EVT 2026-03-25 - Navigation shell build-ready gap review

Session scope

• Discovery root: docs/Domains/DOL English/Product Discovery/
• Lane lock: DOL English V2
• Collaboration mode: autonomous
• Objective: stress-test the new logged-in shell direction and identify which parts are principle-level, which parts still need validation before detailed IA refactor.

Key observations

1. Problem framing

• DOL English is a logged-in repeat-use learning product, not a marketing site.
• The learner returns to resume study, finish assigned work, practice by skill/program, review vocabulary, and inspect progress.

2. Strong principle already supported

• program = context
• task = primary navigation
• cross-program view must not become deep execution view
• each deep object needs one canonical home

3. Main gaps before detailed design is fully build-ready

• state model:
  • when user is in cross-program mode vs program mode,
  • whether program switch is global state or filter,
  • URL/back behavior,
  • timed-attempt switching guardrail.
• ownership model:
  • where course assignment, attempt/result, vocab items, and analytics live canonically.
• default behavior:
  • what opens for single-program users vs multi-program users.

4. Critical UX risks called out

• All Programs can regress into a pseudo-program if allowed to act as a peer execution context.
• Overview / Hôm nay can overlap with Progress / Tiến độ unless their questions are sharply separated.
• Summary surfaces can accidentally become second homes of the same object.

5. Recommended direction from review

• Keep hybrid shell:
  • program as context,
  • task as primary nav.
• Tighten it with:
  • portfolio-only All Programs,
  • explicit context switching,
  • canonical object ownership,
  • asymmetric defaults by learner archetype,
  • accessibility semantics designed at IA layer, not deferred to polish.

Open items after session

• Validate true rate of multi-program users and whether Global Today should beat resume-last-workspace for some cohorts.
• Validate whether Từ vựng deserves permanent top-level placement.
• Validate mobile shell compression separately from desktop baseline.
• Validate data readiness for consistent program + source + canonical_section ownership on surfaced objects.