Marketplace Liquidity Improvement Plan: SF Evening Dog Walking
Executive Summary
This plan targets improving the booking fill rate from 55% to 75% for SF evening slots (5-9 PM) within 6 weeks. The core problem is a supply-demand imbalance during peak evening hours, resulting in an 18-minute p50 time-to-book (target: under 10 minutes) and a 9% cancellation rate. We will deploy a combination of supply-side incentives, demand shaping, matching algorithm improvements, and operational changes within the $25k/month incentive budget and limited engineering capacity.
1. Problem Diagnosis
Current State
- Fill rate: 55% (SF evenings)
- p50 time-to-book: 18 minutes (target: <10 min)
- Cancellation rate: 9%
- Markets: NYC, SF, LA (focus: SF)
- Core action: Request to booked within 10 minutes
Root Cause Hypotheses
- Insufficient evening supply: Not enough walkers available during 5-9 PM peak hours in SF.
- Geographic mismatch: Walkers are concentrated in certain neighborhoods while demand is spread across others.
- Pricing doesn't reflect urgency: Flat or static pricing fails to attract walkers during high-demand windows.
- Slow matching: The current matching algorithm takes too long to find and confirm walkers.
- Cancellation drag: 9% cancellations waste supply and create rebooking delays that inflate time-to-book.
Key Metrics to Track
| Metric | Baseline | Week 3 Target | Week 6 Target |
|---|---|---|---|
| Fill rate (SF evenings) | 55% | 65% | 75% |
| p50 time-to-book | 18 min | 12 min | <10 min |
| Cancellation rate | 9% | 7% | 5% |
| Active evening walkers (SF) | TBD | +30% | +50% |
| Requests per evening | TBD | Stable or +10% | Stable or +15% |
2. Strategy Overview
Four parallel workstreams, ordered by expected impact and feasibility:
| Workstream | Expected Impact | Eng Effort | Budget |
|---|---|---|---|
| A. Supply Activation & Incentives | High | Low | $18k/mo |
| B. Matching & Booking Speed | High | Medium | $0 |
| C. Cancellation Reduction | Medium | Low | $3k/mo |
| D. Demand Shaping | Medium | Low | $4k/mo |
3. Workstream A: Supply Activation & Incentives ($18k/month)
A1. Evening Surge Bonuses (Weeks 1-6)
- What: Offer $5-$10 per-walk bonus for walks accepted during 5-9 PM in SF.
- Tiered structure:
- $5 bonus for walks accepted within 5 minutes of request (speed incentive)
- $8 bonus for walks in underserved zones (geographic incentive)
- $10 bonus for walkers who complete 4+ evening walks in a week (consistency incentive)
- Budget: ~$12k/month (assuming 1,500-2,000 qualifying walks/month)
- Eng effort: Low -- configure bonus rules in existing payment system.
A2. Guaranteed Earnings Floor (Weeks 1-6)
- What: Guarantee walkers $25/hour if they commit to being available 5-9 PM at least 3 evenings per week in SF.
- Mechanism: If a walker's actual earnings fall below $25/hr during their committed window, we top up the difference.
- Budget: ~$4k/month (most walkers will earn above floor; this is insurance against slow nights)
- Eng effort: Low -- manual calculation and payout initially, automate later.
A3. Reactivation Campaign (Weeks 1-2)
- What: Identify churned or dormant SF walkers (inactive 30-90 days). Send targeted push/SMS/email with a "$50 comeback bonus for your first 5 evening walks this week" offer.
- Budget: ~$2k/month
- Eng effort: None -- use existing CRM/messaging tools.
A4. Walker Referral Burst (Weeks 1-4)
- What: Temporarily increase walker referral bonus to $75 (from standard) for SF, specifically for walkers who complete their first evening walk within 7 days of signup.
- Budget: Included in A1 envelope.
- Eng effort: None -- update referral config.
4. Workstream B: Matching & Booking Speed ($0 incremental budget)
B1. Reduce Matching Radius Dynamically (Weeks 1-3)
- What: During evening hours, pre-position the matching algorithm to prioritize walkers who are already nearby rather than waiting for the "best" match.
- Current likely behavior: Algorithm searches for optimal walker (rating, distance, preference). This takes time.
- Change: During peak hours, switch to "first available within 0.5 miles" mode. Expand radius only if no match in 2 minutes.
- Expected impact: Reduce p50 time-to-book by 4-6 minutes.
- Eng effort: Medium -- requires algorithm change, but likely a configuration/flag approach.
B2. Pre-Matching / Predictive Dispatch (Weeks 3-6)
- What: Use historical request patterns to pre-alert walkers 15-30 minutes before expected demand spikes in their zone.
- Mechanism: Push notification: "High demand expected near Marina District at 6 PM -- go online to earn bonus."
- Eng effort: Medium -- requires basic demand forecasting model + notification trigger.
B3. Auto-Accept for Trusted Walkers (Weeks 2-4)
- What: Allow top-rated walkers (4.8+ stars, 50+ walks, <3% cancel rate) to opt into auto-accept mode. When a request comes in, it's instantly confirmed without requiring manual acceptance.
- Expected impact: Eliminates the walker-side acceptance delay (often 3-8 minutes).
- Eng effort: Medium -- new opt-in flow + matching logic change.
B4. Parallel Dispatch (Week 1)
- What: Instead of sequentially offering a walk to one walker at a time (waiting for timeout before trying the next), send the request to 3 walkers simultaneously. First to accept wins.
- Expected impact: Reduces time-to-book by eliminating sequential timeout delays.
- Eng effort: Low-Medium -- depends on current architecture.
5. Workstream C: Cancellation Reduction ($3k/month)
C1. Cancellation Fee Enforcement (Week 1)
- What: Enforce a $10 cancellation fee for walker-side cancellations within 30 minutes of a booked walk. Customer-side cancellations: $5 fee if cancelled within 15 minutes of walk start.
- Current state: Likely lenient or no enforcement, contributing to the 9% rate.
- Eng effort: Low -- policy change + update in app.
C2. Cancellation Replacement Priority Queue (Weeks 2-4)
- What: When a cancellation occurs, immediately re-enter the request into matching with "urgent" priority and offer the replacement walker a $5 rescue bonus.
- Budget: ~$3k/month (assuming ~300 cancellations/month needing rescue)
- Eng effort: Low -- priority flag in matching queue.
C3. Walker Reliability Score (Weeks 3-6)
- What: Introduce a visible reliability score that factors into walker ranking and bonus eligibility. Walkers with high cancel rates get deprioritized in matching.
- Expected impact: Behavioral nudge to reduce cancellations over time.
- Eng effort: Low -- add score calculation, surface in walker app.
6. Workstream D: Demand Shaping ($4k/month)
D1. Off-Peak Discounts (Weeks 1-6)
- What: Offer $3-$5 discount for customers who book walks at 4-5 PM or 9-10 PM (shoulder hours) instead of the 6-8 PM core peak.
- Goal: Flatten the demand curve so supply can serve more requests without adding walkers.
- Budget: ~$3k/month
- Eng effort: Low -- discount code or automatic pricing rule.
D2. Scheduled Walks Promotion (Weeks 1-4)
- What: Encourage customers to schedule recurring evening walks (e.g., every weekday at 6 PM) rather than on-demand requests. Offer a 10% discount for weekly recurring bookings.
- Benefit: Predictable demand allows pre-committed supply; walkers can plan their evenings.
- Budget: ~$1k/month in discounts
- Eng effort: Low -- if scheduling feature exists; Medium if it needs to be built.
D3. Wait Time Transparency (Week 1)
- What: Show customers estimated wait time before they request. "Current wait: ~15 min. Book for 7:30 PM instead for instant confirmation."
- Expected impact: Reduces failed bookings and sets expectations; nudges demand to available slots.
- Eng effort: Low -- surface existing matching data in UI.
7. Implementation Timeline
Week 1: Quick Wins
- Launch evening surge bonuses (A1)
- Activate reactivation campaign (A3)
- Enable parallel dispatch (B4)
- Enforce cancellation fees (C1)
- Show wait time transparency (D3)
- Launch off-peak discounts (D1)
Week 2: Supply & Speed
- Launch guaranteed earnings floor (A2)
- Launch walker referral burst (A4)
- Ship auto-accept opt-in for trusted walkers (B3)
- Launch scheduled walks promotion (D2)
Week 3: Matching Intelligence
- Deploy dynamic matching radius (B1)
- Launch cancellation replacement priority queue (C2)
- Begin demand forecasting model work (B2)
Week 4: Iterate
- Analyze Week 1-3 data. Adjust bonus amounts based on elasticity.
- Tune matching radius thresholds.
- Launch walker reliability score (C3)
Week 5: Scale
- Deploy pre-matching / predictive dispatch (B2)
- Optimize incentive spend: shift budget from underperforming levers to top performers.
Week 6: Stabilize & Measure
- Full measurement against targets.
- Document playbook for replication in NYC and LA.
- Plan incentive taper to find sustainable equilibrium.
8. Budget Allocation
| Category | Monthly Budget | % of Total |
|---|---|---|
| Evening surge bonuses | $12,000 | 48% |
| Guaranteed earnings floor | $4,000 | 16% |
| Reactivation & referral bonuses | $2,000 | 8% |
| Cancellation rescue bonuses | $3,000 | 12% |
| Demand-side discounts | $4,000 | 16% |
| Total | $25,000 | 100% |
9. Measurement & Decision Framework
Weekly Review Cadence
Every Monday, review the following dashboard:
- Fill rate by hour (5 PM, 6 PM, 7 PM, 8 PM, 9 PM)
- p50 and p90 time-to-book by hour
- Active walkers online by hour and zone
- Cancellation rate (walker-side vs. customer-side)
- Incentive spend per incremental booking
- Unit economics: Revenue per walk vs. cost (including incentives)
Decision Rules
- If a lever shows <5% improvement after 2 weeks, reallocate its budget.
- If fill rate hits 70% by Week 3, begin tapering bonuses by 20% to test sustainability.
- If cancellation rate doesn't drop below 7% by Week 3, escalate enforcement (ban repeat offenders).
- If time-to-book doesn't improve with parallel dispatch, prioritize B1 (dynamic radius) engineering work.
Kill Criteria
- Abort a workstream if cost-per-incremental-booking exceeds $15 (not sustainable at typical walk margins).
- If overall fill rate regresses below 50%, pause demand-side promotions to avoid worsening the experience.
10. Risks & Mitigations
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| Incentives attract low-quality walkers | Medium | High | Gate bonuses behind minimum rating (4.5+) and background check completion |
| Walkers game the system (go online but decline requests) | Medium | Medium | Track accept rate; require >80% accept rate for bonus eligibility |
| Demand increases faster than supply (incentives attract customers too) | Low | High | Pause demand-side discounts if fill rate drops |
| Engineering delays on matching improvements | High | Medium | Front-load no-eng changes (Weeks 1-2); matching improvements are additive, not blocking |
| Budget overrun from higher-than-expected bonus qualifying rate | Medium | Medium | Set weekly budget caps with automatic bonus reduction triggers |
| Walker fatigue / churn after incentives end | Medium | High | Taper gradually; build habit through recurring bookings; improve non-monetary walker experience |
11. Success Criteria
Primary: SF evening fill rate reaches 75% by end of Week 6.
Secondary:
- p50 time-to-book < 10 minutes
- Cancellation rate < 6%
- Incentive cost per booking < $10
- No degradation in walker or customer NPS
Stretch: Develop a repeatable playbook that can be applied to NYC and LA evening slots within the following quarter.
12. Post-Campaign Sustainability Plan
Reaching 75% is necessary, but sustaining it without $25k/month in perpetual incentives is the real goal.
- Taper incentives gradually (Weeks 7-10): Reduce bonuses by 25% every 2 weeks. Monitor fill rate elasticity.
- Lock in recurring supply: Convert bonus-motivated walkers into habitual evening walkers through scheduling commitments and reliability rewards.
- Dynamic pricing: Replace flat incentives with algorithmic surge pricing that self-adjusts based on real-time supply-demand ratio.
- Expand the walker base permanently: The referral and reactivation campaigns should yield a structurally larger supply pool that persists beyond the incentive period.
- Product improvements stick: Matching algorithm changes, auto-accept, parallel dispatch, and wait time transparency are permanent improvements with zero ongoing cost.