Menu UX Audit

This document defines how to evaluate the interactive menu UX before changing it.

The goal is to make menu-driven work predictable for a non-expert operator:

• no hidden steps
• no misleading choices
• no internal terminology unless it is necessary
• no dead ends without a clear way back

Scope

Primary user-facing menu sections:

• Sites
• SSL
• PHP
• Database
• Diagnostics
• Logs
• Backup / Migrate
• Applications
• Profiles
• System

Priority flows:

1. Sites -> Create site
2. System -> Update simai-env
3. Applications shared daily-ops flow
4. SSL issuance/status
5. Site info / Site doctor

UX quality gate

Each flow is scored on the criteria below.

Score each criterion:

• 2 = good
• 1 = acceptable but confusing
• 0 = poor / misleading / broken

Criteria

1. Discoverability

• Can a normal operator understand where to go next from the menu labels alone?

2. Prompt clarity

• Does each question use plain language?
• Does the operator understand why the system is asking this?

3. Choice quality

• Are only realistic choices shown?
• Are unavailable options hidden or clearly marked?

4. Cancellation safety

• Can the operator back out safely at every step?
• Does cancel avoid partial or misleading state?

5. Flow continuity

• Does the command progress in a sensible order?
• Are there any surprising jumps, repeated questions, or context switches?

6. Outcome clarity

• After the command ends, is it obvious what was created or changed?
• Is the next recommended action clear?

7. Terminology

• Is the language user-facing rather than implementation-facing?
• Are engineering terms shown only when genuinely needed?

Rating bands

• 12-14: good, can stay as-is
• 9-11: usable, should be polished
• 6-8: confusing, needs redesign
• 0-5: unacceptable, should not ship

Scenario format

Each audited scenario should be documented with:

1. Entry point

• exact menu path

2. Expected operator goal

• what the user thinks they are doing

3. Actual prompt sequence

• what the menu really asks

4. Confusion points

• where the user hesitates or gets surprised

5. Recommended change

• smallest safe improvement first

6. Re-test result

• before/after score

Core scenarios

Scenario A. Create a standard CMS site

Path:

• Sites -> Create site

Goal:

• create a new site with a chosen profile, DB, PHP, and HTTPS

Minimum expected flow:

1. domain
2. profile
3. activity class
4. PHP version
5. HTTPS yes/no
6. DB yes/no when needed
7. summary + next steps

Current known risks:

• hidden defaults
• profile-specific surprises
• too many technical prompts

Scenario B. Finish a CMS install

Paths:

• Applications -> WordPress complete setup
• Applications -> Bitrix complete setup
• Applications -> Laravel complete setup

Goal:

• complete setup after web installer or bootstrap work

Minimum expected flow:

1. choose site
2. apply the expected completion step
3. show clear next steps

Scenario C. Issue HTTPS

Paths:

• SSL -> Issue Let's Encrypt
• or inline during Sites -> Create site

Goal:

• enable HTTPS without understanding certbot internals

Minimum expected flow:

1. choose site/domain
2. confirm email if needed
3. summary + validation path

Scenario D. Inspect and diagnose a site

Paths:

• Sites -> Site info
• Diagnostics -> Site doctor
• SSL -> SSL status

Goal:

• understand whether the site is healthy and what to do next

Scenario E. Pause and resume a site

Paths:

• Sites -> Pause site
• Sites -> Resume site

Goal:

• safely stop and restore a low-traffic site

Work plan

Use this order for UX work:

1. Fix flows scoring 0-8
2. Then fix misleading labels
3. Then simplify repeated prompts
4. Then improve summaries and next steps
5. Then sync docs

Current priority order

1. Sites -> Create site
2. Applications section structure
3. SSL issuance flow
4. System everyday actions
5. remaining menu polish

Notes for review

When reviewing a flow, ask:

• Would a first-time operator understand this without docs?
• Does the menu ask about the right thing at the right time?
• Is the operator forced to know an implementation detail?
• Is there a safe cancel path?
• After success, is the next action obvious?