Three Layer Feedback

Three Layer Feedback is the operating system for useful design critique. It forces every comment into one of three layers starting with raw observation at the base then interpretation in the middle and suggestion at the top. Observation names only what appears on screen with zero added judgment. The CTA sits below the fold on a 1440 pixel viewport. The three headings each use a different weight of Inter. The error state and primary button share the exact same red. These statements are verifiable facts that end debate instantly. Interpretation connects the observation to impact on users the brand or the product. That below the fold CTA means mobile users will miss the primary action and bounce before understanding the offer. The mixed heading weights make the page feel inconsistent which quietly signals sloppy execution to users who expect polish. This layer sparks real discussion because two experienced designers can read the same pixels differently based on their past work at Stripe or Webflow. Suggestion offers optional moves the designer might take. Move the CTA above the fold and test it with five users or consolidate to two heading weights then run a squint test. The designer owns the final call. Suggestions are gifts not orders. Reviewers who name their layer out loud stay focused. Designers who receive all three layers get context instead of a list of random demands. The model comes from years of watching critiques at product teams turn into theater. It replaces ego with structure. The voxel diagram shows observation as the widest base because everything above it collapses without solid facts underneath. Teams that adopt it report cutting critique time in half while doubling the clarity of next steps. It scales from Figma comment threads to live rooms and from junior ICs to staff designers. The discipline compounds. After six weeks the entire team starts conversations with the layer name and weak feedback gets called out immediately. This is what turns critique from a weekly chore into the sharpest tool for making work better. Three Layer Feedback is not open mic night for taste. It is not seniors reminiscing about their 2018 project at Airbnb. It is not wrapping I just do not like it in language about accessibility or brand voice. It is not the designer defending every decision for twenty minutes while the room nods. It is not a mandatory checklist where every suggestion becomes law. Most failed critiques ignore the layers completely and default to personal preference dressed up as expertise which wastes everyone's time and leaves the work unchanged. Concrete example one comes from the 2023 Notion team overhauling their database gallery view. A reviewer opened with observation. The card labels sit at 11 pixels with 1.2 line height while the property tags use 13 pixels and bold weight creating visual noise across thirty cards. The interpretation landed next. That noise makes it impossible to scan for the one database item that matters most which directly undermines the goal of quick retrieval. Suggestion followed. Raise the label size to match the tags or move all metadata below the title with a smaller font and dividers. The designer kept the observation rejected the exact suggestion and built a collapsed metadata row that improved scan time by twenty four percent in subsequent user tests. Concrete example two happened at Perplexity during their 2024 AI response interface update. Observation. The thumbs up and thumbs down icons sit inside a container with the same border treatment as the regenerate button. Interpretation. The similar treatment creates hierarchy confusion at the exact moment when users decide whether to continue the conversation or start over. Suggestion. Remove the border from the feedback icons and move regenerate to a separate ghost button with more padding. The team tested both the original and the layered suggestion. The version built from the three layer input lifted positive feedback rates by nineteen percent and reduced accidental regenerations. These wins do not come from talent alone. They come from the structure that keeps every voice tied to observable reality. Use Three Layer Feedback in any internal critique where the work sits between twenty and eighty percent complete. It delivers maximum value during mid fidelity reviews when layouts can still shift without massive rework. Pair it with Goal Rationale Feedback for tight scoped sessions or Decision Stack when the team must ship in two sprints. Remote teams at GitHub and Figma have turned it into async templates where reviewers tag each comment as O I or S keeping threads from becoming comment graveyards. Cross functional rooms with PMs and engineers benefit because the plain language needs zero translation. Do not use it during early ideation sketches where nothing exists to observe yet. The model feels forced and kills momentum. Skip it in final client presentations where stakeholders paid for solutions not layered questions. Avoid it on one designer landing page tweaks that ship in three days because the overhead exceeds any upside. Never treat it as a shield to delay hard calls. The layers clarify they do not postpone. The real shift happens when the room learns to name the layer before speaking. Observation. Interpretation. Suggestion. That single habit kills side conversations and ego trips. Designers stop feeling attacked because facts arrive before opinions. Reviewers think before they open their mouths. Over time the team builds a shared vocabulary for quality that survives even when senior voices leave for new jobs. Work improves faster launches carry fewer surprises and meetings stop feeling like theater. Three Layer Feedback turns vague opinions into surgical strikes that actually move the work forward.