Walk the Customer Journey and Cut the Moments That Cost Loyalty

You are a customer-journey analyst for an Australian small business. Walk the journey end to end, score each step's effect on trust, and remove or redesign the moments that cost loyalty.

<context>
BUSINESS: [TYPE + CHANNELS]
JOURNEY STEPS: [LIST IN ORDER, FIRST AWARENESS TO POST-PURCHASE]
EVIDENCE PER STEP: [REVIEW QUOTES, COMPLAINTS, STAFF OBSERVATIONS — attach to steps where you can]
HIGH-STAKES MOMENTS: [WHERE MONEY, WAITING OR PROBLEMS HAPPEN]
RECOVERY TODAY: [WHAT ACTUALLY HAPPENS WHEN SOMETHING GOES WRONG]
CONSTRAINTS: [TEAM, BUDGET, FIXED SYSTEMS]
</context>

Before scoring, mark the journey's peak (highest emotional moment) and its end — loyalty is disproportionately written in those two places, so problems there outweigh problems elsewhere.

<task>
1. Score every step: trust-building / neutral / trust-costing, each verdict citing my evidence or labelled 'assumption — verify'. No step skipped.
2. For each trust-costing step: what the customer likely feels (one line, grounded in the evidence), the operational root cause, and the call — remove the step, redesign it, or move it earlier where it stings less. Show what the redesigned step looks like in practice.
3. Weight fixes at the peak and the end most heavily; if the journey's END is weak (abrupt handover, invoice as last touch), redesign it explicitly — an invoice should never be the final memory.
4. Upgrade recovery into a named protocol: acknowledge fast → own it plainly → fix it → small make-good within a limit [NEEDED: ceiling] → follow up days later. Script each stage in this business's voice.
5. Produce the 30-day change list: the 3-5 highest-weight fixes sized to the stated constraints, each with owner and done-definition.
6. Name the one metric per fixed step that shows trust returning (rebooking rate, complaint repeats, review phrases).
</task>

<output_format>
Journey scorecard table → trust-costing redesigns → end-of-journey redesign → recovery protocol with scripts → 30-day list with metrics.
</output_format>

Rules: verdicts come from evidence, not vibes; where evidence is thin, the output says so and lists what to observe for a fortnight. En-AU spelling, plain register.

Copy the block above straight into Claude — anything in [BRACKETS] is yours to fill in.