Figma: measuring the impact of AI on design practice
When a company wants to understand whether AI is changing how designers work, the obvious move is a survey. Shane Johnston, design researcher at Figma, took a different route: three years of structured qualitative interviews, moderated in part by AI, analyzed with AI, and synthesized into a single composite score called the AI Impact Index.
What the research covers
The AI Impact Index runs from 0 to 100 and represents how much AI has changed daily design practice. In 2024, when Figma first ran the study, the score was 31. In 2026, it reached 62 — a doubling in two years, which the team describes as the fastest measured shift in design workflow in the study’s history.
The data behind that number is substantial: 639 interviews conducted across roles and experience levels, producing just over 2 million words of transcript. AI moderated the interviews (handling follow-up questions and probing for specifics), while a separate analytical layer generated 40,701 thematic codes and collapsed them into 11,846 distinct themes. Human researchers then reviewed the synthesized output to validate categories and weight them.
What changed most
The research breaks the index into five dimensions: capability, process, collaboration, quality, and trust. The collaboration dimension showed the largest gain, rising from 32 to 58 between 2024 and 2026. Johnston interprets this as designers using AI to bridge the gap between disciplines — sharing explorations earlier with engineers, running more asynchronous feedback loops, and generating reference materials that non-designers can read directly.
Capability and process dimensions also rose but at a slower rate. Quality and trust remain the most contested dimensions: designers report that AI-generated outputs are improving but that verification overhead has grown alongside output volume.
How the methodology works
The study was designed to be repeatable. Figma built a cross-functional dashboard that tracks index scores by dimension and by demographic segment (years of experience, company size, design specialty). This lets the team compare cohort shifts year over year rather than relying on aggregate averages.
The choice to use AI moderation for the interviews was deliberate. Human moderators, Johnston notes, tend to probe certain answers more than others based on their own assumptions. AI moderation applied consistent follow-up across all 639 participants, which reduced interviewer bias but introduced its own artifacts — particularly in cases where a participant gave a short answer that a human moderator would have pushed on.
Who this is useful for
Design leaders and researchers trying to build a case for or against AI adoption will find the methodology more useful than the score itself. The index structure — qualitative interviews → AI coding → human validation → composite score — is transferable to other contexts. Teams running internal design research can adapt the approach to measure AI’s effect on their own workflows without needing the scale Figma had.
The article also includes charts breaking down dimension scores by experience level, which is relevant for organizations with mixed-seniority design teams. Senior designers show a notably different adoption curve than designers with fewer than three years of experience — a gap that has widened rather than narrowed since 2024.