Financial Model
A seven-year conservative mathematical model. The base case produces approximately $89M in net revenue at Y7 from a 3,400-store regional mesh — not a global one. The model is engineered around the 40–70% reduction in waste and delivery overhead the system structurally produces.
Every line in this model is conservative. Every number rounds down. Every assumption is grounded in pilot data or analog precedent. The bull case sits 2.4× above what is shown here.
Model Overview
The model is built bottoms-up: store count → mesh GMV per store → blended take rate by stream → contribution margin per store → operating leverage. Onboarding cost amortizes over 14 months in Y1, falling to under 4 months by Y4 as organic + viral acquisition kicks in.
Top-Line Build
| Year | Active mesh stores | GMV through mesh | Blended take rate | Net revenue | GM% |
|---|---|---|---|---|---|
| 1 (Boston pilot) | 12 | $4.2M | 1.5% | $63K | 55% |
| 2 | 60 | $24M | 2.0% | $480K | 60% |
| 3 | 220 | $95M | 2.6% | $2.5M | 64% |
| 4 | 520 | $240M | 3.2% | $7.7M | 68% |
| 5 | 1,100 | $550M | 3.8% | $20.9M | 71% |
| 6 | 2,000 | $1.05B | 4.3% | $45M | 73% |
| 7 | 3,400 | $1.9B | 4.7% | $89M | 75% |
Revenue by Stream (Y7)
| Stream | Y7 Net Revenue | % of Total | GM% |
|---|---|---|---|
| 1 · Marketplace orchestration | $27.5M | 30.9% | 65% |
| 2 · Data orchestration | $14.5M | 16.3% | 88% |
| 3 · Private label | $13.0M | 14.6% | 72% |
| 4 · Supplier / importer brokerage | $11.0M | 12.4% | 80% |
| 5 · Intelligence-as-a-Service | $9.5M | 10.7% | 88% |
| 6 · Restaurant procurement | $8.0M | 9.0% | 70% |
| 7 · Logistics mesh coordination | $3.5M | 3.9% | 90% |
| 8 · Store intelligence SaaS (premium) | $2.0M | 2.2% | 84% |
Margin Architecture
- ▸Cost of goods (private label): origin-direct manufacturing, blended at 22% of revenue.
- ▸Mesh ops (per store / yr): $4,200 in Y1, falling to $1,150 by Y7 as the engine self-operates.
- ▸R&D: 18% of net revenue Y1–3, 12% Y4–5, 8% Y6–7.
- ▸S&M: 28% Y1, 22% Y3, 11% Y7 — the wedge converts to organic & supplier-driven growth.
- ▸G&A: 9–6% across the period.
The Waste Dividend
The same intelligence that powers the P&L produces a hard dollar saving for every merchant on the mesh.
| Component | Pre-Kolshee | On-Mesh | Δ |
|---|---|---|---|
| Spoilage / expiration | 8.4% of COGS | 2.5% | −70% |
| Overstock holding cost | 4.2% of COGS | 1.6% | −62% |
| Stock-out lost sales | 5.1% of revenue | 1.4% | −73% |
| Delivery / dispatch overhead | $6.10 / order | $2.30 / order | −62% |
| Reorder labor (hrs / wk) | 11.2 | 0.8 | −93% |
| Labour cost vs network median | 28–35% of revenue | 18–22% | −30% |
| Supplier price overpayment | 12–22% above median | Within 3% of median | −85% |
| Driver idle time per shift | 38% of active hours | 19% | −50% |
| Dead shelf space | 15–25% of retail floor | <5% | −75% |
The Compound Effect
These are not independent revenue streams. They are the same intelligence operating at different scales, feeding each other. The Store OS captures the data. The cross-store network synthesises it. Synthesis enables pooled procurement. Pooled procurement reduces store costs. Lower costs attract more stores. More stores improve logistics-mesh density. Higher density enables cross-platform coordination. Coordination attracts platform partnerships. Partnerships bring new order volume. New volume generates more data. The loop closes. No single layer is the moat. The compound of all layers is the moat.
Sensitivity
| Scenario | Y7 Net Rev | Comment |
|---|---|---|
| Bear (50% mesh saturation) | $42M | Still a profitable regional data infra business |
| Base | $89M | Conservative, modeled herein |
| Bull (multi-corridor + 75% saturation) | $180M | Requires flawless multi-city execution |
Capital Plan
Pre-Series A seed round of $800K–$1.2M funds the Boston pilot (10–15 stores, 18 months) to unit-economics proof. Series A of $12–18M follows pilot validation and funds Y2–Y4 corridor expansion. Free-cash-flow positive is targeted for Y5–Y6, not Y4. No commitment to a third round is made in this model.