I sell design systems. Here are some situations when you shouldn't buy one from me.

The Conflict of Interest, Resolved from the Outset

I have a services page that sells design systems. I built one for my own site, migrated twenty years’ worth of legacy files to Figma, and wrote entire articles on tokens and theming. Needless to say, I have every reason to convince you that you need one. That’s exactly why this article exists: the question “Do you need a design system?” almost always gets a salesperson’s answer, and mine—after twenty years—is more nuanced: most small and medium-sized businesses don’t need one. They need a specific subset, which costs twenty times less. Let me show you where the line is drawn, with hard numbers to back it up.

“Best practices” come from a world that isn’t yours

Almost everything written about design systems comes from an enterprise context, and industry survey data proves it. In zeroheight’s 2026 Design Systems Report, 39% of respondents work in organizations with more than 5,000 employees; while only 11% work at companies with fewer than 100 employees. The 2021 Sparkbox survey showed the same skew: barely 10% of respondents came from companies with fewer than 50 people. When you read “every team should have a design system,” understand this: every team that resembles those in the surveys—dozens of designers, multiple products, constant staff turnover.

And even in that world, the picture isn’t rosy. According to zeroheight 2026: 61% of design system teams say they are understaffed, satisfaction with executive support has dropped from 42% to 32% in one year, adoption is “complete” in only 7% of organizations—this is the fifth consecutive year that adoption has topped the list of challenges—and 5% measure return on investment. Reread that last figure. The industry that sells you on the ROI of design systems doesn’t measure its own in 95% of cases. Even at Sparkbox, back in 2022, only 16% of maintainers were tracking metrics. The design system is one of the few tech investments justified almost exclusively by averages from compiled studies—the famous “+38% design efficiency” figure, which comes from a compilation of case studies from 2016 to 2019, not from a controlled trial.

What It Actually Costs

The few honest public cost estimates are enough to make SMEs’ heads spin. The agency Autentika detailed its calculation for 2024: approximately 100 to 120 hours per component (design, development, testing, documentation), a realistic v1 with 40 components, totaling nearly 4,800 person-hours— “three full-time people for ten months,” about $288,000 USD at their rates. The benchmark ROI formula, published by Smashing Magazine, assumes that a team devotes 30% of its time to the system during the ramp-up phase, then 10% indefinitely—and its authors specify that it is calibrated for a context involving approximately ten designers and thirty developers. Their profitable example (135% ROI over five years) is based on a 15-person product team.

Do the reverse calculation for a typical Quebec SME: two to five people working on the user interface, one product, and a major redesign every three or four years. That one-tenth of the time “forever” amounts to half a day a week spent maintaining the infrastructure instead of delivering. Below a certain team size, a full design system isn’t an investment—it’s a liability disguised as best practices—a garden that requires a gardener you don’t have. Even the Nielsen Norman Group, in May 2026, noted that the most effective design system teams remain tiny by choice; but these “lean” teams of two to five people serve organizations with thousands of employees. That’s not your situation, and it’s not meant as an insult.

The Three-Question Test

When a client asks me for a design system, I ask three questions. How many different people create interfaces or visuals each month? If the number is three or fewer, “consistency across teams” isn’t an issue you face—you are the consistency. How many distinct products or user interfaces? A website and a Facebook page don’t justify a governed component library; two apps, a public website, and a client portal do—that’s actually the point at which my own studio made the switch. How often does the interface actually change? If the answer is “every three years, during a redesign,” a living system has nothing to keep alive.

Zero or an answer “above the threshold”: you need a kit, not a system. Two: gray area—start with the kit and see what doesn’t work. Three: now we’re talking about a real design system, and this is when the +47% speed increase measured on IBM’s Carbon system (Sparkbox study, eight developers, median of 2 hours vs. 4.2 hours for a form) starts to apply to your reality.

The Minimal 2026 Kit—and Why Now Is the Best Time in History to Settle for It

The good news the industry rarely mentions: the underlying infrastructure has become standardized, and the core benefits of a design system are now accessible for just a few thousand dollars. Three building blocks have changed the game. The W3C Design Tokens Community Group’s design tokens spec reached its first stable version on October 28, 2025—after six years in the making—and is supported by Figma, Tokens Studio, Style Dictionary, and others: your colors, typography, and spacing now live in a neutral, portable format—no longer in a PDF. Style Dictionary v5 (1.7 million downloads per week) transforms these tokens into CSS, Tailwind variables, or whatever you want. And shadcn/ui—118,000 GitHub stars, over 5 million weekly downloads of its CLI—provides accessible components that you own in your code, with no dependencies to maintain.

In practical terms, the minimal kit I’m providing consists of three parts. Token-based foundations: a color palette with semantic roles, a typographic scale, and spacing rules—all defined in Figma Variables and exported in DTCG format. A standards page: the logo and its usage guidelines, application examples, and your editorial tone—the brand guidelines whose content I’ve already detailed. And a dozen components that are actually used (buttons, forms, cards, navigation), linked to the tokens, each documented in a single line. It’s a two- to three-week job, not a ten-month one. Based on my real-world experience, this captures 80% of the consistency benefits at a fraction of the cost—and most importantly, it maintains itself, because there’s almost nothing to maintain.

Growth Will Decide, Not Doctrine

The minimal kit isn’t a compromise—it’s a compatible foundation. The day you hire a second designer, when a client portal is added to the site, when inconsistencies start costing real hours—DTCG tokens get versioned, shadcn components get documented, and the kit grows into a system without having to start from scratch. That’s exactly the path I followed for my own studio: years using the equivalent of a kit, and a full-fledged system only once the user interfaces had multiplied.

My final piece of advice—the one that costs me client projects: if a vendor offers you a complete design system before asking how many people work on your interface, be wary. A design system is a tool for scaling. Sell it to someone who doesn’t have the scale, and you’re selling them a second job.

Sources

• Design Systems Report 2026 — zeroheight · Overview of the 2025 edition
• Design Systems Survey 2022 (5th and final edition) — Sparkbox · 2021 edition
• “How much is a design system?” — Autentika, June 2024
• The ROI Formula for a Design System — Smashing Magazine, September 2022
• ROI Study on Carbon (IBM) — Sparkbox, 2021
• First stable version of the Design Tokens spec — W3C DTCG, October 28, 2025
• “Small by Design” — Nielsen Norman Group, May 15, 2026
• Tool adoption: shadcn/ui and npm/GitHub metrics as of July 4, 2026