Case Study

Keepeye
KPI Management Platform

Dojo lived from project to project and wanted a product of its own. I led the research, designed the platform, wrote part of the Angular front-end, and mentored a developer moving into design. Usability tests of the finished tool scored it 8.2 out of 10.

Keepeye KPI Management Platform interface

Company

Dojo Smart WaysInternal product · SaaS

Role

UX EngineerDesign + Front-End

Timeline

2023Built from scratch

Stack

Figma, AngularTailwind CSS

Background

A consultancy betting on its first product.

The Business Case

Dojo Smart Ways ran entirely on project work: revenue stopped whenever a project did. The goal set for Keepeye was explicit: create the company's first product and shift its focus from data projects to products.

It had a candidate: an internal metrics tool Dojo had already built for a client project, and that users kept responding well to. Keepeye was the bet to turn that tool into a SaaS.

The User Problem

The companies we spoke to had no single place to look at their KPIs. Excel files, Google Sheets, and scattered Power BI dashboards, each maintained by a different team, each calculating the same metric its own way.

When leadership asked for a number, the answer depended on who they asked. People spent meetings reconciling reports instead of acting on them.

Everyone had the data. Nobody trusted their version of it.

The goals

Two goals set before any screen.

01

Create the company's first product

Build a solution that could attract new clients and shift Dojo's focus from data projects to products, opening a recurring revenue line beside project work.

02

Automate the data

Replace manual data entry with a robust, easy-to-use platform: safer inputs, automated metrics, and a better day for the people who maintain them.

The process

One Double Diamond, four phases.

01

Discover

Brainstorming with directors
Proto-personas
Market benchmarking
Survey with 40+ professionals
Interviews + usability tests

02

Define

User journey
Word cloud
UX Canvas
RICE prioritization

03

Develop

Blue Squid design system
Wireframes
Low to high fidelity
Visual identity
Angular + Tailwind build

04

Deliver

MVP launch
Usability tests
Metrics monitoring
Feedback + roadmap

01 · Discover

Listening to the people who live in KPIs.

Understanding the problem from the inside out

The existing tool was a starting point, not an answer. Discovery opened with brainstorming sessions involving the dev team and Dojo's directors, connecting business expectations to the pains the product had to solve. The board captured objectives, client complaints, and the features people expected before a single screen existed.

Brainstorming Benchmarking
Brainstorming board from the discovery sessions

↑ The brainstorming board with the dev team and Dojo's directors: objectives, pains, and expectations

Personifying the user

With the product values discerned, proto-personas captured who we were designing for: production engineers turned managers, analysts living inside spreadsheets. They kept the expected user base concrete for the whole team through every later decision.

Proto-personas
Proto-persona Fernando
Proto-persona Juliana

↑ Proto-personas built from the discovery sessions

Product values defined in the brainstorms

aggregate

value, purpose, and practicality

connect

people, data, and goals

simplify

the everyday work around metrics

Validating what we heard at scale

Interviews give depth but can mislead, so we surveyed more than 40 professionals who use metrics tools day to day, asking about their management methods, current platforms, and where those platforms failed them. Then we ran interviews and usability sessions on the existing tool, watching real users hit the exact walls the survey described.

Survey Research Usability Testing
Remote interview and usability session on the existing KPI tool

↑ Interview and usability session on the existing KPI tool, run remotely with real users

Discover · What we learned

Three findings that redefined the product.

One metric, many versions

The most consistent finding: the same KPI produced different numbers depending on which system or export it came from. Not because formulas differed, but because the underlying data was never the same to begin with. The inconsistency was structural.

Built for analysts, not managers

Most KPI tools needed technical configuration or an analyst to produce a usable dashboard, so the managers who needed the numbers depended on someone else for the view. The opening was a tool a manager could own without SQL or a BI specialist.

Bad data leads to paralysis, not bad decisions

When professionals don't trust a number, they don't act on it, they investigate it. Meetings become reconciliation sessions and decisions wait. This reframed what we were building: not a prettier dashboard, but a single source of truth. The formula calculator came directly from this finding.

02 · Define

Turning research into decisions.

User Journey

During the usability tests we mapped every step and interaction users had with the tool, start to finish. The journey showed not just where people struggled, but where the product lost them entirely.

Word Cloud

The quantitative responses were distilled into a word cloud that surfaced the most repeated pains and needs. It kept the loudest problems in front of the team instead of buried in a spreadsheet.

UX Canvas

With the problems understood, I consolidated them into a UX Canvas: one document stating what the product's experience had to deliver. Every later decision was checked against it.

RICE Prioritization

With solutions drafted and development effort estimated, the RICE matrix ranked priorities analytically instead of intuitively. With four people and a fixed window, this decided what the MVP would and would not be.

03 · Develop

A design system first, then every screen.

Development started with Blue Squid, a design system created to standardize components and carry a visual identity that Dojo's future products could inherit. With the system and the information architecture defined, I wireframed every screen of the product.

Screens moved from low to medium to high fidelity, validated at each step. The brand rested on three pillars: security, robustness, and simplicity. I built the components and screens in TypeScript, Angular, and Tailwind, and we shipped the MVP for real-world use.

Keepeye wireframes and early exploration

↑ Wireframes covering every screen, drawn on top of Blue Squid and the information architecture

Keepeye component library

↑ Blue Squid components, later implemented in TypeScript, Angular, and Tailwind

03 · Develop · Visual identity

A brand built on three pillars

The logo construction started from concept pillars defined for the brand: security and robustness in how the product presents itself, aligned with the simplicity the tool offers the user. The identity runs through every screen, starting at login.

Keepeye login screen with the new visual identity

↑ The login screen carrying the new identity

The solution

Four features, each traced to a finding.

The RICE exercise shaped the feature set directly. What scored highest was not always what we expected, but the research consistently pointed toward the same needs: centralization, formula consistency, and managerial autonomy.

Monitoring

KPI Dashboards

All active KPIs in one view: status, trend, goal. The first screen answers "how are we doing?" without a single click. Drill-down exists, but the summary stands on its own.

Data Trust

Metric Formula Calculator

The highest-scoring feature in the RICE exercise, and the one that attacked the trust deficit directly. Each metric connects to a centralized data source with a single formula, so every department reads the same number. No diverging exports, no parallel spreadsheets.

Configuration

KPI and Metrics Management

Create, edit, categorize, and link KPIs to the metrics that feed them. Research showed teams track 15 to 40 active KPIs, each with different owners, cadences, and sources. The management view had to absorb that complexity without becoming a spreadsheet itself.

Governance

User Area and Access Control

Some users configure, some monitor, some only read. Roles, permissions, and data scope decide who sees floor metrics and who sees consolidated results. Role definition came early and shaped the information hierarchy of the entire product.

Keepeye dashboard interface
Keepeye product screens

↑ High-fidelity screens of the shipped product: dashboards, metric management, and user area

Keepeye home with KPI options
Keepeye dashboards
Keepeye metric management
Keepeye user area and access control

↑ Core screens of the MVP: home with KPI options, dashboards, metric management, and the user area

The role

I stayed for the implementation.

UX Engineering

After handoff I joined the build as a front-end developer, writing the design system components and screens in TypeScript, Angular, and Tailwind alongside the team. Designing and implementing the same product removes the translation layer: when a component didn't survive contact with real data, I adjusted the design at the source.

It also meant I understood constraints before they became blockers. When the team hit a component complexity problem, I was already in the room.

Mentoring in Transition

One developer on the team was moving from front-end into UI/UX design. I reviewed her work throughout the project, walked through the reasoning behind decisions, and gave structured feedback on her interface choices.

Explaining why a pattern works forces you to articulate principles you normally apply on instinct. It sharpened my own thinking as much as hers.

04 · Deliver

8.2 out of 10, and how we measured it.

Delivery didn't end at launch. We kept collecting: usability tests on the finished tool, key-metric monitoring, and a backlog of integration ideas straight from user feedback.

8.2/10 Average user satisfaction in the post-delivery usability tests, with minor fixes shipped from the findings
4 Team members, including one designer mentored through her career transition
MVP Launched as Dojo's first product, opening a product revenue line beside project work

Keepeye compressed the whole cycle into one project: research, prioritization, design, code, mentoring. It confirmed a rule I still work by: whatever you skip in discovery, you pay for in delivery.

More work

Multipedidos

Multipedidos

Senior Product Designer at a food platform serving 4M+ orders a month. Design system, consumer product, and AI-assisted workflows.

Product Design · Food Tech

IncludED

IncludED

Co-founder and Lead Product Designer at PixelPunk. Built a SaaS platform for inclusive education from zero to first paying customers.

Product Design · EdTech SaaS

Electrolux

Electrolux

Dashboard visual identity and design system for global markets, via Dojo Smart Ways.

Data Viz · Design System