Private working brief · not linked from the homepage
Points & Prompts — Business Brief
Cost of service · revenue projections · exit path. As of July 31, 2026.
Figures labeled Measured (live system / fetched doc), Reported (secondary source), or Assumption. Codex Round 50 independently verifies the Reported set.
Snapshot
The business is distribution-constrained, not capital-constrained: costs barely matter to the model (doubling Plaid's price costs about 3 margin points). Traffic and conversion are everything. Strategy ruling (July 31): Jeremy operates; build toward a sale rather than holding a small-cap annuity. The automation keeps carry cost near zero while the traffic thesis gets tested.
Product ladder (owner-ratified)
| SKU | Price | Terms that matter |
|---|---|---|
| Free | $0 | Wallet, programs, 3 AI conversations/mo (server-metered, built, audited) |
| One-time Spend Analysis | $20 | Plaid (365+ days requested) or 3 statements. Coverage measured before charging. Under 90 days: no charge, no partial answers |
| Rerun | $10 | Unlimited frequency, one active run. Delta-check first: no meaningful new data means no charge |
| Monthly | $5/mo | $60/yr equivalent |
| Annual | $48/yr | 20% off monthly-equivalent · live site still shows $45, copy fix queued |
| Annual via credit | $25 yr 1 | The $20 analysis fee credits fully toward the annual plan (annual only) |
| Pro reruns | included monthly | Ad-hoc between: $5. Pricing provisional, see fee note below |
Ratified guardrails
- "No AI" for the analysis = no AI recommendations; disclosed model-assisted statement transcription allowed.
- Plaid disconnects after each analysis; raw source data deleted immediately; results kept until user deletes.
- Never stored: card numbers, loyalty-program numbers, passwords.
- Automatic refund on system/coverage/cleanup failure.
Unit economics
Plaid modeled at $0.45/customer-mo (Reported ~$0.30/item × 1.5 items; replaced by real pricing after the Aug 4 Plaid call). Stripe 2.9% + $0.30. Model compute is effectively free (Measured: $0.000235/chat).
| SKU | Revenue | COGS est. | Margin |
|---|---|---|---|
| One-time analysis | $20.00 | $1.40 | 93% |
| Rerun | $10.00 | $1.09 | 89% |
| Ad-hoc Pro rerun | $5.00 | $0.95 | 81% |
| Annual (direct) | $48.00 | $7.19/yr | 85% |
| Annual (credit conv., yr 1) | $25.00 | $6.52/yr | 74% |
The delta-check-first rerun rule is good economics, not just ethics. A ~$0.45 refusal that prevents one chargeback (~$15 fee) pays for 30+ refusals.
Infrastructure & fixed costs
| System | Current | Upgrade trigger → action | Cost |
|---|---|---|---|
| Supabase | FREE (Measured) | Before first paid customer → Pro; verify backups/PITR | $25/mo |
| Cloudflare Pages | Free | Only if build volume explodes (500/mo limit) | $0 |
| Model API | Pay-per-use, $200/mo backstop | Raise backstop with revenue | ~$1–5/mo |
| Resend | Free tier | ~2K+ emails/mo → paid | $0→$20/mo |
| Stripe | Per-transaction | n/a | 2.9% + $0.30 |
| Plaid | Pay-as-you-go (call Aug 4) | Volume discounts later | TBD |
Revenue projections — year 1
Traffic ramps linearly to target over 12 months. Conversion assumptions are the whole model. Everything else is arithmetic.
| Assumption | Conservative | Base | Optimistic |
|---|---|---|---|
| Uniques/mo (month 12) | 2,000 | 10,000 | 40,000 |
| Visitor → $20 analysis | 0.4% | 0.7% | 1.0% |
| Analysis → annual (credit) | 20% | 25% | 30% |
| Rerun attach (yr 1) | 15% | 20% | 25% |
| Year-1 result | Conservative | Base | Optimistic |
|---|---|---|---|
| One-time customers | 53 | 462 | 2,640 |
| Annual subscribers | 17 | 149 | 924 |
| Revenue | $1,716 | $14,636 | $85,536 |
| Gross profit | $977 | $12,357 | $74,463 |
| Month-12 run rate | ~$248/mo | ~$2,078/mo | ~$11,960/mo |
Cheapest tested growth lever: the "validate the Reddit advice" wedge (live-tested, owner-approved, in backlog).
Exit plan: the point of the whole exercise
Owner intent: sell rather than hold. The target is a clean handoff so the next thing gets built. Market context (Reported, Jan 2026): micro-SaaS trades at 2.5–4× ARR / 4–6× SDE; $50–300K marketplace deals close in 30–60 days, and 2026 buyers reward profitable cash flow.
What a sale is worth
| Exit at ~month 18–24 | Trailing ARR | Marketplace (2.5–3.5×) | Strategic |
|---|---|---|---|
| Conservative traction | ~$3–5K | $10–30K asset sale | — |
| Base traction | ~$25–35K | $65–120K | $150K+ |
| Optimistic traction | ~$145–180K | $360–630K | $500K–1M |
Blunt read: a worth-it exit (>$250K) requires near-optimistic traction for two to three quarters. Below that it is a $50–120K flip. The mitigant: carry cost is about $45/mo plus low documented owner-hours, so holding the option while the thesis gets tested is nearly free.
What buyers pay premiums for here
- The verified-data pipeline + honest-receipts architecture (no content site has this)
- The AwardWallet data relationship. Assignability is the contract term to secure now
- Prerendered SEO surface; transferable crawl assets
- Provably low owner-dependence: the autonomous Claude↔Codex ops relay, with logs
- A diligence-ready audit trail: 50 numbered audit rounds, immutable verdicts, negative-controlled tests
- Plaid production access (pending) — approvals are assets small buyers can't easily get
Likely acquirers
| Tier | Who | Why they buy |
|---|---|---|
| 1 · Strategic product | AwardWallet (existing data partner and the first conversation), CardPointers, Kudos, MaxRewards, Travel Freely, point.me, Roame, Seats.aero | Consumer advice layer / analysis engine on their data or search |
| 2 · Media platforms | The Points Guy (Red Ventures), NerdWallet, Bankrate, Upgraded Points, 10xTravel, Frequent Miler, Daily Drop | SEO + a real tool; the verified-data pipeline is the pitch |
| 3 · Financial buyers | Acquire.com individuals ($50–300K sweet spot), micro-PE (≥$10K MRR) | Cash-flowing asset, fast close |
The plan
- Phase 0 (now): LLC + every account papered to the entity from the start; AwardWallet assignability review; MRR/churn instrumentation; owner-hours logging. All of it serves operations anyway.
- Phase 1 (launch → month 9): prove conversion; quarterly metrics snapshots into the data room.
- Decision gate at about $2K MRR: (a) list on Acquire.com at ~3× ARR, take $65–120K, and move on; or (b) push 2–3 more quarters to $8–12K MRR and open Tier-1 conversations, targeting $350K+. Marketplace listing stays the BATNA.
- Broker: not under ~$500K. Marketplace plus a deal lawyer at close.
Decision log (July 31)
- Jeremy operates; wife-handoff / women-owned certification dropped (certification only ever made sense with genuine operator control)
- Pricing: $5/mo · $48/yr · $20 analysis ($20 credit → $25 yr-1 annual) · $10 reruns unlimited/one-active/delta-check-first · Pro monthly rerun + $5 ad-hoc
- Coverage: 90-day floor, ≥365-day Plaid pulls, 3-statement fallback, charge only after coverage clears
- Transcription: model-assisted allowed, disclosed; recommendations stay non-AI
- Retention: user-controlled results; raw data deleted; no card/loyalty/password storage ever
- LLC before charging; sale-readiness items start now
Open items
- Aug 4, 12:30 PT Plaid pricing call: replace the $0.45 placeholder and re-run the model (scripted)
- Codex Round 50: independent verification of every Reported figure + upgrade-trigger table
- $5-rerun fee structure: keep vs prepaid credits
- Supabase Pro upgrade + backup verification (launch blocker)
- Pricing copy fix on the live site ($45 to $48); owner approves wording
- LLC formation + AwardWallet assignability review