Delta Brief
Delta Brief turns two accounting CSV exports into a client-ready month-end variance brief, with a ranked movers table, a waterfall chart, and footnoted commentary, in under five minutes.
One product, one founder, thirty days. Everything below is built to be executed inside the contest window — the official rules still decide what counts.
Decide in 20 seconds
The facts that decide whether this is worth thirty days of yours.
Solo or 2-5 person fractional CFO and outsourced-accounting practices in the US, UK, Canada and Australia running 5-30 SMB clients on QuickBooks Online or Xero, during the first ten days of the month.
Operational simplicity 4.5/10
Remove regulated data, human review steps and custom infrastructure; keep the product self-serve and stateless on managed services.
Why this could work in 30 days: one founder can ship the MVP in 5 days; the plan targets a first sale on day 11; 21 sales at $49 clears $1,000 of collected revenue; strongest signal is speed to first revenue at 7.6/10.
Score and projections come from the generated report and are not guarantees. The $100K Hackathon is run by Ship or Die — read the official rules before you build.
Your first 24 hours
12 concrete actions. Tick them off as you go — nothing is saved or sent.
Product opportunity and positioning
Target customer
Solo or 2-5 person fractional CFO and outsourced-accounting practices in the US, UK, Canada and Australia running 5-30 SMB clients on QuickBooks Online or Xero, during the first ten days of the month.
Why now
Month-end close is a fixed calendar event, the input files already sit on the buyer's desktop, and 2026 model APIs make a footnoted narrative cost under $1 per brief to produce.
Problem
Every close the practitioner re-exports two P&Ls per client, rebuilds the variance table in Excel and rewrites near-identical commentary — 2-4 unbilled hours per client, every month.
Solution
Upload prior and current P&L or trial balance exports, confirm column mapping once per client, get a deterministic variance table, ranked movers, waterfall chart and line-cited commentary you edit inline, then export a PDF and read-only link.
Delta Brief — 30-Day Launch Plan
JEV score: 61.9/100 (band C, confidence 0.67). Window: Nov 16 2026 13:00 UTC → Dec 16 2026 13:00 UTC. One founder, zero paid marketing.
1. Product opportunity and positioning
Delta Brief sells to a named person on a named calendar date. A solo or 2-5 person fractional CFO practice runs 5-30 SMB clients on QuickBooks Online or Xero, and in the first ten days of every month it owes each client a month-end pack. Building that pack means re-exporting two P&Ls, rebuilding a variance table in Excel, and rewriting commentary that is 80% identical to last month's. It takes 2-4 hours per client and is the least billable work in the engagement.
The product is narrow on purpose: two files in, one PDF out. No warehouse, no OAuth sync in this launch, no team roles, no budget-vs-actual. That is the difference from the BI category it descends from and from generic chat tools: variances are computed deterministically in TypeScript before any model is called, and the model may only narrate figures that already exist in the variance table. A sentence carrying an unsupported number is rejected at validation, not caught by a human reviewer.
JEV scores speed to first revenue highest (7.58) — the buyer has the input files locally, a practice credit card, and a deadline. It scores operational simplicity lowest (4.46), so this plan keeps the product stateless: files held per session, raw financial exports deleted within 24 hours of export, no OAuth tokens, no human review queue, no 24/7 support promise, managed services only.
Why now: narration constrained to a computed table costs under $1 per brief in 2026, so a $49-$149 one-time price clears cost of delivery by a wide margin, and the buyer's alternative is 2-4 unbilled hours.
2. Smallest sellable MVP
Upload two P&L exports → confirm column mapping → see the variance table → get a one-page PDF with waterfall, top-10 movers and line-cited commentary → edit sentences inline → pay one credit to export. That is the whole product at day 5.
Explicitly excluded until after the 30 days: QBO/Xero OAuth, multi-entity consolidation, XBRL, SSO, approval workflows, forecasting, white-label portals, scheduled email delivery. Column mapping is manual on first use and remembered per client profile thereafter — the founder does not build an auto-detector before knowing which export formats real buyers send.
3. Technical implementation plan
Next.js on Vercel; Supabase for Postgres, session file storage and row-level auth; papaparse and SheetJS for ingestion; Recharts charts rendered server-side to PDF via Playwright on a serverless container; Anthropic or OpenAI called with a strict JSON schema whose allowed figure set is built from the variance table; Stripe Checkout plus webhooks for credits; Resend for receipts.
The build order is fixed by risk: ingestion first, math second, PDF third, gating last. The variance engine is pure functions with fixture tests, because a wrong number in a client-facing PDF is a trust failure no marketing recovers from. The commentary call receives the top movers as JSON and must return paragraph objects that cite account_line and figure; a post-validator rejects and retries once, then falls back to a template paragraph using only computed values.
The likeliest breakage is export-format drift — QBO and Xero column headers differ by plan and region. The mapping screen is therefore the first screen shipped, not the last.
4. Pricing and monetization
Credit model, no subscription required to start: Single Close $49 (1 export), Practice Starter $149 (5 exports, saved client profiles, materiality presets), Practice Monthly $99/month (15 exports per month). The $149 pack is the conversion target because it prices against 2-4 unbilled hours on a single client. Export is gated on a paid credit; the free sample the founder delivers is a PDF sent by email, not a free account.
Upsell path: practices that exhaust five credits inside one close cycle move to $99/month, which converts a one-time sale into recurring revenue the 30-day plan does not depend on.
5. First-customer acquisition
One channel: direct LinkedIn outreach to people whose headline contains "fractional CFO", "outsourced CFO" or "outsourced accounting", filtered by US/UK/CA/AU and by QuickBooks or Xero mentions, plus public practice directories. Target 200 names by day 7.
The message offers labour, not software: send two files, receive a finished brief free within 24 hours. Conversion happens in the same thread with a Stripe link and one question — how long does this take you now? Follow up on day 3 and day 7 to non-responders. Capacity is the real constraint: the founder can deliver roughly 5-8 sample briefs per day manually, so 200 messages is the ceiling, not 200 calls.
6. Execution roadmap
Days 1-5 build the MVP to a live, exported brief. Day 6 adds Stripe gating so no export is free. Day 7 prepares 200 names plus a demo P&L pair used for the sample that goes out before any payment ask. Days 8-14 are the first outreach wave, manual sample delivery, and message rewrite on measured reply rate. Days 15-21 add saved client profiles, the read-only link, and a hardened figure guard. Days 22-28 add a second title segment (bookkeeping/outsourced accounting), funnel instrumentation, dispute policy, referral asks and revenue reconciliation against TrustMRR. Days 29-30 ship one fix, refresh the demo, and push every warm thread to pay before the 13:00 UTC cutoff on December 16.
7. Revenue and profit scenarios (projections only)
Three scenarios, all projections built from stated hypotheses, not verified revenue. Conservative: 3 customers × $82 blended ARPU = $246 gross. Base: 13 customers × $107 = $1,391 gross, one $49 refund, ~5% Stripe fees ($70), zero ad spend, $45 operating. Upside: 26 customers × $110 = $2,860 gross, one $149 refund, $145 fees, $80 operating. Hackathon profit deducts refunds, chargebacks, processing fees and paid marketing only; hosting and model tokens sit in business profit, not the contest number. No figure here is a promise: the base case depends on roughly 200 messages producing ~45 delivered samples and 13 payments.
8. Risks and validation
Reply rate is the first gate: if the first 100 messages produce fewer than 8 replies by day 9, rewrite the message once; if 200 messages still produce under 5 file-sending replies, the channel is wrong and the plan stops. Commentary quality is the second gate: three delivered samples must be sendable after ≤10 minutes of editing, or the product is a variance calculator, not a brief. Format drift is the third: the first three real buyer files must ingest without founder code changes, otherwise mapping becomes a paid onboarding step the founder cannot afford. Trust is the fourth: financial exports demand a clear retention answer (deleted within 24 hours) stated on the upload screen. The fifth is funnel physics — manual sample delivery caps volume, so the plan assumes no more than 45 samples and never promises same-hour turnaround.
9. First 24 hours
Hours 1-2: create repo, Vercel project, Supabase project, domain, Stripe test account. Hours 3-4: page-scaffold upload of two files and a static variance table from a hardcoded fixture. Hours 5-6: writable variance engine with unit tests on the fixture pair. Hours 7-8: draft the 3-line outreach message, build the first 25 names, and write the upload-screen data-retention sentence. By the end of day one the founder has a deployed skeleton, a tested variance function, and 25 names ready.
Smallest sellable MVP
A paid-gated web app where a practitioner uploads two P&L exports, confirms column mapping, and exports one client-ready PDF variance brief with waterfall, top-10 movers and line-cited commentary.
Must-have features
- CSV and XLSX upload of two periods with a column-mapping screen
- Deterministic variance engine: absolute variance, percent variance, configurable materiality threshold
- Mover classifier tagging each line volume, timing, one-off, reclass or unmapped
- Top-10 movers table and waterfall chart rendered into a print-ready PDF
- Commentary generator that only cites account lines and figures present in the variance table, with a rejection guard
- Inline sentence editing plus regenerate-single-paragraph
- Shareable read-only HTML link
- Stripe Checkout credit purchase that gates export
Explicitly out of scope
- Live QuickBooks or Xero OAuth sync
- Multi-entity consolidation, XBRL or audit exports
- Budget-vs-actual, forecasting and cash-flow modules
- SSO, roles and approval workflows
- White-label client portal and scheduled email delivery
- Automatic export-format detection for unknown column headers
- Any human review or done-for-you service tier
Demo flow (what the buyer sees)
- Founder selects the demo client profile with saved column mapping
- Uploads prior-month and current-month P&L CSV exports
- App shows mapped columns and the computed variance table with materiality threshold applied
- Movers are auto-classified and ranked; waterfall renders
- Commentary paragraphs appear, each citing an account line and dollar figure
- User edits one sentence and regenerates another
- User buys a credit and exports the PDF plus read-only link
Technical implementation plan
Stack
Architecture
Browser uploads two files to a Supabase session bucket; a server route parses them with papaparse or SheetJS and applies the saved column mapping for that client profile. A pure TypeScript variance module computes absolute and percent variance, applies the materiality threshold and classifies movers. Those computed lines are the only input to the model, which returns paragraph objects citing account_line and figure; a validator drops any paragraph referencing a figure outside the allowed set. Charts and commentary render in a page that Playwright prints to PDF. Stripe webhooks write credits to the account row; the export route checks credits before rendering. Session files are deleted 24 hours after export by a scheduled job.
Build steps
- Create Next.js repo, Vercel project, Supabase schema for accounts, client profiles, sessions and credits
- Build the two-file upload screen with session storage and delete-after-24h job
- Build the column-mapping screen and persist mapping per client profile
- Write the variance engine as pure functions with fixture tests on a demo P&L pair
- Add the mover classifier and materiality threshold presets
- Render the variance table, waterfall and top-10 movers in the app
- Add the model call with the allowed-figure schema and the rejection/retry guard
- Add inline sentence editing and single-paragraph regeneration
- Add Playwright PDF export route and read-only share link
- Wire Stripe Checkout, webhook credit grant and pre-export credit check
- Add Resend receipt and export-ready email
- Add funnel events: upload started, mapping completed, export blocked, credit purchased
- Deploy to production domain and run one end-to-end export with a real-format file
AI components
- Commentary generation constrained to a JSON schema whose allowed figures come from the computed variance table
- Optional single-paragraph regeneration using the same constrained schema
- No AI is used for variance math, classification thresholds, ranking or PDF layout
Pricing and monetization
Single Close
Entry offerFor A practitioner testing the tool on one client at one month-end
- 1 client variance brief export
- PDF plus shareable read-only link
- Inline commentary editing
Practice Starter
For A solo or small practice converting after the free sample
- 5 client brief exports
- Saved client column mapping profiles
- Materiality threshold presets
- Waterfall, ranked movers and footnoted commentary
Practice Monthly
For A practice that exhausted credits and does this every close
- Up to 15 brief exports per month
- Saved client profiles
- Priority regeneration on model errors
Model: Prepaid export credits with one-time packs, plus a monthly plan for practices that close every month.
First revenue offer: Send a finished, free variance brief for one client within 24 hours — no signup, no card — then offer the $149 Practice Starter pack with a Stripe link in the same thread.
Expected ACV: $149
Upsell path: Practices that burn all five credits inside one close cycle move to the $99/month plan, and each additional client in their book is another credit-consuming profile.
First-customer acquisition
Direct LinkedIn outreach to fractional CFO and outsourced-accounting practitioners, preceded by a free manually delivered sample brief.
Where to find them
- LinkedIn search on headline keywords 'fractional CFO', 'outsourced CFO', 'outsourced accounting' combined with QuickBooks or Xero
- LinkedIn searches filtered by city clusters in US, UK, Canada and Australia
- Public accounting-practice directories and small-firm listings
- Accounting and bookkeeping communities where the founder confirms promotion is permitted
- Inbound replies from the first wave, asked for one referral intro
Exact outreach motions
- Message 1: offer to run two P&L files free and return a finished PDF within 24 hours
- Message 2 (day 3, non-responders): resend the same offer with a redacted sample brief attached
- Message 3 (day 7, non-responders): one line asking whether month-end packs are still manual this cycle
- Sample delivery email: attach PDF, ask the single question 'how long does this take you now?'
- Conversion reply: $149 Practice Starter Stripe link in the same thread, no new pitch
- Post-close reply: ask for one intro to another practitioner with the same close process
Message template
Hi {first_name} — I built a small tool that turns two month-end P&L exports (QBO or Xero) into a client-ready variance brief in about 5 minutes: waterfall, top-10 movers, and commentary that cites each account line. Send me one client's two files and I'll run it free and send the PDF back within 24 hours. No signup, no card. Worth a look?30-day execution roadmap
Week 1 · Build & validate · 7 days
Repo, hosting, database, domain
- Create Next.js repo, Vercel project and domain
- Create Supabase project with accounts, client profiles, sessions, credits tables
- Add Stripe test keys and a hardcoded demo P&L fixture
Deliverable: Deployed skeleton with schema and fixture data
Two-file upload and column mapping
- Build CSV/XLSX upload to session storage
- Build column-mapping screen with account line selection
- Persist mapping per client profile
Deliverable: Two files uploaded and mapped, mapping reloads on return
Deterministic math with tests
- Implement absolute and percent variance as pure functions
- Add materiality threshold and rank movers
- Write fixture unit tests on the demo P&L pair
Deliverable: Tested variance table rendering in-app
Waterfall and mover tagging
- Render waterfall and top-10 movers with Recharts
- Add volume, timing, one-off, reclass, unmapped classifier
- Add materiality preset selector
Deliverable: Full brief page with charts and ranked movers
Model guard plus export
- Call the model with a schema limited to computed figures
- Add rejection and single retry, fallback to template paragraph
- Add Playwright PDF export and inline sentence editing
Deliverable: End-to-end brief exported as PDF
Credit gating
- Wire Stripe Checkout with three prices
- Add webhook that grants export credits
- Block export until a credit exists, email receipt via Resend
Deliverable: Pay-to-export live in production mode
List and sample assets
- Build 200-name LinkedIn list in a tracking sheet
- Produce one redacted demo brief PDF
- Write the 3-line message and follow-up variants
Deliverable: 200 names, demo PDF, message scripts
Week 2 · Launch & first sale · 7 days
First 40 messages
- Send 40 LinkedIn messages with the free-sample offer
- Log send date and reply state per name
- Prepare 24-hour sample delivery checklist
Deliverable: 40 messages sent and tracked
Manual briefs, turnaround
- Deliver every received file pair as a finished brief
- Record turnaround time and edit minutes per brief
- Ask the 'how long does this take you now' question
Deliverable: 3-6 sample briefs delivered with timings logged
Volume and follow-ups
- Send 40 more messages
- Follow up day-3 non-responders with the sample PDF
- Post a redacted before-and-after where promotion is allowed
Deliverable: 80 total sent, follow-ups logged
Convert warm threads
- Send $149 Stripe links into every thread that received a sample
- Offer $49 Single Close to anyone who hesitates
- Log payment in the revenue sheet
Deliverable: First paid credit (target) and payment log started
Real buyer files
- Fix mapping failures from actual QBO and Xero exports
- Add any new header alias to the mapping presets
- Re-run the fixture tests plus one real file
Deliverable: Ingestion handles the first real export formats
Copy from objections
- Rewrite landing page around the objections heard in replies
- Add data-retention sentence and sample brief preview
- Instrument page view to checkout events
Deliverable: Revised landing page with funnel events live
Reply and conversion rates
- Compute reply rate, sample rate and sample-to-paid rate
- Write message v2 based on the best-performing reply
- Decide whether to widen or narrow the headline filter
Deliverable: Metrics sheet and message v2
Week 3 · Repeat & systematize · 7 days
Saved profiles and presets
- Ship saved client mapping profiles in the UI
- Ship materiality threshold presets per client
- Confirm profiles reduce second-close setup to under 60 seconds
Deliverable: Repeat-client setup under one minute
Message v2 at scale
- Send 40 messages using v2
- Segment names by practice size
- Track which segment replies
Deliverable: 160 total sent with segment tagging
Read-only output
- Ship read-only HTML brief link with expiry
- Deliver queued samples in batches
- Add link to the sample delivery email
Deliverable: Shareable link live and used in samples
Day-7 sequence
- Send day-7 follow-up to all non-responders
- Convert warm threads with $149 and $99 options
- Log cumulative paid credits
Deliverable: 3-5 cumulative paid credits (target)
Zero invented figures
- Add regression tests for the figure-rejection guard
- Log every rejected paragraph with reason
- Tighten prompt to name account lines explicitly
Deliverable: Guard test suite passing on 20 generated briefs
Verified allowed channels
- Check promotion rules in two accounting communities
- Post one redacted sample in each allowed venue
- Log referral traffic and replies per post
Deliverable: Two permitted posts with traffic logged
Pricing decision
- Review mix of $49 versus $149 versus $99 purchases
- Publish the $99 monthly price if repeat buyers exist
- Update outreach message with the strongest proof line
Deliverable: Pricing decision recorded and price live
Week 4 · Scale what works · 7 days
Second title segment
- Add bookkeeping and outsourced-accounting titles to the list
- Send 40 messages to the new segment
- A/B the opening line against the CFO segment
Deliverable: 200 total sent across two segments
Drop-off diagnosis
- Instrument mapping-abandonment events
- Identify where sample recipients stall before paying
- Fix the single highest drop-off step
Deliverable: Funnel report with one fix shipped
PDF layout
- Improve footnote layout and number alignment in the PDF
- Add client name and period header fields
- Re-render the three most recent briefs for comparison
Deliverable: PDF template v1.1 shipped
Every warm thread gets a link
- Send a Stripe link to every thread without one
- Offer the $99 monthly plan to anyone with 5+ clients
- Log who opened but did not pay
Deliverable: All warm threads have a paid offer open
Terms and refund policy
- Publish terms, refund policy and data-retention page
- Add a refund request path inside the receipt email
- State the 24-hour file deletion on the upload screen
Deliverable: Policy pages live and linked from checkout
Buyer-sourced intros
- Ask every buyer for one intro to a peer practice
- Send final outreach wave to remaining names
- Deliver all outstanding samples within 24 hours
Deliverable: 240+ sent, referral asks logged
Numbers match TrustMRR
- Reconcile Stripe payouts against the revenue sheet
- Confirm every sale is inside the contest window
- Recount refunds, fees and any marketing spend
Deliverable: Revenue reconciliation sheet balanced
Days 29-30 · Consolidate · 2 days
Highest-impact fix only
- Ship the single fix that unblocks the most stalled users
- Refresh the public demo brief with v1.1 layout
- Write a changelog entry for the 30 days
Deliverable: One fix shipped, changelog published
Cash before cutoff
- Push every open thread with a deadline-neutral nudge before Dec 16 13:00 UTC
- Confirm Stripe captures all pending payments inside the window
- Export the final revenue and profit summary
Deliverable: Final verified revenue summary and 30-day retro
Revenue and profit scenarios
How the numbers are defined
Hackathon profit — Collected revenue − refunds − chargebacks − payment processing fees − paid marketing spend.
Business profit — Hackathon profit − operating costs (hosting, domains, tooling, AI/API delivery costs).
The contest ranks entrants on hackathon profit only. Hosting, tooling and AI/API delivery costs are not deducted by the organisers, but each sale still has to cover them — so we show both. Definitions verified on 2026-11-15 against the official rules.
Stated assumptions
- Conservative: 3 paid conversions from roughly 60 messages and ~12 delivered samples; blended ARPU $82 because early buyers take the $49 Single Close.
- Base: 13 conversions from ~200 messages and ~45 delivered samples; ARPU ~$107 from 6 × $149, 4 × $49 and 3 × $99.
- Upside: 26 conversions; ARPU ~$110 from 14 × $149, 8 × $49 and 4 × $99.
- Payment fees assume Stripe at 2.9% + $0.30 plus roughly 1.5% currency and cross-border, about 5% blended on collected revenue.
- Refunds: 0 conservative, 1 × $49 base, 1 × $149 upside; chargebacks assumed 0 in all cases with a policy page shipped on day 26.
- ad_spend_usd is 0 in all three scenarios: the channel is founder LinkedIn outreach and manual sample delivery, with no ads, sponsored posts, influencers or affiliate payouts.
- Operating costs cover Vercel, Supabase, domain, Resend and model tokens at under $1 per brief plus fixed tooling (~$25-$80 per month).
- All three scenarios are projections built from the concept's stated hypotheses, not verified revenue, customers or traction.
Conservative
3 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $12
- Paid marketing
- $0
- Hackathon profit
- $234
- Operating costs
- $25
- Business profit
- $209
Base case
13 customers- Refunds
- $49
- Chargebacks
- $0
- Payment fees
- $70
- Paid marketing
- $0
- Hackathon profit
- $1,272
- Operating costs
- $45
- Business profit
- $1,227
Optimistic
26 customers- Refunds
- $149
- Chargebacks
- $0
- Payment fees
- $145
- Paid marketing
- $0
- Hackathon profit
- $2,566
- Operating costs
- $80
- Business profit
- $2,486
Model your own numbers
Runs in your browser · nothing is saved- Collected revenue
- $637
- Payment fees (3.9%)
- $25
- Paid marketing
- $0
- Hackathon profit
- $612
- Business profit
- $567
Hypothetical arithmetic, not a forecast. The fee rate is a flat 3.9% assumption, refunds and chargebacks are left at zero here, and none of this is tracked data.
All figures are scenario projections derived from stated hypotheses, not verified revenue, customers or traction.
Main risks and validation experiments
Risk 1Cold LinkedIn reply rate is far below the assumed 15 replies per 200 messages, so too few samples exist to convert.
Track replies daily; rewrite the message once at day 9 using the best-performing reply wording, and widen to bookkeeping and outsourced-accounting titles.
Send 40 messages on day 8 and count replies and file offers within 72 hours.
If 200 messages produce fewer than 5 file-sending replies by day 16, stop cold outreach and switch to warm intros inside verified accounting communities.
Risk 2Variance math is right but commentary reads generic, so buyers do not trust or forward the PDF.
Constrain every paragraph to a named account line and a figure from the variance table; edit three real samples and time the human fix-up.
Deliver three sample briefs on day 9 and record minutes of editing required to make each sendable.
If real samples need more than 20 minutes of editing each after two prompt iterations, reposition as a variance table tool at $49 and drop the brief promise.
Risk 3QBO and Xero export formats vary by plan and region, so the first real buyer file breaks ingestion.
Ship column mapping before charts; store header aliases per client profile and add a raw-header paste fallback.
Ingest five distinct real exports from different plans and log every failure mode.
If two of the first five real files cannot ingest without founder code changes, make mapping a paid assisted setup and price it explicitly.
Risk 4Practices refuse to upload client financials to an unknown tool.
State 24-hour file deletion on the upload screen, offer the founder-run sample path where the buyer keeps the account, and keep no OAuth tokens.
A/B the upload screen copy on the first 40 landing visits and count uploads started.
If fewer than 20% of sample recipients will upload their own files after the retention copy change, the product cannot be self-serve in 30 days.
Risk 5Manual sample delivery caps the funnel at a few briefs per day for one founder.
Cap promises at 24-hour turnaround, batch delivery into two windows per day, and never promise same-hour service.
Log minutes per delivered sample across days 9-17 and compute the daily ceiling.
If average delivery exceeds 30 minutes per sample, stop offering samples above 3 per day and shift budget to landing-page conversion.
Hour-by-hour: the first 24 hours
Hours 1-2
- Create Next.js repo and deploy an empty page to Vercel
- Create Supabase project and the accounts, client_profiles, sessions and credits tables
- Point a domain at the deployment
Output: Live skeleton with schema applied and domain resolving
Hours 3-4
- Build the two-file upload page with a demo P&L fixture pair
- Render a static variance table from the fixture
- Add the 24-hour file-deletion sentence to the upload screen
Output: Upload page showing a computed variance table from fixture data
Hours 5-6
- Write the variance engine as pure functions with unit tests
- Add materiality threshold and percent variance handling for zero and negative base
- Run tests against the fixture pair
Output: Passing test suite for absolute and percent variance
Hours 7-8
- Draft the 3-line outreach message and the day-3 and day-7 follow-ups
- Build the first 25 LinkedIn names into a tracking sheet
- Write the one-question sample-delivery email
Output: Message scripts, 25-name list, and delivery email ready to send
Full 30-day launch report
Read the full generated report
Delta Brief — 30-Day Launch Plan
JEV score: 61.9/100 (band C, confidence 0.67). Window: Nov 16 2026 13:00 UTC → Dec 16 2026 13:00 UTC. One founder, zero paid marketing, zero customers at start.
1. Product opportunity and positioning
Delta Brief sells to a named person on a named calendar date. A solo or 2-5 person fractional CFO practice runs 5-30 SMB clients on QuickBooks Online or Xero, and in the first ten days of every month it owes each client a month-end pack. Building that pack means re-exporting two P&Ls, rebuilding a variance table in Excel, and rewriting commentary that is 80% identical to last month's. It takes 2-4 hours per client and is the least billable work in the engagement.
The product is narrow on purpose: two files in, one PDF out. No warehouse, no OAuth sync in this launch, no team roles, no budget-vs-actual. That is the difference from the BI category it descends from and from generic chat tools: variances are computed deterministically in TypeScript before any model is called, and the model may only narrate figures that already exist in the variance table. A sentence carrying an unsupported number is rejected at validation, not caught by a human reviewer.
JEV scores speed to first revenue highest (7.58) — the buyer has the input files locally, a practice credit card, and a deadline. It scores operational simplicity lowest (4.46), so this plan keeps the product stateless: files held per session, raw financial exports deleted within 24 hours of export, no OAuth tokens, no human review queue, no 24/7 support promise, managed services only.
Why now: narration constrained to a computed table costs under $1 per brief in 2026, so a $49-$149 one-time price clears cost of delivery by a wide margin, and the buyer's alternative is 2-4 unbilled hours.
2. Smallest sellable MVP
Upload two P&L exports, confirm column mapping, see the variance table, get a one-page PDF with waterfall, top-10 movers and line-cited commentary, edit sentences inline, pay one credit to export. That is the whole product at day 5.
Excluded until after the 30 days: QBO/Xero OAuth, multi-entity consolidation, XBRL, SSO, approval workflows, forecasting, white-label portals, scheduled email delivery. Column mapping is manual on first use and remembered per client profile thereafter — no auto-detector is built before knowing which export formats real buyers send.
3. Technical implementation plan
Next.js on Vercel; Supabase for Postgres, session file storage and row-level auth; papaparse and SheetJS for ingestion; Recharts charts rendered server-side to PDF via Playwright on a serverless container; Anthropic or OpenAI called with a strict JSON schema whose allowed figure set is built from the variance table; Stripe Checkout plus webhooks for credits; Resend for receipts.
The build order is fixed by risk: ingestion first, math second, PDF third, gating last. The variance engine is pure functions with fixture tests, because a wrong number in a client-facing PDF is a trust failure no marketing recovers from. The commentary call receives the top movers as JSON and must return paragraph objects that cite account_line and figure; a post-validator rejects and retries once, then falls back to a template paragraph built only from computed values.
The likeliest breakage is export-format drift — QBO and Xero column headers differ by plan and region. The mapping screen therefore ships first, not last.
4. Pricing and monetization
Credit model, no subscription required to start: Single Close $49 (1 export), Practice Starter $149 (5 exports, saved profiles, materiality presets), Practice Monthly $99/month (15 exports). The $149 pack is the conversion target because it prices against 2-4 unbilled hours on one client. Export is gated on a paid credit; the free sample is a PDF sent by email, not a free account.
Upsell path: practices that exhaust five credits inside one close cycle move to $99/month, converting a one-time sale into recurring revenue the 30-day plan does not depend on.
5. First-customer acquisition
One channel: direct LinkedIn outreach to people whose headline contains "fractional CFO", "outsourced CFO" or "outsourced accounting", filtered by US/UK/CA/AU and by QuickBooks or Xero mentions, plus public practice directories. Target 200 names by day 7.
The message offers labour, not software: send two files, receive a finished brief free within 24 hours. Conversion happens in the same thread with a Stripe link and one question — how long does this take you now? Follow up on day 3 and day 7 to non-responders. Capacity is the real constraint: the founder can deliver roughly 5-8 sample briefs per day manually, so 200 messages is the ceiling, not 200 calls.
6. Execution roadmap
Days 1-5 build the MVP to a live, exported brief. Day 6 adds Stripe gating so no export is free. Day 7 prepares 200 names plus a demo P&L pair used for the sample that goes out before any payment ask. Days 8-14 are the first outreach wave, manual sample delivery, and a message rewrite based on measured reply rate. Days 15-21 add saved client profiles, the read-only link, and a hardened figure guard. Days 22-28 add a second title segment, funnel instrumentation, dispute policy, referral asks and revenue reconciliation against TrustMRR. Days 29-30 ship one fix, refresh the demo, and push every warm thread to pay before 13:00 UTC on December 16.
7. Revenue and profit scenarios (projections only)
Three scenarios, all projections built from stated hypotheses, not verified revenue. Conservative: 3 customers × $82 blended ARPU = $246 gross, no refunds, $12 in Stripe fees, $25 operating. Base: 13 customers × $107 = $1,391 gross, one $49 refund, $70 in fees, zero ad spend, $45 operating. Upside: 26 customers × $110 = $2,860 gross, one $149 refund, $145 in fees, $80 operating. Hackathon profit deducts refunds, chargebacks, processing fees and paid marketing only; hosting and model tokens sit in business profit, not the contest number. The base case depends on roughly 200 messages producing ~45 delivered samples and 13 payments — a hypothesis, not a forecast.
8. Risks and validation
Reply rate is the first gate: if the first 100 messages produce fewer than 8 replies by day 9, rewrite the message once; if 200 messages still produce under 5 file-sending replies, the channel is wrong and the plan stops. Commentary quality is the second gate: three delivered samples must be sendable after ≤10 minutes of editing, or the product becomes a variance calculator at $49. Format drift is the third: the first three real buyer files must ingest without founder code changes. Trust is the fourth: financial exports demand a clear retention answer, stated on the upload screen before anything else. The fifth is funnel physics — manual sample delivery caps volume, so the plan assumes no more than 45 samples and never promises same-hour turnaround.
9. First 24 hours
Hours 1-2: repo, Vercel project, Supabase schema, domain. Hours 3-4: upload page and static variance table from a fixture pair. Hours 5-6: tested variance engine with threshold handling. Hours 7-8: outreach message, 25 names, and the sample-delivery email. By the end of day one the founder has a deployed skeleton, a tested variance function, and a list ready to contact.
This brief was generated from public data about established startups, then narrowed and scored for a solo 30-day launch. Revenue and profit figures are scenario models, not verified results, guarantees or evidence of existing customers. Validate demand yourself before you spend a month on it — and check the contest rules for what counts toward your score.
Related ideas
All ideas →Bridge Brief
Turns a founder's own LinkedIn connections export into a ranked warm-intro map against a specific target list, with a ready-to-send ask for every path.
AnswerProof Audit Kit
A white-label audit generator that lets small SEO and content agencies produce a client-ready PDF showing how ChatGPT, Perplexity, and Gemini describe a client's brand versus its competitors, complete with a 30-day fix list.
Zéro Écart
A fixed-price catch-up service for French small businesses that are months behind on bookkeeping: upload a bank export plus a pile of receipts, and receive a finished, importable accounting journal, renamed receipts, and an itemised list of exactly what could not be matched.