Kuebiko

Helping teams navigate, transform, and trust their data, with or without SQL

Desktop app

Lead designer

Timeline
Mar 2023 - Oct 2024
‍
Team
CTO, 2 Engineers, and myself

Highlights
‍Used by dataPlor's global data operations team, handles +150 million POI’s, it's 200 times faster than its predecessor tool.

Main screen of Kuebiko's data manipulation tool with SQL tool

Main screen of Kuebiko's data manipulation screen with SQL tool

Overview

Dataplor's mission was to deliver high-quality, human and AI-verified Point of Interest (POI) data to the global marketplace, with kuebiko we elevated data verification rates from 30 to 200 monthly datasets (that's a lot of new POIs = a lot of more revenue).

The Problem

Internal teams were stuck between tools. Some users relied on SQL editors to extract and manipulate data, while others needed simpler, UI-driven ways to explore and verify information. There was no single space that supported both workflows cleanly, which made collaboration and iteration harder than it needed to be.

AdminNext vs Kuebiko

Approach & Research

Through shadowing sessions and detailed interviews, we identified critical workflow blockers and behavior patterns across roles. Surveys revealed that 70% of users were dissatisfied with Adminext's performance, and 30% found the overall experience frustrating or confusing.

I interviewed 8 users (4 technical, 4 non-technical) and ran two internal workshops. We also benchmarked task speed across available tools. This helped define two key workflows: SQL-based querying and UI-driven validation. From there, we focused on building a shared environment that supported both.

Maria user persona

Jack user persona

How do you support both tech-savvy users who need flexibility, and less technical users who need structure, clarity, and guardrails?

Jack's user journey

Pain points

Scattered tools, fragmented workflows

Teams had to juggle spreadsheets, SQL editors, dashboards, and manual scripts. It slowed them down and made collaboration messy.

Gaps between user types

Non-technical users struggled to validate data. Technical users spent extra time cleaning or translating issues instead of moving forward.

Gaps between user types

No shared space existed to track what had changed or why. Teams couldn’t trace data confidently, leading to errors and delays.

What I Did

Audited the legacy tool (Adminext) to understand pain points in data validation and manipulation
Studied existing data analysis methods and platform architecture to define a foundation that supported both user needs and improved the way data was cleaned and verified
Defined early functionality, flows, and user types
Designed interaction patterns for filtering, inspecting, and verifying data across different levels of technical experience
Created user training material tailored specifically to data analysts’ level of technical expertise, and documentation.

POI comparison tool

Key Design Work

A hypothesis

Giving teams a single tool to explore, validate, and manipulate data, whether through SQL or UI, would reduce friction, replace scattered workflows, and build trust in the results.

Table Interaction
Made it easier to explore, filter, and edit data without breaking anything. Focused on small details that helped users feel confident making changes.

‍Visual Modes
Designed flexible layouts so users could choose how much data to see at once. This was key for reducing overwhelm during busy workflows.

‍Reading Messy Data
Worked on ways to show raw and cleaned data side by side, so users could understand how values were changing and why.

‍Flexibility for the Future
Built patterns that could support more than just filtering and editing, like comparing datasets or creating reports, without needing to start from scratch.

Main screen of Kuebiko's data manipulation tool

List of deliveries available to export

A raw JSON file display for fill rates

Kuebiko's POIs comparison tool

Design Iterations

Version 1 — OG AdmiNext
The earliest version was functional but overly technical. It used language and patterns familiar only to advanced users, which made it hard for others to understand or trust the outputs.

‍Version 2 — AdminNext
We introduced a more visual, interactive layout focused on spatial exploration. Users could work with data tied to specific locations, improving clarity for certain workflows but still falling short in flexibility.

‍Version 3 — Kuebiko
This version balanced both needs. It supported SQL users and non-technical users equally, offering just enough structure and customization for each to work confidently without stepping on each other's workflows.

Kuebiko iterations

Outcomes

Reflection

This project taught me the value of designing for layered complexity. Supporting both workflows in one tool pushed me to think deeply about defaults, mental models, and visual clarity. Every detail had to earn its place.

Replaced multiple fragmented workflows with a single, centralized platform
Became the trusted source of truth for operations and analysis teams
Eliminated the need for external spreadsheets and disconnected scripts
Improved validation speed and accuracy across teams
Positively impacted revenue by enabling faster decisions and fewer errors

A list view of kuebiko deliveries

Next Steps

The next step for Kuebiko is to introduce Validators, a new user type with fast-paced, high-volume tasks. This addition will allow the data team to apply different validation methods in a fraction of the time.

Kuebiko continued to evolve through iterative design, qualitative testing, and direct user feedback. Each improvement brings the product closer to serving a broader, more diverse user base, while keeping speed, clarity, and trust at the center.

A task of the Validators new persona