I migrated 20 years of PSD files to Figma. Here's what nobody says about design systems.

Twenty years of PSD: a heritage, not an archive

I know this situation from the inside. After more than twenty years of practice, I've accumulated thousands of Photoshop, Illustrator and InDesign files - hundreds of projects, dozens of clients, years of design decisions frozen in a format that no longer meets the demands of modern collaboration. Each PSD is a creative snapshot, but a dead one: each variation is a layer, each state a group, each adaptation a copy. Figma, on the other hand, offers a living ecosystem - connected components, shared variables, modes, real-time collaboration. The gap between the two paradigms is enormous, and overcoming it requires a strategy, not a magic button.

The urgency was heightened by the mothballing of Adobe XD: teams who had migrated to XD found themselves changing tools for a second time in just a few years. Figma emerged as a credible final destination. I've been through this process several times, and I'll tell you what the marketing guides fail to mention.

The first thing nobody says: Figma doesn't import PSD.

This is surprise number one. Many designers expect a native import and fall flat on their faces. Figma correctly imports .sketch files, SVGs from Illustrator and vector PDFs - but not PSDs. You have to go through intermediaries, each with their own compromises.

Method	How	Result
Photopea	opens PSD, exports layers as SVG/PNG	free, tedious, reliable
Sketch/Figma/XD/PSD web converter	variable depending on complexity
Plugins "PSD to Figma" from the Community	uneven, handles layer effects poorly
Manual export from Photoshop	SVG if vector, high-def PNG if bitmap	maximum quality, intelligent decisions

For complex files, I prefer manual export from Photoshop: SVG when vector content, high-resolution PNG when bitmap. It's tedious, but it preserves quality and, above all, it forces me to make a decomposition decision at every stage - a decision that no automatic converter will make for me.

Migration in five phases

The process I experienced can be broken down into five phases. First, the audit: I catalog each file - content, last modification, usage level, strategic value. This mapping invariably reveals that 20% of the files concentrate 80% of the value. It's this core that we migrate first, not the thousands of dormant PSDs that no one has reopened for five years.

The second phase is the most important, and the one we most often botch: defining the target design system. This is the time to decide not to reproduce the existing chaos in a new tool. I take this opportunity to rationalize - reduce the number of colors, harmonize spacing, unify typographic styles, identify recurring components. Each simplification improves the coherence of the future system. The last three phases follow: building the Figma foundation (variables, modes, collections - I detail the mechanics in my articles on theming and design systems), rebuilding components natively with Auto Layout, then migrating static assets to optimized SVG and PNG. The key word is rebuild. A button is no longer a group of fixed layers: it's a component with variants, properties and constraints.

Pitfalls nobody mentions

The other pitfalls are more insidious. Text formatting is lost: custom kerning, text effects and distortions are not transferred. Colors drift between RGB and CMYK. Raster effects - complex shadows, bevels, layer textures - have no native equivalent and must be redesigned with Figma tools or exported as images. Every discrepancy must be documented and validated with the parties involved, otherwise it will resurface at the validation meeting as a "it's not exactly the same" that will hold up the project.

But the most underestimated challenge is not technical, it's human. A legacy of thousands of PSDs cannot be migrated in a few weeks, and the resistance of a team used to Photoshop for fifteen years is real. That's what nobody is saying: migration rarely fails for technical reasons, it fails because we underestimated the change in habits. I recommend a gradual approach - a complete pilot project, measured gains, a team trained on this concrete case, then an extension - with a planned depreciation of the old system so that nobody feels abandoned.

What AI means for the process in 2026

Since I started these migrations, the tooling has changed. Figma has rewritten the architecture of its design systems, making large files significantly more responsive - a concrete relief when handling massive libraries. Its layer renaming AI and third-party conversion tools have improved, and Figma's MCP server now allows an agent to read a file's structure. All this speeds up the repetitive work of renaming, arranging and extracting.

But - and this is the point I hammer home to my customers - AI speeds up conversion, not reconstruction. Deciding that a group of layers should become a three-variant component rather than a fixed instance is a design system judgment, not a sorting task. AI saves me hours on plumbing; it doesn't make any of the decisions that make a system good or bad. To confuse the two is to end up with a clean, tidy PSD in Figma - which is to say, still a PSD.

Accompany the change, measure the return

Technical migration is only half the job. The other half is human support. Thinking in terms of reusable components, reasoning in terms of Auto Layout constraints, collaborating in real time, living with integrated versioning - every aspect of Figma is a change in mentality that technical training alone cannot produce. I organize workshops where designers reconstruct in Figma a model they know by heart in Photoshop: starting from a familiar result to discover a new method reduces anxiety and accelerates learning. I measure progress not by speed, but by the moment when a designer spontaneously starts thinking in terms of variants and properties - that's when the cultural shift is won.

The return on investment is substantial. The teams I've worked with report a 50-70% reduction in production time for mock-ups after a four- to six-week adaptation phase. Once in place, the design system becomes a gas pedal whose output increases with each project. But the real gain is in terms of quality: visual consistency throughout the organization, simply impossible with thousands of independent PSDs. That's what we're really migrating - not files, but a way of working.