← All materials

Product: "One Alumnus, One Record"

A self-hosted Alumni Data Hub with baselines Robbie can inspect

What it is

A self-hosted hub in the AAO office that keeps each alumnus's record current and links every event and form submission back to that record. It includes an on-prem, human-approved AI assistant for the matching and data-entry work staff handle manually today.

What it does (5 jobs)

  1. Keeps records current. LinkedIn refresh where legally possible, plus a one-tap "Still at [employer]?" confirmation at login.
  2. Returns event data to the record. Every sign-up source (CEMS / art-mate / PFS) reports who registered / who actually attended back to the alumni record.
  3. Auto-fills common alumni tasks. Start with parking: 10 retyped fields → 3 taps.
  4. AI assistant, self-hosted and human-approved. It proposes matches and duplicate merges. Staff approve each one. PII never leaves campus.
  5. Optional event layer: styled / dynamic QR + wallet identity. Useful for events and alumni networking. A plain gate QR only proves entry.

Where it can go after parking: PolyU's current portal asks alumni to "update your record". This gives them services: parking, events, bookings, benefits, a real digital card, personalized news. All hang off one record (reference: SCAA member portal on AVEEGO; EventBinder's own Alpha student platform). Start with parking; grow the hub. The record underneath is the defensible part. App screens are easy to copy.


How we measure it

Baselines get measured in the paid discovery (Week 1–2). Targets are what we commit to beat and report on a live dashboard. We do not quote an improvement % as fact before the baseline exists.

# Outcome The metric (measurable) Today / baseline Target
1 Record freshness % of records confirmed-current in last 12 mo; monthly confirmation volume ~1% at launch (4k in 10 days); ongoing rate set in discovery +30–50% records current within 12 mo
2 Event data reaches the record % of event registrations auto-linked to a record; % attendance in the record < 24h ~0% automated today (manual / none) ≥ 90% linked · attendance < 24h
3 Manual matching (Robbie's driver) staff-hours/week on matching + data entry set in discovery (hrs/week) −50 to −70% via AI-proposes / human-approves
4 Parking friction (the visible proof) fields typed per request; completion time; abandonment 10 fields typed every time 3 fields (−70%) · time & drop-off measured in pilot
5 Data quality duplicate rate; % of records with a verified identity link set in discovery −X% duplicates · +X% verified
6 Consent coverage % of contactable records with current PDPO / Part VIA consent set in discovery (likely thin) ≥ target% with valid consent
7 Speed to record time from event/form → data in AMS days / manual / never < 24 hours

North-star metric

An "Alumni Record Health" score = freshness × linkage × consent, shown monthly. One number that moves from baseline → target and tells Robbie whether the system is working.


Rule for the room

Discovery's first job is to measure the real baselines. How many records are actually current? What share of event sign-ups ever reaches the record? How many staff-hours go into manual matching? Once measured, every target above becomes a contract number we can report against. No vanity metrics. No unverified percentages in front of Robbie.