goal-directed-design-and-research
Ground design in user goals, personas, and observed behavior, and validate candidates with users.
SKILL.mdsubagents/ui-design-advisor/skills/goal-directed-design-and-research/SKILL.md
GitHubGoal-Directed Design and Research
Ground design in user goals, personas, and observed behavior, and validate candidates with users.
When this applies
- You need to characterize an audience or validate assumptions about them (P005).
- Synthesizing research into a design-guiding model of the audience (P009).
- Defining or evaluating what a product should do (P011).
- Building a persona set from research (P012).
Procedure
Apply these principles to the situation under review; for each, name the user goal at stake and the trade-off the choice carries.
- Use empirical user research, especially qualitative observation and interviews, to understand users' behavior, expectations, vocabulary, and mental models; use quantitative methods for validation rather than as a substitute for design insight. (P005)
- Identify and prioritize user goals before tasks, and base the persona hypothesis on likely behavior patterns and their differentiating factors rather than purely on demographics, which should serve only as a proxy for behavior. (P009)
- Keep users' goals as the bedrock throughout the pressures of the development cycle, treating features and functions as a limited way to define a product, because a goal-directed process supplies a clear rationale that makes the design not guesswork or personal preference. (P011)
- Synthesize personas from observed behavior rather than fiction, keeping detail no deeper than the research supports, inferring goals from behavioral connections (about three to five end goals, zero or one life goal, zero to two experience goals), making each persona distinct in at least one significant behavior, expanding into a one-to-two-page narrative, and remembering personas are design tools, not an end in themselves. (P012)
- Use ethnographic interviews by observing and interviewing users in their own context, avoiding fixed or leading questionnaires and solution proposals, eliciting specific stories, keeping recording unobtrusive, and verifying interpretations with users. (P015)
- Use usability testing to evaluate a concrete candidate design, not to create it: do qualitative research before ideation, test late enough to have something real but early enough to change it, and make findings measurable and actionable. (P021)
- Run the five-step iterative Requirements Definition, connecting business to usability in a problem and vision statement, brainstorming to clear preconceptions, recording each persona's expectations, keeping context scenarios broad, shallow, and textual, pretending the interface is magic in early stages, and extracting requirements as objects, actions, and contexts. (P025)
- Involve the programming team intimately throughout Refinement and construction, deliver a form and behavior specification detailed enough to code from, and stay vigilant that the design vision is translated faithfully into the final product. (P032)
- Prefer personas over user roles, profiles, and market segments, and augment them appropriately: create customer and served personas for non-users, treat a user profile as a persona only if it is research-based, use market segments merely to filter interview scope, and fall back to clearly labeled provisional personas, drawn as sketches and documented, only when rigorous research is impossible. (P049)
- Use persona-based scenarios, especially key path and validation scenarios, to refine and validate interaction design, focusing on likely paths before necessary variants and edge cases; use cases are mainly for late validation. (P051)
- Design in short buildable cycles, ship the smallest useful version first, and avoid representing future functionality as present capability. (P081)
- Decide honestly whether most users will be perpetual beginners, occasional users, or experts, and design accordingly: optimize occasional/one-time interfaces for learnability (simplified, wizards, explained along the way) and frequent/expert interfaces for efficient operation. (P091)
- Optimize the interface for perpetual intermediates, the large majority of users, devoting the bulk of effort to them while still letting beginners and experts be effective, with the threefold goal of moving beginners quickly into intermediacy, not obstructing intermediates who want to become experts, and keeping perpetual intermediates happy. (P092)
- Support beginners without penalizing others: make beginner help removable rather than fixed in the interface, give beginners overview information such as scope and a guided tour rather than reference, reflect the user's mental model so concepts are grasped quickly, and treat users as intelligent but very busy by giving brief, targeted instruction plus cause-and-effect understanding. (P093)
- Model three user-goal types, experience, end, and life goals, mapping to Norman's visceral, behavioral, and reflective levels: meet end goals for the product to be worth users' time and money, never egregiously violate experience goals since that alone dooms the product, and address life goals to turn a satisfied user into a loyal one. (P094)
- Use narrative and persona-based scenarios as a primary creative design tool, with the persona serving as the tangible agent that keeps scenarios focused on goals rather than only tasks, and keep early solutions at the level of sketchy plot points. (P095)
- Expect users to change goals mid-task and defer choices: keep choices available, do not lock users into a choice-poor environment without good reason, support reentrance (resume without losing data), ask only the minimum upfront, and let users return to deferred fields later. (P096)
Principles applied
- P005 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P009 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P011 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P012 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P015 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P021 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P025 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P032 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P049 (medium) — grounded in claims/evidence and chunk anchors in
sources/. - P051 (medium) — grounded in claims/evidence and chunk anchors in
sources/. - P081 (medium) — grounded in claims/evidence and chunk anchors in
sources/. - P091 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P092 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P093 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P094 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P095 (high) — grounded in claims/evidence and chunk anchors in
sources/. - P096 (high) — grounded in claims/evidence and chunk anchors in
sources/.
Provenance
Grounded in principles P005, P009, P011, P012, P015, P021, P025, P032, P049, P051, P081, P091, P092, P093, P094, P095, P096, their backing claims and evidence records, and paragraph-level source anchors under sources/anchors/. Every cited id resolves into this package's distilled spine; see provenance-ledger.md and reports/faithfulness-report.yaml.