Start with a slice: 7 steps to get your new design system moving
2025-10-09 • Paul Love
Without a concrete goal, design system projects get stuck weighing up the options. Here’s how to choose a narrow, visible slice that reveals the right problems early and demonstrates progress everyone can see.
Getting a design system off the ground is hard. Teams often start too broadly. Without a concrete goal, it's easy to get stuck weighing up the options.
In this post we share how to take a more focused approach by choosing a narrow slice to build to show value quickly.
1. Start with a visible, bounded slice
Choose one or two target pages to begin with. These pages should be highly visible but clearly bounded. You want a manageable scope, but also show an immediate impact.
A focused slice helps reveal the right problems early and demonstrates progress everyone can see.
The homepage is often a strong choice. It’s rarely representative on its own, but it's often politically non-negotiable. People will inevitably ask, “… but what about the homepage?” Including it is a pragmatic way to avoid hassle.
Pairing it with a second, more typical page helps make sure you're solving recurring problems. This could be a product listing or dashboard, for example.
Avoid pages that make heavy use of overlays or modals. That’s too much complexity, too early.
2. Build a shadow world with Storybook
Create a shadow world. Recreate your product in a low-stakes space your team owns. It’s not live. It’s not connected to real data. It doesn’t need to be perfect. It’s a place to explore and demonstrate what your design system can do.
Set up a Storybook environment to develop and present components without delivery pressure. This removes risk by letting you experiment, validate decisions and get buy-in before touching production code.
Storybook lets you test components in isolation, explore states and variants, and prototype static pages. This creates a shared frame of reference and enables fast iteration.
3. Categorise your components deliberately
Next, analyse your target page. Divide what you see into three categories:
Potential design system components: Reusable patterns like buttons, cards or form fields.
Site furniture: Recurring structural elements like headers, footers, and navigation.
Custom components: Unique, often complex, items.
Design tokens tie all three together with consistent styling. This resists the trap of becoming the department responsible for all the things rather than a team that enables others.
4. Decide what to tackle and what to park
Be strict about scope. Account for layout mechanics and logic underpinning visible components. Keep a list of future needs, but resist tackling them now. (This makes a great starting point for an opportunity roadmap.)
Look out for what we call icebergy componenets: seemingly simple elements that hide complexity. For example, a bell icon might open a complex messaging panel. A text input requires validation states, complex accessibility features, and error handling.
Start with the simplest version that works. You can always add complexity later, but you can't ship if you try to solve every edge case upfront.
5. List components, then order your steps
List the components needed for your target page. Group them by type. Start with something basic like a button. Jot down what it needs to get it working.
This simple exercise reveals more than you'd expect. Particularly if you're new to design systems, or if your role isn’t hands on. it surfaces foundational work that wasn't on your radar. You'll discover you need to:
- Define and create a starter set of design tokens
- Set up tooling: code repositories, Storybook deployments, tests
- Figure out versioning and publishing
- Write basic getting started docs
- Factor in cross-cutting concerns like accessibility, security, and internationalisation
One by one, move to the next component, noting only new tasks. The net-new work shrinks rapidly as foundational elements fall into place.
This approach reveals everything you need for an alpha release and exposes hidden complexity early so you can plan for it rather than scramble.
6. Create the right conditions
This approach demands a collaborative, cross-functional team. It helps to have T-shaped people: developers who care about design and designers with technical empathy.
You need a shared mindset embracing modern product development. Without it, teams fall back into old patterns. Design becomes a one-off handover rather than a collaboration, and the whole approach fails.
Set expecations: progress and learning in tandem are key.
Clarity, momentum, and progress
A narrow, visible slice gives your team a shared focus and a tangible goal. It helps you solve the right problems early, without overthinking.
The trick isn't having all the answers upfront. It's building something real that teaches you what questions to ask next. Get your slice working, and momentum takes care of the rest.