LoopcardFinancial plan — 18 September 2026
Bar used in the challenge pass: ~$5,000 MRR by month 12 on one VPS, PLG only, no enterprise sales.
Result: no killed idea clears that bar on a base case. Slip’s stretch case required six named levers and still sat on unproven WTP.
This is not audited accounting. VPS is already paid (sunk).
1. What “viable” means here
At self-serve prices:
| Price | Paying accounts to hit $5k MRR |
|---|---|
| $19/mo | 263 |
| $29/mo | 173 |
| $49/mo | 102 |
| $99/mo | 51 |
cryptobook.space: ~65 unique IPs / week. Even a heroic 5% visitor→paid on a real product would not create 100 paying accounts from this domain in a year.
2. Shared cost sheet (non-LLM-hot-path SaaS)
| Line | At ~150 paying / ~4k free | Note |
|---|---|---|
| VPS | $0 incremental | Already running TMC / Virtus / CryptoBook |
| Bandwidth | $0 unless we proxy LLM streams | Do not become Helicone |
| Disk | $0 if hashes/receipts only | Full traces become a storage problem |
| LLM we pay | $0–40 if English compile is local; $400–2,000 if we run a tutor | Do not eat GPU or frontier API |
| Stripe | ~4–8% of MRR | $19 SKU is Stripe-heavy |
| Support | the real COGS | Proxy/auth products run hot on founder hours |
Gross margin on a decision API with no token proxy: ~85–92% after Stripe. Nobody dies on COGS. They die on WTP and distribution.
3. Idea P&L (challenge pass, ideas folder empty → BRIEF A–D + patches)
| Idea | Verdict | M12 MRR base | Clears $5k? |
|---|---|---|---|
| A MCP permit gateway @ $29 | FIX as specified | $1.4–2.2k | Only after changing the SKU; still no demand |
| B agent-permit.yaml GitHub App | KILL | $0.3–0.8k | No |
| C freelance receipts | KILL | $0.05–0.3k | No |
| D quantum classroom | KILL | $0–0.4k | No |
| Langfuse clone | KILL | — | Instant |
| Slip / Permit Inbox stretch | Demand KILL | Finance once modeled ~$4k base / $7–9k stretch | Stretch assumed WTP that the demand pass refuted |
Do not use the Slip stretch model. It assumed people would pay for unwatermarked receipts. ShieldMCP’s actual paid object is a $49 PDF with 5 PH upvotes.
4. Holding-page economics (recommended spend)
| Item | Amount |
|---|---|
| Design/implement holding page | Founder / this VPS, hours only |
| Ads | $0 |
| Domain cryptobook.space | Already held |
| Stop old cluster later | Frees ~4 Node workers × up to 2 GB heap each — optional ops, not revenue |
5. If a later GO is an application SaaS
Re-build this file with:
- Price, conversion (use 2–3.5% free→paid if OSS alternative exists; 4.5% median freemium otherwise)
- Churn 4–8%/mo for set-and-forget tools
- Support hours cap: if >10 h/week under $5k MRR, the SKU is wrong
- Stripe + any LLM COGS
- Distribution not equal to cryptobook.space residual traffic
Until then, the financial plan is: do not spend engineering on a SKU with no WTP.