01 / RUSHED COMMUTER“I have two minutes.”
Find a nearby store, opening hours, and the quickest route to pickup information.
Speed · Findability · Clarity02 / KEYBOARD-ONLY USER“Let me navigate.”
Reach useful homepage controls with the keyboard, see focus, and understand what each control does.
Navigation · Focus · Labels03 / FIRST-TIME CUSTOMER“Where do I start?”
Understand the offerings and find a clear next step without assuming brand familiarity or an existing account.
Orientation · Trust · Next step Copy one prompt into your AI coding agent
Use an agent with browser access. It should inspect the actual homepage and keep each persona’s goals separate.
Review the current https://www.starbucks.com/ homepage from three AI-simulated personas:
1. A rushed commuter with two minutes to find nearby-store, opening-hours, and pickup information.
2. A keyboard-only user who needs to navigate the homepage, see keyboard focus, and understand control labels without using a mouse.
3. A first-time customer who wants to understand the offerings and the next step without an account or prior brand knowledge.
Inspect the page with your available browser tools. Capture the viewport and relevant screenshots. For the keyboard-only persona, actually try Tab, Shift+Tab, and appropriate keyboard activation on safe homepage controls. Do not infer keyboard usability from a screenshot. If you cannot interact, clearly identify what remains untested. Keep the review focused on the homepage; note where a link goes without reviewing the full ordering journey.
For EACH persona, return:
- A short first-person qualitative impression, explicitly labeled AI-simulated feedback.
- What works well, what causes friction, and which visible elements support that feedback.
- Separate 1–5 star ratings for clarity, ease of use, and trust. Use 1 = serious friction, 3 = mixed, 5 = clear and effective. Explain each rating with evidence. Use “not assessed” when evidence is missing; do not invent a rating.
- The three most useful improvements for that persona, ranked by impact.
- Observed issues versus assumptions or questions to investigate.
Finish with a comparison table: persona, goal, clarity stars, ease-of-use stars, trust stars, and top concern. Identify shared concerns and places where the personas disagree. Do not force every persona to find a problem or give different ratings.
Treat ratings as qualitative AI judgments, not measurements from real customers or proof of WCAG compliance. Do not invent interviews, demographic stereotypes, or completed actions. Browse public content only: no sign-in, personal information, orders, payments, or intrusive tests. Stop at access restrictions or CAPTCHA.
This page only supplies the exercise prompt. It does not call an AI service. Persona feedback is simulated—not a substitute for feedback from real users.