From rampstack-skills
Audits brand, UX, and site design of leading sites in a vertical to produce cross-site experience patterns and the design bar a new build must meet or beat.
How this skill is triggered — by the user, by Claude, or both
Slash command
/rampstack-skills:competitor-experience-auditThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Capture the brand, UX, and site design that the leading sites in a vertical observably share, and where they fall short. The output is the experience bar a new build must meet or beat, grounded in cross-site patterns rather than aesthetic opinion.
Capture the brand, UX, and site design that the leading sites in a vertical observably share, and where they fall short. The output is the experience bar a new build must meet or beat, grounded in cross-site patterns rather than aesthetic opinion.
Stack-agnostic. Works on any SiteShape (ecommerce-catalog, inventory-listing, hospitality-food, b2b-manufacturer, and so on). Auto-parts catalogs are the first test case, not the only target.
This skill is the brand / UX / design counterpart to a technical audit like seo-onpage or accessibility-audit. Technical findings are objective (a missing canonical is missing). Design is more subjective, so this skill is deliberately written to make the subjective assessable: every dimension is judged by observable questions about what the leading sites actually do, reported as patterns and gaps, and marked not_assessable whenever the evidence does not support a judgment.
design-standards or brand-style-guide for production design decisions on a known build)seo-competitor)creative-direction and brand-discovery)ux-research, usability-testing)seo-onpage, accessibility-audit)If the list of leading sites is unknown, ask. Picking the wrong field produces an experience bar against the wrong vertical.
Some dimensions can be partially assessed from static HTML (primary navigation, the catalog/category surface count, the presence of search). Most cannot. Layout density, brand register, motion, trust-signal prominence, and the rendered hierarchy of the first viewport all need the rendered page to judge honestly. If you only have static HTML for a site, mark every rendered-only dimension not_assessable for that site rather than guessing.
Each dimension is judged by observable questions across the leading sites in the field, then reported as a cross-site pattern plus the gap (where the leaders converge, and where even they fall short).
Score each dimension as Pattern present (the field converges on a recognizable convention), Mixed (the field splits across two or three approaches), or Gap (no convention emerges, or every leader misses it). Mark not_assessable when the evidence cannot support a judgment.
Never write Good design or Bad design or any aesthetic verdict. The dimensions are about what the field observably does, not whether you like it.
Does the site lead with the visitor's primary job, or bury it?
Observable signal: count the elements that compete for first attention, identify which one the layout actually privileges, name what the field converges on as the primary task.
Is the layout retail-dense, editorial, or marketing-airy? Cross-reference creative-direction aesthetic positions (editorial restrained, polished standard, controlled maximalist, expressive maximalist) for vocabulary; do not redefine that here.
Observable signal: count modules above the fold across the field; describe the register the leaders share, not whether you find it attractive.
How much of the catalog or offering is surfaced at once, and in what shape?
Observable signal: count the category entry points and describe the navigation depth pattern the field uses.
Do the sites serve both the know-what-I-want path and the browse path?
Observable signal: list the paths the field offers, score whether each path is primary, secondary, or missing.
What brand posture do the leading sites take, and is it consistent across the page?
Cross-reference the creative-direction and brand-archetype-system skills for the vocabulary. Report the register the field converges on. Do not rate the taste; describe what posture the leaders take.
What recurring trust and conversion conventions does the field use?
Observable signal: which signals appear across the field, where they appear, and which the leaders share.
The patterns that recur across the leaders and define what a credible site in this vertical is expected to have. This is the synthesis of dimensions 1 through 6 into a short list of "if your build does not do this, it will read as off-vertical."
Observable signal: which conventions recur on 3 of N or more of the leading sites; which are absent across the field (the gap that a positioning opportunity exploits).
This is not negotiable. The credibility of an experience audit erodes the moment it slides into taste:
art-direction or design-standards, not here.not_assessable for that site or that pattern. Common cases: layout and density cannot be judged from un-rendered static HTML; motion and animation cannot be judged from a screenshot; brand voice cannot be judged from a single page.If the user asks "is this design good", redirect to observable patterns ("the leading sites in the field share X, Y, Z; this build does X and W but not Y or Z; that is the gap"). If the user asks for an aesthetic call, point them at design-standards or art-direction.
references/experience-dimensions-checklist.md to score each site on each dimension. Mark not_assessable wherever the evidence is thin.references/audit-template.md. Default output is a markdown experience bar, generalized to pattern-level if it will feed a public showcase workup (no named competitors in the public version).creative-direction, design-standards, information-architecture, and the relevant build skill (landing-page-copy, frontend-component-build) read as the experience bar to meet.When you spot one of these, push back before delivering.
not_assessable and get the rendered view. Static HTML supports only partial assessment; do not guess at design quality you cannot see.partial and call out what is assessed vs what is not.Default output is a markdown experience bar at experience-audit-[vertical-slug].md in the project root. Structure:
Not_assessable items (what could not be judged and why)Keep the bar under 1500 words. If a dimension needs deeper treatment, link to a deeper appendix rather than expanding inline.
If the audit will feed a public showcase workup, scrub for named competitors before publishing. The named version stays operated-side; the public workup carries the patterns and the gaps in generalized form. This mirrors how Basano's competitor review already handles named sites in its model-backed audit.
references/audit-template.md - Fillable experience-audit template, copy and use. Generalized and named variants both present.references/experience-dimensions-checklist.md - The observable questions per dimension, in scoring order.npx claudepluginhub rampstackco/claude-skills --plugin rampstack-skillsBenchmarks design patterns and UX conventions from category-leading competitors. Use during redesigns to identify industry conventions and differentiation opportunities.
Composes pages and sites to vertical-specific layout, density, and merchandising conventions so builds read as credible category members rather than generic SaaS pages.
Pesquisa concorrentes, analisa tendências visuais de nicho, gera moodboards e produz dossiê estratégico para UI/UX. Use para benchmarking visual, redesign ou direcionamento de interface.