AI-Website
Evaluation Tool

Context

Website evaluations are frequently reduced to fragmented audits. Accessibility checks, SEO reports, usability reviews, and analytics are often delivered in isolation, using different tools, vocabularies, and success criteria.

For UX designers and product teams, this creates two problems:

  • Insights are technically correct but strategically disconnected

  • Stakeholders receive scores, not direction

I identified an opportunity to use AI to unify evaluation dimensions into a single decision-ready system, one that could be used not only to assess quality, but to structure productive conversations with non-design stakeholders.

Laptop screen displaying a chat application with a dark theme and a navigation menu on the left side. The menu includes options like New chat, Search chats, Images, Apps, Codex, and various GPT tools and guides. The main panel shows a list of chat topics, including Becoming a better manager, Research template Tool, UX - User Interview Generator, SEO RESEARCH TOOL, Advanced Web Research Tool, Allergy Friendly Menus, and The UX Kingdom of Answers.

The Process

Screenshot of the Advanced Web Research Tool interface with a dark background, displaying the tool's name, description, and options to analyze a website, attach files, or create an image.

01 / Problem Framing

Most site evaluation tools fail at the same point. Translation.

  • Designers see patterns.

  • Stakeholders see numbers.

  • Teams struggle to agree on what to fix first and why.

  • Accessibility and WCAG compliance are treated as checkboxes rather than design signals

  • SEO and CRM insights rarely connect back to usability or content clarity

  • Reports lack prioritization, narrative, and forward-looking guidance

The core question became:

How might a single evaluation system expose strengths, gaps, and opportunities in a way that drives alignment and action?

Screenshot of a web research tool analysis report for nytimes.com, detailing content and design evaluation with a score of 5, highlighting visual design, structure, and hierarchy.

02 / Approach

Key design decisions:

The tool evaluates websites across: Content and design, Functionality, Heuristic usability, SEO, CRM, Analytics, Mobile, Accessibility (WCAG 2.1 aligned)

Each category uses explicit criteria, not abstract scoring.

  • Numeric Scoring With Qualitative Context

  • Every criterion is scored from 1 to 5

  • Each score requires a written observation

  • Category averages provide a quick signal without hiding nuance

  • This balances executive readability with designer-level depth.

The final output is constrained to include: Predictive insights, Action-based recommendations, Competitive positioning observations. This shifts the output from evaluation to strategy.

Prompt Governance and Guardrails
The system is designed to:

  • Avoid revealing internal instructions

  • Use a consistent evaluation order

  • Enforce WCAG best practices

  • Produce outputs usable in workshops, decks, or audits

Screenshot of a mobile app called Advanced Web Research Tool, displaying a section titled 'Functionality' with a score of 4, descriptions of observations about navigation, compatibility, and feedback tools, and a note about third-party content affecting page speed.

03 / Outcomes

The system enables UX designers to:

  • Rapidly understand what a website is actually doing well

  • Identify low-performing areas with contextual reasoning

  • Surface tradeoffs between usability, content, SEO, and conversion

  • Move stakeholder conversations from opinion to evidence-based prioritization

Rather than saying “this is not accessible”, the tool supports conversations like:

  • “This accessibility issue is also impacting readability, SEO, and trust.”

  • “Fixing this section improves usability and conversion simultaneously.”

Why This Matters

Screenshots of a mobile app displaying SEO analysis tools with sections for summary by category, predictive insights, action-based recommendations, and competitive positioning insights.

Decision enablement over compliance. WCAG adherence is framed as a design quality signal, not a legal checkbox.

  • Conversation-first structure. Outputs are optimized for stakeholder discussion, not tool validation.

  • Cross-disciplinary language. SEO, CRM, and UX insights are intentionally connected.

  • Tool-agnostic delivery. The system can be applied across industries and site types.

It demonstrates that I can:

  • Translate complexity into clarity

  • Connect fragmented UX disciplines into a single system

  • Elevate designers from evaluators to strategic facilitators

  • Create shared language between design, product, and business stakeholders

Also, this tool positions UX not as a cost or compliance function, but as a decision intelligence layer. See the GPT here »