Into Design Systems 2025: The more things change, the more they stay the same

This year’s Into Design Systems conference held in May once again brought together designers, engineers, and system leads for three days of everything design systems related. The conference began a few years ago as a space to exchange component-building tips, but has grown into an annual forum for exploring governance models, accessibility regulations, measurement strategies, and new tooling, among other things.

Recognized industry leaders and top talent gather around the campfire to share best practices and lessons learned, from token naming structure to transforming hundred year old companies. The speakers included people from eBay, Pinterest, Equinor, ExxonMobil, Posti (Finland!), REWE and Preply – and of course scene heavyweights like Samantha Gordashko, Romina Kavcic and Nathan Curtis. 

Topics varied across the board from technical detail (eg. naming strategies for tokens) to organizational challenges, such as demonstrating ROI to executives. Design systems are becoming deeply embedded in how organizations structure their product development processes, with conversations extending way beyond component details. Maybe the most noticeable overarching theme was the ever-so-important accessibility, and it was nice to notice how central an aspect it is becoming.

Tokens, modes and more

Tokens, tokens, tokens. The omnipresent design system talking point of the last ten years. Almost every design system these days has already leveraged – at least to some extent – the massive power tokens bring to product development, but the shop talk has gotten increasingly intricate. 

The most advanced systems these days can create so-called dynamic tokens. It means roughly that things like brand attributes are dished out to products via mathematical functions: the common example would be calculating the color shades from the brand color “on the fly” while preserving saturation (a common problem with certain bright colors – simple math just won’t do). People at Tokens Studio have been at the subject for a couple of years already, and while tried-and-true applications are still scarce, it is certainly an interesting and potentially very powerful approach.

Tokens, dynamic or not, also offer a foundation for easy scalability for design systems, especially in cases like multi-brand theming. A single component “skeleton” can be “dressed up” for different purposes with comparatively little effort. However, to be able to leverage all this, tokens need to be classified and named appopriately.

It can be especially tricky since there is no one-size-fits-all solution –the final architecture depends on your product ecosystem. Matching design tokens’ Figma representation (variables) with actual tokens that the developers use can also be tricky – as Daniel Yuschick (Posti) and Nathan Curtis pinpointed in their presentations. 

This time, let’s make it a variable

Speaking of Figma variables and variable modes – what they can and cannot do? Well, the consensus at IDS felt like they are in a transitional phase: the feature is without a doubt strong enough to carry individual design decisions, like colors and light/dark mode. But people like Victor Nystad from Equinor have researched ways to push it a little further. 

He showcased a set-up where variables and modes also store things like component layout, icon size, typography, et al – which means steering away from Figma component properties (which still in many cases means drawing multiple variants of the same component) and dive into a more CSS-like world of styling. While using variables this way has its drawbacks – such as Figma’s mode inheritance from the parent layer – it also has definite potential, especially in multi-brand organisations.  

I was especially delighted to see that presentations like Victor’s were not abstract concepts but concrete walkthroughs of how variables are set up and maintained in everyday work-flows. It is much easier to estimate their long-term value that way. Variables may not solve every problem, but they are becoming more and more integral part of the design system toolkit.

Accessibility as a system-level concern

European Accessibility Act came into effect in June 2025, and so accessibility naturally had a prominent place on the IDS agenda. The shift in framing was more obvious than ever: instead of talking about accessibility just as meeting the WCAG criteria, the sessions were more concerned on how to embed accessibility directly into systems and use it as a foundation for product design – not something to plaster on afterwards. Since so many aspects of accessibilty propagate easily from design systems, it is a critical discipline to master.

It is also essential for the designer to understand exactly why accessibilty is so important and not treat it just as a box to tick. Marcelo Paiva’s and Sarah Massengale’s presentation framed nicely why exactly people with neurodiversity struggle with using digital products and how small changes can make a huge difference.

If you document it, people will use it… or will they?

Accessibility doesn’t concern just the end-users. Internal design system documentation also needs to be accessible, especially in the sense that the information has to be easy to find and engaging – otherwise it will not be utilized. eBay’s new Playbook demonstrates this beautifully by tying together both a brand book and a design system documentation. It showcases in a powerful way where the design decisions actually come from and how they can be deployed effectively into the product. The link remains concrete every step of the way, and avoids the pitfall of many design systems, ie. designing in a vacuum and forgetting the end user’s mental model.

eBay’s Anna Zaremba, Cordelia McGee-Tubbs and Ryan Tinsley also demonstrated beautifully how building documentation that teams love to use doesn’t have to be a grueling, years-long effort. Using agile methods and active contribution from every type of stakeholder, they have built an engaging documentation with a core team of just three people in a relatively short timeframe. The team has also built intricate tools for managing their design system documentation directly from Figma, as well as annotating accessibility easily in Figma.

How to be sure that whatever we do actually matters

Teams often use different analytics for design systems, but they tend to concentrate on the designers’ use of styles, variables and components, partly for the reason that those are the easiest to track from Figma. The end user benefits are often discussed with anecdotal evidence only. But Stefano Magni from Preply, a language tutoring platform, has built with his team a much more sophisticated tool.

To put it simply, they’ve created a library that measures the visual impact of the design system components that actually show on the user’s interface – in real time. It counts the amount of pixels that the components occupy on the user’s screen at any given moment, and gives those pixels a weighted value according to the component’s importance for the user experience (a call-to-action element gets a higher value than a profile picture, even if they have the same amount of pixels).

Under the hood it is pretty sophisticated stuff, but the value is easily presentable to the stakeholders. Preply has a data platform with a neat dashboard, where they can pinpoint each team’s adoption of the DS and how much it actually benefits their users. Concrete data also shows the possible gaps – and what improvements might have the most impact.

Best of all, it’s open source.

…but what about AI?!

Nothing to report, really! I mean of course, design system designers and developers use AI tools as much as anyone else – but they seem to be leveraged mostly as a sparring companion and automation tool. How long that will be the status quo, and is there a gamechanger on the horizon is anybody’s guess, but I feel the good old craftmanship isn’t leaving this domain anytime soon.

As a sidenote, the Figma MCP server was released right after the conference – I expect to see some interesting applications in the near future.

Closing reflections

Compared to last year, when dynamic tokens were all the buzz, this year’s conference didn’t bring too many new concepts on the table, but instead strengthened the existing ones. The focus was more on empowering designers with representation, tooling and ways of working – ever so important in today’s challenging environment, where it is harder and harder to justify the cost of design system work.

As always, the range of Into Design System 2025 was admirable: there were highly specific techniques that anyone can apply immediately, but also broader discussions on governance, measurement, and accessibility. Besides those, collaboration and scalability seem to be the larger concepts that most everyone is keen to immerse themselves into right now. 

This all points to maturity – the formative period of the DS discipline is well behind us now, and every system should be creating value. Design systems are no longer experiments but investments for a better future.