Expertise · DIY, hire, or hand over to the team

Using an existing template or creating custom from scratch

This question does not have a single correct answer for every case, but there are clear principles for you to make the right decision.

Quick summary

For most businesses, especially internal products or those with a tight budget, building everything from scratch is wasteful. Standing on a mature open platform like Google's Material or IBM's Carbon, and then applying a unique identity on top with a token layer (a customizable design value unit) is a more practical choice. Exception: when a consumer brand needs a truly deep identity or a unique experience that the platform cannot support, building from the ground up is worth the cost.

Quick comparison
You should choose this direction when
  • Internal products or management tools, need to be quick, with a tight budget.
  • borrow open architecture, replacing all token layers and visible layers
Not needed when.
  • a consumer brand that needs a strong identity and unique experience
  • places too much burden on the open platform; every update is a struggle

Before discussing technical details, you need to answer a foundational question: how different does your brand need to be compared to what users have seen in other products? That answer largely determines the direction for the design system.

Two directions and real differences

An open system like Google's Material Design or IBM's Carbon is not just a beautiful interface to copy. It is a system that has already solved hundreds of foundational issues: keyboard navigation, WCAG standards (guidelines for accessibility for people with disabilities), mobile form behavior, data loading states. Rebuilding these from scratch takes a lot of time and is prone to omissions.

Creating entirely from scratch makes sense when the product requires a truly different experience that no existing platform provides, or when a consumer brand needs such deep visual identity that borrowing from outside becomes a barrier.

Existing platform (Material, Carbon, etc.): Faster launch due to pre-built components and basic behaviors. Accessibility standards are already established. Suitable for internal products, management tools, and lean budgets. Risk: if customization is superficial, the result may still resemble the original system and lose its identity.

Create from scratch: Absolute control over visuals and experience. Suitable for consumer brands needing a strong identity or specific experience. Risk: it takes a lot of time to rebuild what has already been solved, and requires a team capable of maintaining the system long-term.

The token layer is the key to the hybrid direction.

Design tokens according to the DTCG standard of W3C are the smallest value units in the system: colors, font styles, spacing, and corner shapes. When you intervene at the token layer instead of overriding the internal structure of a component, you achieve both: structure and behavior from the open foundation, and a purely identifiable visible layer of your own.

This is the direction Sinh Vũ typically recommends for mid-sized businesses, meaning not a simple internal product but also not needing an extreme identity that hinders the foundation.

Design tokens that are neutral according to DTCG standards allow for the separation of identity from the technical background. This way, you can borrow the component structure of an open system while still maintaining your own colors, typography, and spacing.

Design Tokens Community Group (DTCG), W3C

Factors that determine the direction

  • Brand differentiation level required: If identity only requires a specific color and font, a flexible background is acceptable. If shape, movement, and layout must be entirely unique, a flexible background will become more of a burden than a support.
  • Actual customization cost: Overlaying on an open platform to achieve a unique identity often requires more effort than one might think, especially with complex components like data tables or date pickers.
  • Coverage level of the underlying system: Carbon is strong in data and business context. Material excels in multi-device experience. If your needs deviate from those strengths, the underlying system will be less helpful.
  • Team's long-term maintenance capability: a proprietary system requires a team capable of sustaining it long-term. An open-source system has a community and available documentation for recruiting or training additional personnel.
  • Lock-in risk: If Google or IBM makes significant changes in their systems, the elements you have built upon may conflict and need to be reworked.

Common errors to avoid

  • Using an open background and layering too much at a shallow level, resulting in a loss of identity and difficulty upgrading according to the original background when they release a new version.
  • Custom building from scratch and spending years rebuilding open-source solutions that have already been solved, such as keyboard navigation, accessibility standards, and mobile forms.
  • Creating multiple branches (forks) from the root to patch each individual case, and when the root updates, it's unclear whether to merge or rewrite from scratch.

Sinh Vũ's viewpoint

Sinh Vũ does not worship a fixed foundational system. The question is not "Material or Carbon or DIY" but rather "which layers can be borrowed, and which layers must be created separately." For most clients, the answer is: borrow architecture and the principles of approach, and create all token and visual layers separately.

Components are built according to the technology stack the client is using, which can be React, Vue, or Web Components, without forcing the client to adopt an unfamiliar stack just because the underlying system requires it. The ultimate goal is for you to own a self-sustaining internal system, independent of external vendors for each small operation.

The tool brings back.

Decision checklist

Topic: Using existing templates or creating a design system from scratch. Sinh Vũ guide, sinhvu.com

0 more than 6 items

Select each item you find appropriate, then print or save as PDF to take with you.

Sign indicating that you should take action
Questions to answer before deciding

If you have marked most of the signs above, this is the time to discuss in more detail. Sinh Vũ can help you review and propose a direction.

References

Design Tokens Format Module, DTCG (W3C Community Group); IBM Carbon Design System; Material Design (Google); summary of Material vs Custom practices (Edana); practical experience from Sinh Vũ.

Frequently asked questions

Can customized Material or Carbon really create a unique identity?

Yes, but with conditions. If you only change the button color and font at a superficial level, the result will still look very similar to the original system. To achieve a truly distinct identity, you must intervene from the token layer down: color, typography, spacing, and corner shapes. This task is not light, but it is still less than rebuilding everything from scratch.

If using an open template, will upgrades be difficult later?

Whether it is difficult or not depends on how much you have built upon the original system. If you build little, following the token layer, the upgrade is relatively straightforward. If you create many branches separate from the original foundation, each time they release a new version, it requires reworking. The principle: build on the token layer, not on the internal component structure.

What is a hybrid approach and is it suitable?

A hybrid approach borrows the architecture and principles of an open platform (including WCAG accessibility standards) but replaces the entire visible layer with a unique identity. This is the approach Sinh Vũ often recommends as it balances launch speed, brand identity, and future maintainability.

← Back to Digital design system