# Match Rate Optimisation — Ownership & Strategy Analysis
**Date:** 2026-02-26 | **Author:** Harper (Finance/Legal)
**Priority:** HIGH | **Context:** Match rate is the #1 variable in our financial model. A 10-point swing shifts break-even by 3–5 months. No agent currently owns this metric.

---

## THE PROBLEM

My sensitivity analysis shows:

| Match Rate | Month 6 Hires | Break-Even Month | Cash Impact vs Base |
|-----------|--------------|-----------------|-------------------|
| 10% | 6/mo | Month 18+ | -$2,888 worse |
| **25% (base)** | **28/mo** | **Month 9** | **Baseline** |
| 35% | 53/mo | Month 7 | +$2,463 better |

**A 15-point difference between pessimistic and optimistic = $5,351 cash impact in 6 months and 11 months difference in break-even timeline.**

Currently: no agent owns match rate. Builder built the algorithm. Susan generates supply/demand. Cog monitors system health. Nobody is watching whether matches convert to hires, diagnosing why they don't, or systematically improving the rate.

---

## 1. SHOULD WE ASSIGN A DEDICATED AGENT?

**No — not a dedicated agent. Assign it as a PRIMARY METRIC to an existing agent.**

Reasoning:
- We don't have spare agent capacity (all 8 are assigned)
- Match rate optimisation is cross-functional — it requires data analysis (Cog), algorithm tuning (Builder), supply/demand awareness (Susan), and financial tracking (Harper)
- What we need is an **owner** who tracks, diagnoses, and coordinates — not someone who does it all

### Recommended Owner: **Cog**

Why Cog:
- Already monitors system health and data freshness
- Has ops mandate — match rate IS operational health
- Can pull database queries to analyse patterns
- Can coordinate between Builder (algorithm) and Susan (supply/demand) without being either
- Currently underutilised relative to capability (inbox curation, data freshness — important but not full-capacity work)

Why not others:
- **Builder:** Conflict of interest — built the algorithm, may be biased toward "it works fine"
- **Susan:** Her job is filling the pipeline, not analysing conversion. She's supply-side, match rate is full-funnel.
- **Harper:** I'm the one who identified the gap and I'll track the financial impact, but I don't have database access to diagnose match failures
- **Radar:** Intel-focused, not operational
- **Sentinel:** Infrastructure, not product metrics

---

## 2. SPECIFIC METRICS TO TRACK

### Primary Metrics (Track Daily in Week 1, Weekly After)

| Metric | Definition | Target | Alert Threshold |
|--------|-----------|--------|-----------------|
| **Match Rate** | Hires ÷ Matches Shown | 25%+ | <15% for 7 consecutive days |
| **Job-to-Match Rate** | Jobs with ≥1 match ÷ Total jobs posted | 80%+ | <50% = supply problem |
| **Match-to-View Rate** | Matches viewed by contractor ÷ Matches created | 70%+ | <40% = notification/UX problem |
| **View-to-Hire Rate** | Hires ÷ Matches viewed | 35%+ | <20% = quality/trust problem |
| **Time-to-First-Match** | Hours from job post to first match shown | <4 hrs | >24 hrs = supply gap |

### Secondary Metrics (Track Weekly)

| Metric | Definition | Why It Matters |
|--------|-----------|---------------|
| **Match Score Distribution** | Avg/median/p25 of DB match scores | Low scores = algorithm serving poor matches |
| **Rejection Rate** | Contractor rejects ÷ total matches | High = algorithm quality issue |
| **No-Response Rate** | Matches with no contractor action in 48hrs | High = notification or engagement problem |
| **Hire-to-No-Show Rate** | No-shows ÷ total hires | Quality of matched workers |
| **Repeat Hire Rate** | Contractors who hire 2+ times | Product-market fit signal |
| **City-Level Match Rate** | Match rate broken by city | Identifies geographic supply gaps |

### Funnel Visualisation

```
Job Posted (100%)
  ↓
Matches Found (target: 80%+)     ← Supply problem if low
  ↓
Matches Viewed (target: 70%+)    ← UX/notification problem if low
  ↓
Hire Completed (target: 35%+)    ← Trust/quality problem if low
  ↓
Worker Shows Up (target: 90%+)   ← Vetting problem if low
  ↓
Contractor Satisfied (target: 80%+) ← Match quality problem if low
```

Each stage has a different root cause and different owner for the fix.

---

## 3. OPTIMISATION LEVERS

### Supply-Side Levers (Susan Owns)

| Lever | Action | Impact on Match Rate |
|-------|--------|---------------------|
| **Worker volume** | More workers = more potential matches per job | HIGH — most jobs fail because no workers match the trade/location |
| **Trade coverage** | Recruit specific trades that have job demand but no supply | HIGH — a formwork job with 0 formworkers = 0% match rate |
| **Geographic coverage** | Workers in cities where jobs are posted | HIGH — Sydney job with only Melbourne workers = 0% |
| **Profile completeness** | Workers fill in skills, certs, availability | MEDIUM — better profiles = higher match scores |

### Algorithm Levers (Builder Owns)

| Lever | Action | Impact on Match Rate |
|-------|--------|---------------------|
| **Scoring weights** | Current: Trade 40%, Availability 20%, Experience 15%, Certs 15%, Location 10%, Ratings 10% | MEDIUM — weights may need tuning based on what predicts successful hires |
| **Minimum score threshold** | Lower threshold = more matches shown (but lower quality) | MEDIUM — trade-off between quantity and quality |
| **Match limit** | Currently 20 max per job. May be too many (choice overload) or too few | LOW-MEDIUM |
| **Location radius** | How far is "close enough"? Construction workers travel. | MEDIUM — too tight = missed matches, too loose = irrelevant |
| **Availability matching** | How strictly does worker availability need to match job dates? | MEDIUM |

### UX/Engagement Levers (Builder + Herald Own)

| Lever | Action | Impact on Match Rate |
|-------|--------|---------------------|
| **Match notifications** | SMS/push when matches found — speed matters | HIGH — if contractor doesn't see matches quickly, they go elsewhere |
| **Match presentation** | How matches are displayed — highlight strengths, ratings | MEDIUM — better presentation = more contractor confidence to hire |
| **One-click hire** | Reduce friction from "view match" to "pay $50" | MEDIUM — every click is a dropout point |
| **Social proof** | Show ratings, hire count, verification badges | MEDIUM — trust drives conversion |

### Pricing/Policy Levers (Harper Owns)

| Lever | Action | Impact on Match Rate |
|-------|--------|---------------------|
| **Refund policy** | "First hire money-back guarantee" removes contractor risk | HIGH — biggest hire barrier is "what if the worker is terrible?" |
| **Free first match** | Let contractor see one match free, pay $50 to unlock contact | MEDIUM — reduces commitment barrier |
| **No-show protection** | Automatic credit if worker no-shows | MEDIUM — repeat hire confidence |

---

## 4. RECOMMENDED OWNERSHIP STRUCTURE

### The Match Rate Accountability Framework

| Role | Agent | Responsibility |
|------|-------|---------------|
| **Match Rate Owner** | **Cog** | Tracks all metrics daily. Diagnoses bottlenecks. Coordinates fixes. Reports weekly. |
| **Supply Fix** | **Susan** | Recruits workers to fill trade/location gaps identified by Cog |
| **Algorithm Fix** | **Builder** | Tunes scoring weights, thresholds, matching logic based on Cog's data |
| **UX Fix** | **Builder + Herald** | Notification speed, match presentation, conversion flow |
| **Financial Tracking** | **Harper** | Tracks match rate → revenue correlation. Reports financial impact. |
| **Competitive Benchmark** | **Radar** | What match rates do competitors achieve? Industry benchmarks? |

### Weekly Match Rate Review (Proposed)

Every Monday, Cog produces a 10-line "Match Rate Brief" with:
1. Current match rate (7-day rolling)
2. Funnel breakdown (where are we losing people?)
3. Top bottleneck (supply, algorithm, UX, or trust)
4. Recommended action for the week
5. Financial impact estimate (Harper provides formula)

This goes to Rivet for prioritisation. If match rate drops below 15% for 7+ days, it auto-escalates to Michael.

### Escalation Rules

| Trigger | Action | Owner |
|---------|--------|-------|
| Match rate <15% for 7 days | Alert Michael + emergency analysis | Cog → Rivet |
| Job-to-match rate <50% | Susan: emergency worker recruitment in affected trades/cities | Cog → Susan |
| View-to-hire rate <20% | Builder: UX/trust audit on match presentation | Cog → Builder |
| No-show rate >20% | Harper: review vetting process, consider refund policy | Cog → Harper |

---

## 5. FINANCIAL FRAMEWORK FOR COG

To translate match rate into dollars, Cog should use this formula:

```
Monthly Revenue = (Job Posts × Match Rate × $48.85)

Where:
- Job Posts = f(active contractors × job post rate)
- Match Rate = hires ÷ matches shown
- $48.85 = net revenue per hire

Example (Month 3, Realistic):
- 85 contractors, 40% post → 34 job posts
- 25% match rate → 8.5 hires → $415 revenue
- If match rate were 35% → 11.9 hires → $581 revenue
- 10-point improvement = +$166/month = +$1,992/year
```

**Every 1 percentage point of match rate improvement is worth ~$17/month at Month 3 scale, scaling to ~$74/month by Month 6 (realistic scenario).** By Month 12, each point is worth ~$150/month.

---

## RECOMMENDATIONS

1. **Assign Cog as Match Rate Owner** — add to Cog's queue via Rivet. Not a new agent, just a new primary responsibility.
2. **Builder to expose match analytics** — we need a query or dashboard that shows the funnel metrics. Currently only raw DB tables exist.
3. **Week 1 post-launch: daily match rate check** — Cog reports to Rivet, Harper tracks financial impact.
4. **Pre-launch: set the minimum score threshold** — Builder should confirm what `min_score` default is used and whether it's been tested against real data (spoiler: it can't be, because we have no real hires yet. Plan to tune it after first 20 matches).
5. **Month 1: implement "first hire guarantee"** — if the first worker from a match doesn't show up or is clearly unsuitable, refund the $50. This removes the biggest barrier to a contractor's first hire. Cost: maybe $200–$500 in refunds. Value: dramatically higher conversion rate.

**The first-hire guarantee is the highest-impact lever that nobody is discussing.** Every contractor's mental model is: "What if I pay $50 and the worker is useless?" Remove that fear and match-to-hire rate jumps. I recommend we implement this before launch.