What is segmentation & why it matters
Segmentation is the process of dividing a broad customer base into smaller groups of people who share similar characteristics — behavior, demographics, or purchasing patterns.
In data terms, it means analysing datasets to discover natural groupings, then using those groupings to personalize experiences, target campaigns, and prioritize product decisions.
A retail chain with thousands of customers, for example, might discover a set of high-frequency, low-basket shoppers who respond to entirely different offers than occasional big-basket buyers — insight that stays invisible in an unsegmented view.
Segmentation enables businesses to better understand their customers, tailor experiences to distinct needs, and make sharper, faster product decisions.— Segmentation Tool Design Document, v2.0
Four personas, one unified platform
The tool serves four distinct user types, each with different skills, goals, and reasons for opening it. Every screen has to make sense to all four at once.
Two modes for every use case
Not every team starts from the same place. Some are exploring a dataset for the first time; others are applying a segmentation model they already trust. Both paths start from the same entry point.
New Segmentation
Discovers new segments from unlabeled data using unsupervised clustering — ideal for the first time you're exploring how a customer base naturally groups together.
Existing Segmentation
Applies a segmentation model you've already validated to new or updated data, keeping every classification consistent with the original definitions.
Color, defined with intent
Every color in the system carries meaning before it carries decoration. Segmentation and classification each get a distinct hue so users always know which mode they're in; status colors stay identical across both themes.
| Purpose | Dark theme | Light theme | Why |
|---|
Step-by-step user journeys
Every flow is documented as the design reference for engineering handoff — from the first file upload to the final saved segment.
End-to-end process flow
The complete decision flowchart covering both modes, from the first upload to the final health check — one source of truth for design and engineering.
Six core feature modules
Every module in the platform, with design intent and functional scope, organized by what it's responsible for.
How this design came together
From early alignment across disciplines to the specific UX gaps found in early flows — and what shipped to close them.
Before any pixel was placed, I sat down with frontend engineers, backend engineers, and the design lead to map what was technically feasible, what data was available at each step, and where the biggest UX risk lived — the moment a user has to trust an AI recommendation without seeing the reasoning behind it.
| UX gap found | What I designed |
|---|
Feature prioritisation by business value
Not all features carry equal design weight. This map plots each module by business value against design effort, to sequence what ships first.
| Feature / module | Business value | Design effort | Notes |
|---|
Built with a little AI help
This case study — the copywriting, content structuring, and parts of the build — was put together with the help of AI tools, ChatGPT and Claude, alongside my own design and research work.
View source file on Figma