Frontend System Design · Lesson 19 of 86 · 4 min
Component-Based Architecture Principles
This is how you explain breaking down a UI without sounding like you just read a textbook.
In a frontend system design interview, you will be asked to break down a complex UI into smaller pieces. Your ability to justify where you draw the lines between components reveals your seniority. This lesson is about drawing those lines effectively.
How to Decompose a UI
Your first move is always to partition the interface. Look at the mock or the requirements and draw boxes. Each box is a potential component. The guiding rule for drawing these boxes is the Single Responsibility Principle (SRP).
A component should do one thing. "One thing" means one primary job or reason to change. A DatePicker component's job is to select a date. A UserAvatar component's job is to display a user's image. If you find yourself naming a component UserProfileAndSettingsForm, it's doing too much. Split it.
This decomposition creates a component tree, a hierarchy with a single root component (often called App) at the top.
The trap: Candidates often draw too few boxes. They'll describe a "feed page" as one giant component that fetches data, renders a list, handles infinite scroll, and manages the state for each item. This is a red flag. A senior engineer sees a
FeedPagecontainer, which holds aFeedList, which renders multipleFeedItemcomponents, each of which might use anAvatar,LikeButton, andCommentInput. Show that you can see the smaller, reusable parts.
Defining a Component's Public API
Once you've identified a component, you must define its contract with the outside world. This contract is its public API, consisting of its inputs and outputs. Everything else is an implementation detail.
Inputs are how a parent component passes data and configuration down. In React, these are props. In Angular, @Input() decorators.
Outputs are how a component communicates events up to its parent. This is typically done with function props or callbacks (onClick in React) or event emitters (@Output() in Angular).
A well-defined API makes a component reusable and predictable. You can swap out its internal logic or styling, and as long as the API remains the same, the rest of the application won't break. This is encapsulation.
This Button component has a clear API. Its inputs are label, onClick, and an optional variant. Its output is the call to the onClick function. The internal class name logic is a private detail.
// components/Button.tsx
import React from 'react';
interface ButtonProps {
label: string;
onClick: () => void;
variant?: 'primary' | 'secondary';
}
const Button: React.FC<ButtonProps> = ({ label, onClick, variant = 'primary' }) => {
const buttonClasses = `px-4 py-2 rounded ${
variant === 'primary' ? 'bg-blue-500 text-white' : 'bg-gray-200 text-gray-800'
}`;
return (
<button className={buttonClasses} onClick={onClick}>
{label}
</button>
);
};
export default Button;
Prefer Composition to Configuration
A common junior pattern is to create components with dozens of boolean props for configuration: showIcon, isRound, hasBorder, useDangerColor. This creates a rigid component that is hard to extend.
A better approach is composition. Instead of adding more props, design components to accept and render other components. In React, the children prop is the primary mechanism for this.
Configuration (brittle):
A Card component with props like withHeader, withImage, imagePosition. What happens when you need a video instead of an image? You add another prop, withVideo. This is not scalable.
Composition (flexible):
A Card component that simply renders whatever you pass to it.
// components/Card.tsx
import React from 'react';
interface CardProps {
children: React.ReactNode;
}
const Card: React.FC<CardProps> = ({ children }) => {
return <div className="border rounded-lg shadow p-4">{children}</div>;
};
// Usage:
// <Card>
// <img src="..." alt="..." />
// <h2>Card Title</h2>
// <p>Some descriptive text.</p>
// <Button label="Click Me" onClick={() => {}} />
// </Card>
The parent component now has full control over the card's content and layout, using other standard components. The Card itself does one thing: it provides a styled container.
In an interview, suggesting a compositional API for a component like a Modal, Card, or Layout demonstrates a mature understanding of building flexible design systems.
The Trade-offs of Granularity
How small is too small? The answer depends on reusability and complexity.
Over-atomization is a common mistake. If you break components down too far, you end up with a mess of tiny files that do almost nothing, and you spend all your time wiring them together. This often leads to "prop drilling," where you pass a prop through five layers of components just to get it to a deeply nested child that needs it.
Under-atomization results in monolithic "god components" that are impossible to test, reuse, or reason about.
A good heuristic: if a piece of UI has its own distinct state, has a complex internal structure, or is used in more than two places, it's a strong candidate for its own component. If it's just a div with a class name, it's probably not.
Prop drilling isn't always a mistake, but it's a signal. If you're passing the same prop through more than two or three layers, you should question your component boundaries. It might be a sign that a child component needs to be pulled up a level, or that you need a state management solution.
Try it
Take a screenshot of a complex UI, like the Twitter/X feed or the Amazon product detail page. Open it in an image editor and draw boxes around the elements you would break into components. Name each component and draw arrows to show the parent-child hierarchy. Spend 15 minutes justifying your decisions based on the principles of SRP, reusability, and composition.
Next: We'll look at the data that flows through these components in Props, State and Component Lifecycle.
Lock it in — apply this lesson under interview conditions.