Customer Journey Map

Current manual workflow — Broker → Admin → Seller — for a single property proposal at Auction Central

2–4 hrs
Per proposal (manual)
7
Pain points identified
6
Automation opportunities
100/mo
Current capacity cap
Broker
Admin
Seller
Time cost
Pain point
Broker
1
Sources Property
Broker identifies a potential auction property through networking, referrals, or canvassing
→
2
Sends Details
WhatsApp message or email to admin with property address, P24 link, seller name
⚠ Unstructured data
→
Broker waits...
⏱ 4–24 hrs idle
→
8
Reviews Comps
Manually reviews comparable sales tables and suburb trends from Lightstone data
⏱ 20–30 min
→
9
Sets Reserve Price
Determines valuation / reserve based on gut + comps. No systematic methodology
⚠ Subjective — no AI assist
→
11
Presents to Seller
In-person meeting. Walks through Word/PDF proposal. Answers questions on valuation
⏱ 45–60 min
Admin
idle
idle
→
3
Pulls Lightstone Report
Logs into Lightstone portal. Searches by address. Downloads 6-page PDF report
⏱ 10–15 min
→
4
Finds on Property24
Searches P24 for listing. Copies description, features, photos. Sometimes link is broken
⚠ Manual copy-paste
→
5
Opens Word Template
Opens branded .docx template. Begins filling in property-specific fields
→
6
Assembles Proposal
Pastes description, formats comps table, screenshots suburb trend charts from Lightstone PDF
⏱ 45–90 min
⚠ Bottleneck — caps at 100/mo
→
7
Sends to Broker
Emails completed Word doc or PDF to broker for review. No tracking, no analytics
⚠ No engagement tracking
Seller
Seller is unaware — no touchpoint until broker presents
No visibility into pipeline status or proposal progress
→
10
Receives Proposal
Gets a static PDF or Word attachment via email/WhatsApp from broker
⚠ No interactivity
→
12
Reviews Valuation
Examines comparable sales and reserve price. Key decision point: do I trust this number?
⚠ Trust gap — no AI transparency
→
13
Signs or Rejects Mandate
Signs the auction mandate (10-clause, 60-day) or rejects and considers alternatives

🔴 Pain Points

1. Unstructured Intake

Brokers send property details via WhatsApp messages, emails, voice notes — no standard format. Admin must parse and interpret each one differently.

Steps 1–2 · Broker → Admin handoff

2. Manual Data Assembly

Admin manually pulls Lightstone PDFs, searches Property24, copies data into Word templates. Screenshots charts instead of generating them from data.

Steps 3–6 · 45–90 min per proposal

3. Capacity Bottleneck

At 2–4 hours per proposal with 1–2 admin staff, Auction Central is hard-capped at ~100 proposals/month. Marco says they could do 200+ if this was removed.

Step 6 · Primary scale blocker

4. Subjective Valuations

Reserve prices are set by broker intuition + quick comp review. No systematic methodology, no transparent reasoning, no historical accuracy tracking.

Step 9 · Core sprint test: broker trust

5. Zero Engagement Tracking

Once the PDF is emailed, nobody knows: did the seller open it? How long did they spend? Which sections did they focus on? Complete blind spot.

Step 7 · No Qwilr-style analytics

6. Static Output Format

Proposals are flat PDFs. Sellers can't explore comparable sales on a map, interact with valuation reasoning, or sign mandates inline.

Step 10 · Lost conversion opportunity

7. No Data Capture

After auction outcomes (sold/unsold, price achieved), data is not fed back into the system. No learning loop. No market intelligence. No data moat.

Post Step 13 · Missing flywheel

🟣 AI Automation Opportunities

🤖 WhatsApp AI Intake

Replace unstructured messages with a conversational AI that asks the right questions, validates addresses, and structures data automatically.

Steps 1–2 → 60 seconds instead of ad-hoc

🤖 Auto Data Enrichment

From a property address, automatically pull Lightstone data, municipal valuations, P24 listing data, and comparable sales — no manual portal logins.

Steps 3–4 → Instant instead of 15–25 min

🤖 AI Proposal Generation

Generate the complete branded proposal — description, comps table, suburb trends chart, cover letter — from structured data in minutes.

Steps 5–6 → 3 min instead of 45–90 min

🤖 AI Valuation with Transparency

AI analyses top-5 comparable sales, suburb trends, municipal cross-check, and produces a reserve recommendation with full reasoning chain.

Step 9 → Transparent methodology, broker can override

🤖 Interactive Web Proposals

Replace static PDFs with web-based proposals: explore comps on a map, see live trend data, sign mandate inline. Track engagement in real-time.

Steps 10–13 → Qwilr-style with view tracking

🤖 Outcome Feedback Loop

After auction, capture outcome data (sold/unsold, achieved price) automatically. Feed back into AI valuation model. Build data moat passively.

Post Step 13 → Data flywheel, no extra admin work

🎯 Sprint Target Moment

The broker reviews the AI-generated proposal before presenting it to the seller.

If the broker doesn't trust the output enough to present it, nothing else matters. The broker is the gatekeeper between the AI engine and the seller.

The Friday test must answer: "Would you send this to a seller without edits?"