executives judge a working product early.
Project highlights
The challenge — Served everyone but him
Competitor intelligence assembled by hand every quarter, landing weeks late as a deck. The tool built to fix it answered the analyst’s questions, so the executive it was meant for never opened it.
The approach — Design for minutes
Four immersion sessions on how leaders consume information, not what they would ask for. Then a native app that calls the user, in a palette that keeps the data the hero.
The outcome — Opened, not just shipped
Ten verticals self-serve. Six dependent steps compressed to three. Weeks to intelligence became minutes. Introduced through the CEO’s office and spread by pull, not mandate.
Gallery
Before and after — the quarterly report they were sent, and the screen they open instead.
Wireframes carrying real data — med-fi ran on real quarterly figures, so feedback landed on substance.
The secondary palette — muted so data stays the hero, distinct so it reads as theirs.
Six steps to three — the journey, compressed.
Results — “Spread by pull, not mandate.”
What changed for the user
The CXO opens it. Share prices, flag-chart heatmaps, eight-quarter evolution and deep dives sit in one place, with AI summaries and the sources one tap away. Minutes to intelligence, from weeks.
What it was worth
≈200 hours returned per cycle — Fractal’s published figure.
10 verticals self-serve. 6 → 3 dependent steps. Introduced through the CEO’s office, then adopted by pull rather than mandate — which, for an executive tool, is the adoption measure that matters.
The honest line
The research standoff cost a month. The first native build meant correcting feasibility estimates mid-flight. Both lessons travelled into later engagements.
≈200 hours returned per cycle. Ten verticals. Fifty countries.