--- name: case-study-review description: Review one product design portfolio case study at a time using ten scored dimensions, a stated target level, and concrete revision priorities. created: 2026-10-20T10:50 updated: 2026-10-20T13:37 --- # Case Study Review By Julia Wang. A ten-dimension rubric for reviewing one portfolio case study at a time as a prospective employer sees it. ## How to use Attach this file to your AI conversation, share the direct link to the case study you want reviewed, and say: > Use the attached Case Study Review skill (SKILL.md) to review this case study: [case-study URL]. > > Evaluate it against the Staff Product Designer bar. Review the desktop and mobile experience, including motion where available. Score it across all ten dimensions, show evidence for the scores, and give me the five most important fixes. > > Review only this case study. Do not edit my website. Change the target level if needed. Screenshots or a PDF of the case study work when the live page is unavailable. This file contains the full rubric; no other files are needed. ## Evaluation level **Suggested review level: Staff Product Designer.** This lens emphasizes strategic judgment, ownership, cross-functional influence, and strong design craft. Replace it with your target level, or provide a job description for a more tailored review. Use Staff when no other target is supplied. State the chosen level in the report. - **Early career:** Fundamentals, clear contribution, reasoning, and learning within the available scope. - **Senior:** Independent ownership, strong execution, informed decisions, and cross-functional delivery. - **Staff:** Problem definition, strategic choices, alignment, and influence beyond individual execution, supported by strong craft. - **Principal:** Direction and durable influence across multiple teams or product areas, supported by concrete examples. Titles vary by company. Judge the project within its actual scope and stage; a solo builder need not demonstrate a large organization's stakeholder process. ## Scoring scale Score each dimension from 1 to 5 against the stated level: - **5:** Exceptionally convincing evidence for that level. - **4:** Strong, with minor improvements possible. - **3:** Understandable, but generic, thin, or incomplete. - **2:** Present, but does not make its case. - **1:** Missing from the inspected case study or actively undermines the story. Score what the case study demonstrates, not the designer's underlying ability. Missing explanation can lower a score; unavailable inspection material should be marked **Not assessed**, not scored as a failure. If craft is only partly inspectable, label its score **Provisional** and state what remains unchecked. ## The ten dimensions 1. **Problem framing.** Who faced what problem? Why was this problem worth solving rather than another one? Explain the user and business context, not just a confusing interface. 2. **Stakes.** Why did it matter? Show the consequences for people, the business, or the community. A meaningful example can establish stakes when financial metrics are unavailable. 3. **Constraints and landscape.** What made the work hard, and how did that affect the approach? Look for relevant technical limits, priorities, policies, resources, or ambiguity. 4. **Research rigor.** What evidence informed the work? Is the method and scope clear, and does a finding connect to a decision? Judge the evidence's usefulness and limitations, not the number of methods named. 5. **Solution divergence.** What options were considered, and why was this direction chosen? A meaningful rejected option or a clear explanation of a constrained choice can show judgment. Do not demand extra concepts just to fill a process diagram or apply an automatic score cap. 6. **Decisions and trade-offs.** Are the pivotal choices and their reasoning clear? What was deliberately excluded or deferred, and what did that choice cost? Prioritize judgment over a chronological list of activities. 7. **Stakeholders and dead ends.** Who contributed what? How did the designer navigate differing needs, setbacks, or a change in direction when relevant? Look for ownership and influence without requiring conflict or overstating individual credit. 8. **Outcome loop-closing.** Does the result answer the original problem? Distinguish shipped work, measured results, early signals, and open questions. Unshipped work can demonstrate a valuable direction or decision; do not invent adoption or impact. 9. **Craft, visualization, and motion.** Review both the featured designs and the case-study page: - **Visual consistency:** Look for a design system or a coherent visual language across typography, colors, spacing, alignment, components, icons, illustrations, and motion. Cite specific inconsistencies. A named design system is not required. The page and featured product can have different styles; historical screens and rejected concepts need not match the final design. - **Visual explanation:** Do annotated screens, comparisons, flowcharts, illustrations, icons, and arrows make the decisions or behavior easier to understand? Flag decorative visuals, unexplained screenshots, and illegible labels. - **Desktop and mobile:** Check hierarchy, layout, text wrapping, image and diagram legibility, overflow, touch targets, and access to details. Mobile should be usable, not simply a smaller desktop page. Assess the featured product's responsive design only where relevant to its intended platforms. - **Motion:** Does timing, easing, continuity, and interaction feedback clarify behavior or guide attention? Flag distracting loops, delayed content, and scroll interference. Check playback controls and reduced-motion behavior where applicable. Absence of animation is not a weakness by itself. 10. **Writing quality and digestibility.** Can readers scan the page, understand the main argument, and explore the detail? Flag long paragraphs, consecutive text-heavy sections, repetition, weak headings, and poor grouping or spacing. Look for takeaway headings, clear ownership, and explained terminology. Recommend where to shorten, regroup, or replace prose with a visual while preserving the reasoning. Judge reading effort, not a word count or image quota. For density or visualization findings, name the section and suggest a specific treatment, such as a labeled flowchart, side-by-side comparison, or annotated screen. Explain what text it could replace. Visuals should explain, not decorate; icons, arrows, and color need enough labeling to be understood. ## What to inspect Review one selected case study as visitors see it. If only a homepage or several projects are supplied, ask which case study to start with. Keep homepage positioning and project selection outside this review. For a live page, inspect desktop and narrow mobile layouts, scroll through the story, and try relevant interactions when tools permit. Record viewport sizes and whether mobile was emulated or tested on a device. Assess motion from the live page or a recording, not still screenshots. Text-only material supports narrative review, not visual craft assessment. Report access limits without assuming the site is broken. ## Return this report 1. **Review target:** Case-study title and link or file, target level, desktop/mobile coverage, motion inspected, and any access limits. 2. **Case-study scorecard:** Ten rows, with dimension, score, and a specific reason. Include a total out of 50 only when all ten scores are fully assessed, with none provisional. Scores guide revision; they do not predict hiring outcomes. 3. **Fixes with evidence:** Quote or locate the issue, explain the gap, and suggest a concrete change for every score of 3 or below. Briefly identify strengths worth keeping. 4. **Panel questions:** Two or three skeptical questions grounded in the case study's most consequential claims. 5. **Fix first:** Up to five changes, ranked by their effect on clarity, credibility, and relevance to the target role. ## Review rules - Review only. Do not edit or publish unless asked. - Say "the story does not show a research finding," not "you did poor research." Missing evidence and weak work are different conclusions. - Do not invent facts to improve the story. Ask for missing evidence before suggesting a factual claim. - Flag apparent confidentiality or attribution problems at the top without repeating sensitive information. - Compare with a previous review when supplied, using the same target level and case study. - Be direct. Generic advice and inflated scores do not help revision.