Stop Lack of a Design System Blocking UX Research Quality — Gemini Prompts for Agency UX Researchers

You are an expert UX research strategist with 13 years of experience writing design briefs and usability research frameworks for digital agencies where the absence of a design system means every usability test is evaluating inconsistent UI rather than the user's actual interaction with a coherent product, producing test results that identify symptoms of design system absence rather than genuine usability problems. Help me write a design brief so I can improve usability test results and create a brief that separates the design consistency problems from the genuine usability problems before the test begins, so the research findings are actionable rather than a list of visual inconsistency issues that require a design system to resolve. My situation: - Agency client type and product being researched: [e.g., "a travel booking platform undergoing a homepage and search flow redesign — the existing product has no design system and 3 different component libraries that evolved independently across 4 years of development"] - Primary usability test objective: [e.g., "understand why the search-to-results conversion rate is 34% below the industry benchmark — the product team believes it is a UX problem, the design team believes it is a visual consistency problem creating cognitive friction"] - Current design system status: [e.g., "no formal design system exists — the Figma file has 12 frames with inconsistent component naming, 3 competing button styles, and 4 different card layouts used interchangeably across the product"] - Usability test participant profile: [e.g., "8 participants, ages 28 to 55, booking travel for leisure and business — they have used competing platforms (Booking.com, Skyscanner, Google Flights) in the last 3 months"] - Test format and tooling: [e.g., "moderated remote usability test using Maze for task completion tracking and Zoom for verbal protocol — 45 minutes per participant, 5 tasks covering search initiation, filter application, result comparison, and booking initiation"] - Landing page redesign scope: [e.g., "the homepage and search results page are in scope — both have been redesigned in Figma but have not been built, so the prototype fidelity is high enough to test but inconsistencies between the homepage and results page components are visible"] - Research findings audience: [e.g., "the client's CPO and head of product — they will use the research findings to prioritize either a UX fix or a design system investment, and the design brief must frame the research question in a way that produces a clear recommendation for one or the other"] Deliver: 1. A design brief for the usability test — a structured document covering the research question (does the search-to-results drop-off reflect a navigational or cognitive load problem, or a visual consistency problem creating distrust), the hypothesis for each possible cause, the specific task scenarios designed to isolate each hypothesis, the metrics for each task (completion rate, time on task, error count, and subjective confidence rating), and the success criteria that would confirm a UX fix versus a design system investment 2. A pre-test design consistency audit — a structured process for cataloguing the visual inconsistencies in the current prototype before participants interact with it, identifying which inconsistencies are likely to surface as test observations and which are likely to be invisible to non-designers, so the research findings can separate design system problems from genuine usability problems in the analysis 3. A task scenario script for the five test tasks — each task written as a realistic scenario grounded in the participant's own travel booking motivation, with the specific prototype start state, the success condition, and the facilitator prompts for when a participant is blocked without giving directional help 4. A think-aloud protocol guide — the specific facilitator instructions for capturing verbal protocol that distinguishes between a participant expressing confusion about visual hierarchy (a design system problem) versus expressing confusion about information architecture or interaction model (a genuine UX problem), with the verbal pattern that indicates each type of friction 5. A competitive benchmark comparison brief — a structured comparison of the search initiation flow in three competitors (Booking.com, Skyscanner, and Google Flights) against the prototype, identifying the three specific interface moments where the prototype deviates from the established mental model of travel search and the expected cognitive impact of each deviation 6. A research findings presentation template for CPO and head of product — a five-section format covering the test objective, the three most significant findings ranked by severity, the specific UI moment associated with each finding, the design system versus UX root cause classification for each finding, and the recommended investment priority with the expected conversion impact of addressing each root cause 7. A design brief approval process — a structured review of the design brief by the client's product and design leads before recruitment begins, covering the alignment check on the research question, the prototype fidelity sign-off, and the findings criteria agreement, preventing the common failure mode where research findings are disputed because the CPO and research team defined success differently 8. A post-test design brief retrospective — a one-hour team session held immediately after the test debrief, reviewing which brief assumptions the test confirmed or refuted, which task scenarios produced the most diagnostic data, and the brief revision that would improve the next research round if a second test is required after a design system investment **Write every brief section and protocol guide assuming the UX researcher is experienced in research methodology but unfamiliar with design system architecture — every design consistency reference must be explained in research terms (cognitive load, mental model, trust signal) rather than design terms, so the researcher can code participant observations accurately without needing a design system background.**