Score C · 6.45/10Free to read$100K Hackathon · November 16, 2026 → December 16, 2026Nov 15, 2026

Discount Leak Ledger

Shopify store owners drop in two CSV exports and get a per-code Discount Leak Ledger — which discount codes leaked margin, which to kill or cap, and the exact settings to change — delivered as a PDF and fix-pack CSV within 10 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.

Speculative6.45/10
MVP build (solo)
4 days
Launch price
$199
Sales to $1,000 revenue
6 customers
First paying customer
Personalised X and LinkedIn DMs to DTC founders who are publicly promoting a Black Friday or December discount code, each carrying a hand-checked teaser finding on their own public codes.
Who pays

Owner-operator of a DTC Shopify store doing roughly $30K-$1M per year, running 15+ discount codes, in the US/UK/CA/AU, triggered by Black Friday-Cyber Monday or a major seasonal promotion, paying with a company card and needing no procurement or finance approval.

Biggest execution risk

Market demand 4.8/10

Narrow the buyer to a segment with an urgent, already-budgeted problem and reference the specific trigger event that creates willingness to pay.

Why this could work in 30 days: one founder can ship the MVP in 4 days; the plan targets a first sale on day 10; 6 sales at $199 clears $1,000 of collected revenue; strongest signal is speed to first revenue at 8.8/10.

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

13 concrete actions. Tick them off as you go — nothing is saved or sent.

0/13
Entry price
$199
Solo build
4 days
30d base revenue
$2,988
30d hackathon profit
$2,684
Section 1 of 8

Product opportunity and positioning

Target customer

Owner-operator of a DTC Shopify store doing roughly $30K-$1M per year, running 15+ discount codes, in the US/UK/CA/AU, triggered by Black Friday-Cyber Monday or a major seasonal promotion, paying with a company card and needing no procurement or finance approval.

Why now

The Black Friday-Cyber Monday trigger sits inside the 30-day window, so merchants already running 15+ codes have an urgent, dated reason to buy a per-code answer today rather than next quarter.

Problem

Discount codes are created ad hoc all year, scraped by coupon aggregators, stacked with other offers and applied to low-margin SKUs, but the P&L shows only one lump-sum discounts number, so the merchant cannot tell which code, segment or public listing caused the leak and repeats the same codes next season.

Solution

Merchant uploads a discount-code export and an order-level discount usage export, enters a COGS percentage, and the engine computes per-code redemptions, discount dollars, effective AOV versus store baseline AOV, new-vs-returning split and margin after discount, assigns KILL/CAP/KEEP with the exact setting to change, and emails a ledger PDF plus fix-pack CSV inside 10 minutes with a stored re-download page.

Differentiation · Shopify analytics shows one lump-sum discounts line; agency discount audits cost roughly $2,000 and take weeks; this is a same-day, per-code, margin-adjusted artifact that ends in a kill list instead of a dashboard to interpret.

1. Product opportunity and positioning

Discount Leak Ledger sells one thing: a per-code answer to the question every DTC operator asks after a sale, which is which of my discount codes actually cost me money. It is not a dashboard, not an app install, not a monitoring subscription. Two CSV exports go in. A PDF ledger and a fix-pack CSV come out inside 10 minutes.

The buyer is narrow on purpose. One person running a Shopify store between roughly $30K and $1M a year, maintaining 15 or more discount codes, in the US, UK, Canada or Australia, triggered by Black Friday-Cyber Monday or any major seasonal promotion. That person holds the company card, has no finance team to consult and no procurement step to clear. The purchase decision is a single afternoon.

The problem is structural, not a knowledge gap. Codes get created ad hoc by whoever runs the promo, get scraped by coupon aggregators within hours, get stacked with other offers, and get applied to the lowest-margin SKUs. At month end the merchant sees one line item called discounts. They cannot see that code SPRING20 was redeemed 412 times mostly by first-time buyers on a 22%-margin product, while code VIP15 was redeemed 38 times by repeat customers who would have paid full price. So they repeat the same codes next season, and the leak compounds.

The alternatives are weak in exactly the way that matters. Shopify analytics gives an aggregate. Agency discount audits are priced around the $2,000 mark with a multi-week turnaround, which is a non-starter for a one-person store in November. Spreadsheet reconstruction is possible and nobody does it, because pulling and joining two exports by hand is an hour of miserable work with a real chance of getting the join wrong.

The wedge is the deliverable, not the software. The output ends in a verdict, KILL, CAP or KEEP, mapped to one concrete settings change per code, exported as a CSV the merchant can work through in an afternoon. The PDF carries a one-page summary, a top-10 leak table, a season-long projection with a confidence band, and a re-run checklist for next season.

Because the output is a document, the first ten audits can be run by hand before automation exists. That removes the classic solo-founder trap of building for six weeks before hearing a no. The founder can also test the entire value proposition with public codes scraped from a store's own banners, which costs nothing and produces a personalised hook.

One limit is stated up front: the engine flags codes that look publicly exposed based on single-use share, low repeat rate and bursty redemption timing. It never claims proven coupon-site attribution. That framing is a product requirement, not legal boilerplate, because overstating attribution is the fastest route to a refund and a public complaint.

2. Smallest sellable MVP

The MVP is a paid upload form with a working ledger behind it. Concretely: a column mapper that accepts the Shopify discount-code export and the order-level discount usage export, with a plain two-column fallback template for merchants whose exports do not match; a deterministic engine that computes per-code redemptions, discount dollars, effective AOV against store baseline AOV, new-vs-returning split and margin after discount from a merchant-entered COGS percentage; verdict rules mapping KILL, CAP and KEEP to a specific setting change; a React-PDF ledger with summary page, top-10 leak table, season projection with confidence band and next-season checklist; a fix-pack CSV; and a Stripe checkout webhook that creates the job, emails both artifacts via Resend and leaves a job page for re-download.

Out of scope for 30 days: Shopify OAuth, automatic code disabling, scheduled re-runs, a coupon-site crawler, multi-currency or multi-store consolidation, team seats and any agency white-label portal. The merchant applies the kill list themselves.

3. Technical implementation plan

Next.js on Vercel serves the landing page, upload flow and job page. Supabase Postgres and storage hold jobs, uploaded files and generated artifacts. A Node serverless function runs the leak engine and verdict rules with no model involvement in any number. An LLM writes only the narrative explanation and anomaly callouts around numbers it is given. React-PDF renders the ledger and Resend delivers it. Stripe handles checkout and webhooks and is connected to TrustMRR before the first payment is accepted.

Build order: land the column contract on real exports, write the engine and unit-test it against two synthetic stores with known answers, add verdicts and the fix-pack CSV, render the PDF, wire the webhook-to-email pipeline, then run one live $199 purchase end to end and time it. Build time is four days, with day five reserved for the end-to-end paid test.

The hardest technical risk is not the engine, it is export format variance. Shopify export column names differ by plan and by what the merchant selects, and a paying buyer who uploads the wrong file has already paid. Mitigation is a column mapper with a preview step, a fallback template, and a same-day human fallback where the founder runs the parse manually.

4. Pricing and monetization

Two one-time tiers. Single Store Audit at $199 covers the ledger PDF, per-code verdicts, the fix-pack CSV and one free re-run after fixes are applied. Four-Store Pack at $699 covers four audits, one consolidated cross-store leak summary and a fix-pack CSV per store, aimed at e-commerce agencies and fractional CMOs. Blended expected order value is $249, assuming roughly 90% singles and 10% packs.

One-time pricing is deliberate. It matches a dated trigger event, it clears a company card without a subscription approval, and it is deliverable without a promise of 24/7 support. The upsell path is the free re-run converting into a second audit next season, and the pack buyer converting into a repeat agency customer for each client store.

5. First-customer acquisition

The only channel is personalised DMs to DTC founders who are publicly promoting a Black Friday or December code. The founder builds a list of 150 stores, records each store URL and three public codes, hand-checks those codes for stacking, missing usage caps or storewide application to already-discounted items, and sends the specific finding as the opening line. The offer is the $199 Single Store Audit with a redacted sample attached. No deck and no demo call required.

After Cyber Monday the same list gets a second message with the post-sale angle: you just ran 40 codes through Cyber Monday, here is what they actually cost you. Week-one goal is 150 rows in the sheet and the first 25 DMs sent with a real finding in each.

6. 30-day execution roadmap

Days 1-5 are build: column contract, engine, verdicts and CSV, PDF and delivery pipeline, then one live paid end-to-end test on day 5. Days 6-7 are proof and list: a conservative redacted sample ledger and a 150-row prospect sheet. Days 8-11 are the first outreach wave, roughly 25 to 30 DMs per day with hand-checked teasers. Day 10 is the first-sale target, with delivery done live on a screen share. Days 12-15 ride Black Friday and Cyber Monday, with concierge delivery and the post-sale message queued for day 15. Days 16-21 switch to the post-BFCM angle, re-contact non-responders once and open agency outreach for the Four-Store Pack. Days 22-30 are conversion, delivery quality, referral asks and a final push before the window closes on December 16 at 13:00 UTC.

7. Revenue and profit scenarios

All three scenarios are projections from stated assumptions, not verified revenue. Conservative: 3 audits at a $249 blended price, $747 collected, one refund line at zero, payment fees $22, no paid marketing, $25 of hosting and inference. Base: 12 audits, $2,988 collected, $199 of refunds, $105 of payment fees, no paid marketing, $70 of operating cost. Upside: 30 audits, $7,470 collected, $398 of refunds, $262 of payment fees, $120 of paid marketing and $160 of operating cost.

Payment fees assume Stripe at roughly 2.9% plus $0.30 per transaction, blended to about 3.5% on mixed $199 and $699 orders. The $100K hackathon target is not projected here: reaching it would need roughly 400 audits at $249 plus a working agency resale channel, which is a stretch path rather than the base case.

8. Main risks and validation experiments

Export friction is risk one: every extra upload step cuts conversion. Test it by watching three merchants pull their own exports on a screen share in week one.

Cold DM reply rate is risk two and the biggest swing factor in the model, assumed at 8-12% and entirely unvalidated. Test it with 25 DMs before building anything beyond the engine.

The public-exposure heuristic is risk three. If it overstates coupon-site attribution, it produces refunds and public criticism. Test it by showing the flag to three merchants and asking whether it reads as directional or as a claim.

Timing is risk four: urgency collapses after December 1 unless the post-sale message lands. Test it by sending the post-sale angle to ten non-responders on December 1 and measuring replies.

Export format variance is risk five. Test it by parsing the exports of three different stores before the first paid sale and recording every column-name mismatch.

9. First 24-hour action plan

Hours 1-2: pull real exports from a volunteer store, record exact column names, write the fallback template. Hours 3-4: register the domain, create the Stripe account, connect it to TrustMRR, verify a test payment lands. Hours 5-7: build the deterministic engine skeleton and check it against a hand-computed per-code table. Hours 8-10: build the column mapper, upload form and job record. Hours 11-12: publish a one-section landing page with the $199 price and a placeholder sample. Hours 13-16: start the 150-store prospect sheet, recording URL and three public codes per store.

Section 2 of 8

Smallest sellable MVP

A paid upload form that accepts two Shopify CSV exports, runs a deterministic per-code leak engine with a COGS input, and returns a ledger PDF plus a fix-pack CSV by email within 10 minutes.

Must-have features

  • Two-CSV intake with column mapper and a plain fallback template
  • Per-code engine: redemptions, discount dollars, effective AOV versus store baseline AOV
  • New-vs-returning customer split and margin after discount from a merchant-entered COGS percentage
  • Public-exposure heuristic flagged as directional, not proven attribution
  • KILL / CAP / KEEP verdict per code mapped to one concrete setting change
  • Fix-pack CSV export of verdicts and settings
  • Discount Leak Ledger PDF: one-page summary, top-10 leak table, season projection with confidence band, re-run checklist
  • Stripe checkout webhook creating a job, Resend email delivery, job page for re-download
  • Landing page with redacted sample ledger and two-tier price table

Explicitly out of scope

  • Shopify OAuth or any automatic store connection
  • Writing to the store or auto-disabling codes
  • Scheduled re-runs or a recurring monitoring dashboard
  • Coupon-site crawler or attribution database
  • Multi-currency, multi-store consolidation, ERP or margin-system integrations
  • Team seats, agency white-label portal, or client management

Demo flow (what the buyer sees)

  1. Merchant clicks Buy on the landing page and pays $199 via Stripe
  2. Stripe webhook creates a job record and shows the upload screen
  3. Merchant uploads the discount-code export and the order-level discount usage export
  4. Column mapper previews the parsed rows and the merchant confirms the mapping
  5. Merchant enters a COGS percentage and submits
  6. Engine computes per-code metrics and assigns KILL / CAP / KEEP
  7. LLM writes the narrative summary around the computed numbers
  8. PDF ledger and fix-pack CSV are generated, stored, and emailed via Resend
  9. Merchant opens the job page to re-download both artifacts
Build plan

Technical implementation plan

Stack

Next.js on Vercel for landing page, upload flow and job pageSupabase Postgres and storage for jobs, uploads and generated artifactsNode serverless function for the deterministic leak engine and verdict rulesReact-PDF for ledger renderingOpenAI or Anthropic API for narrative explanation and anomaly callouts onlyStripe for checkout and webhooks, connected to TrustMRR from day oneResend for transactional email deliveryCSV parsing library with an explicit column-mapping layer

Architecture

The landing page hands off to Stripe Checkout. A Stripe webhook creates a job row in Supabase and emails an upload link. The merchant uploads two CSVs to Supabase storage, confirms a column mapping, and submits. A Node serverless function parses both files, joins discount codes to order-level usage, computes per-code metrics and applies verdict rules. Numbers are produced only by that deterministic function. A single LLM call then writes the narrative around the already-computed figures. React-PDF renders the ledger, both artifacts are stored against the job, and Resend emails links. The job page reads from Supabase and allows re-download without a second payment.

Build steps

  1. Pull real Shopify discount-code and order-level discount usage exports; document exact column names
  2. Write the fallback two-column template for merchants whose exports do not match
  3. Build the deterministic engine: redemptions, discount dollars, effective AOV versus baseline, new-vs-returning split, margin after discount
  4. Unit-test the engine against two synthetic stores with hand-computed expected answers
  5. Add KILL / CAP / KEEP rules and map each verdict to one concrete setting change
  6. Generate the fix-pack CSV from the verdict set
  7. Add the public-exposure heuristic and label it directional in every output
  8. Render the ledger PDF: summary page, top-10 leak table, projection with confidence band, checklist
  9. Wire Stripe webhook to job creation and upload link
  10. Add the two-CSV upload flow with column mapper preview
  11. Add LLM narrative call that consumes computed numbers and cannot alter them
  12. Wire Resend delivery and the re-download job page
  13. Publish the landing page with redacted sample and price table
  14. Run one live $199 purchase end to end and time upload to inbox

AI components

  • LLM writes the narrative explanation and anomaly callouts around computed per-code numbers
  • LLM drafts the plain-language rationale text for each KILL / CAP / KEEP verdict from the deterministic rule output
  • No model touches, estimates or rounds any number in the ledger
Hardest technical risk: Shopify export column names and formats vary by plan and by which fields the merchant selects, so a paying buyer can upload a file the parser cannot read after the money has moved; mitigated by a mapper preview, a fallback template and same-day manual parsing.
Section 3 of 8

Pricing and monetization

Single Store Audit

Entry offer
$199 one-time

For One Shopify store owner running 15+ discount codes, typically right before or after a seasonal sale

  • Discount Leak Ledger PDF with per-code leak table and season projection
  • KILL / CAP / KEEP verdict for every code
  • Fix-pack CSV of exact settings to change
  • One free re-run after the fixes are applied

Four-Store Pack

$699 one-time

For E-commerce agencies and fractional CMOs managing several Shopify client stores

  • Four Single Store Audit ledgers
  • One consolidated cross-store leak summary
  • Fix-pack CSV for each store

Model: Two one-time charges: a $199 single-store audit and a $699 four-store pack, paid by card with no subscription.

First revenue offer: I checked three codes you are promoting right now and found a stacking issue plus a missing usage cap. Full per-code ledger on your two Shopify exports is $199, delivered inside 10 minutes, redacted sample attached.

Expected ACV: $249

Upsell path: The free re-run after fixes converts into a second audit next season, and a single-store buyer at an agency escalates to the $699 Four-Store Pack for their client book.

Target customers
48
Revenue model
$11,952
Costs
$480
Profit model
$11,472
Section 4 of 8

First-customer acquisition

Primary channel

Personalised X and LinkedIn DMs to DTC founders who are publicly promoting a Black Friday or December discount code, each carrying a hand-checked teaser finding on their own public codes.

Where to find them

  • Stores whose own site banner or popup advertises a live BFCM or December code
  • Founder posts on X announcing a discount code or promo thread
  • Founder posts on LinkedIn promoting a seasonal sale
  • Instagram brand accounts running a code in the bio or a story sticker
  • Shopify store owners active in DTC founder communities and Slack groups that allow member asks
  • Agency and fractional CMO operators publicly listing Shopify clients

Exact outreach motions

  • DM with the specific flaw found in their public codes and a screenshot, then the $199 offer
  • Email the store support address when no founder contact is public, same finding, same offer
  • Follow up once, 72 hours later, with the second finding rather than a reminder
  • On December 1 send the post-sale angle to every non-responder: you just ran 40 codes through Cyber Monday, here is what they cost
  • Ask each buyer for one referral to another store owner at delivery
  • Ask each buyer for permission to quote an anonymised leak number, no logos without written consent

Message template

Hey [name] - your code [CODE] is stacking with [OTHER] and has no usage cap, so it is applying to already-discounted items. I ran the numbers on three of your public codes. Full per-code ledger on your two Shopify exports is $199, back in 10 minutes as a PDF plus a settings CSV. Redacted sample here: [link]. Want me to run yours?
Week-one goal: 150 stores in the prospect sheet with URL plus three public codes each, and the first 25 DMs sent with a genuine finding in every message.
Section 5 of 8

30-day execution roadmap

MVP deadline: day 5First sale target: day 1030 days planned
Week 1 · Build & validate · 7 days
Day 1Data contract

Lock the input format before writing UI

  • Pull a real discount-code export and an order-level discount usage export; record exact column names.
  • Write the fallback two-column template for merchants whose exports do not match.
  • Register the domain, open Stripe, and connect the payment account to TrustMRR before any charge.

Deliverable: Documented column contract, fallback template, live Stripe and TrustMRR account.

Day 2Leak engine

Deterministic per-code maths

  • Build per-code redemptions, discount dollars, and effective AOV versus store baseline AOV.
  • Add new-vs-returning split and margin after discount from a COGS input.
  • Unit-test against two synthetic stores with hand-computed expected answers.

Deliverable: Engine module that returns a per-code ledger JSON.

Day 3Verdicts

Turn metrics into a kill list

  • Implement KILL / CAP / KEEP rules, each mapped to one concrete setting change.
  • Generate the fix-pack CSV with code, verdict, setting and suggested value.
  • Add the public-exposure heuristic and label it directional in the output.

Deliverable: Fix-pack CSV produced from any processed job.

Day 4Delivery

PDF and payment-to-email pipeline

  • Render the ledger PDF: summary page, top-10 leak table, projection with confidence band, checklist.
  • Wire Stripe webhook to job creation, PDF generation and Resend email delivery.
  • Build the job page with a re-download link so buyers never pay twice.

Deliverable: A paid job produces an emailed PDF and fix-pack CSV.

Day 5MVP liveMVP shipped

One real paid end-to-end run

  • Run a live $199 purchase on production keys using a real store export pair.
  • Time upload to inbox; it must land under 10 minutes.
  • Publish the landing page with a redacted sample ledger and the two-tier price table.

Deliverable: Working paid product end to end, timed and live.

Day 6Proof asset

Sample report and prospect list

  • Produce a conservative redacted sample ledger from a volunteer store.
  • Build a 150-store list of DTC stores promoting a BFCM or December code with three public codes each.
  • Write the teaser DM template and the finding checklist.

Deliverable: Sample ledger PDF and a 150-row prospect sheet.

Day 7First wave

25 hand-checked teasers

  • Hand-check three public codes per store for stacking, missing caps and storewide application.
  • Send 25 personalised DMs, each with a specific finding and the sample link.
  • Log every reply, objection and unsubscribed contact in the sheet.

Deliverable: 25 DMs sent, each with a real finding attached.

Week 2 · Launch & first sale · 7 days
Day 8Outreach volume

Scale the same message

  • Send 30 more DMs using the same angle and template.
  • Run free three-code teaser audits for every reply and return them the same day.
  • Capture the top three objections verbatim for the offer rewrite.

Deliverable: 55 total DMs and the first teaser audits returned.

Day 9Objection handling

Fix the offer, not the product

  • Rewrite the offer line using the top objections from days 7 and 8.
  • Send 30 more DMs with the revised opener.
  • Offer two repliers an on-screen-share walkthrough this week.

Deliverable: Revised offer live and 85 total DMs sent.

Day 10First saleFirst sale target

Close and deliver live

  • Close the first paid Single Store Audit at $199.
  • Deliver on a 20-minute screen share: export the two CSVs live and send the ledger the same session.
  • Ask for one referral and permission to quote an anonymised leak number.

Deliverable: First verified payment collected and first ledger delivered.

Day 11Wave three

Keep top of funnel moving

  • Send 30 more DMs with the revised offer.
  • Follow up once with day 7 and 8 non-responders using a second finding.
  • Publish an anonymised one-paragraph result from the day 10 delivery, no logos.

Deliverable: 120 total DMs and one anonymised public proof note.

Day 12Pre-BFCM

Urgency before the sale peaks

  • Send 30 DMs with the pre-sale angle: your codes go live Friday, here is what last season cost.
  • Publish a short public post on how to cap a Shopify code in three clicks.
  • Confirm delivery capacity for the next five days and write the manual parse fallback.

Deliverable: 150 total DMs and a written delivery fallback.

Day 13Black Friday

Concierge delivery

  • Deliver every queued audit in under 10 minutes and watch for parse failures.
  • Push the pre-sale angle to anyone who replied but did not buy.
  • Log every support question for the re-run checklist wording.

Deliverable: All queued audits delivered on time.

Day 14BFCM weekend

Same-day turnaround

  • Send 25 new DMs to stores visibly running stacked codes this weekend.
  • Deliver all audits within the 10-minute promise.
  • Collect one buyer quote with written permission.

Deliverable: Weekend deliveries complete and one consented quote.

Week 3 · Repeat & systematize · 7 days
Day 15Cyber Monday

Peak code volume

  • Monitor jobs and deliver every audit the same day.
  • Ask each buyer whether the kill list matched what they suspected.
  • Draft the post-sale message for December 1.

Deliverable: Cyber Monday deliveries done and post-sale message drafted.

Day 16Post-sale pivot

Re-engage the whole list

  • Send the post-sale angle to all 150 contacts: you just ran 40 codes through Cyber Monday, here is what they cost.
  • Send it first to the 100+ non-responders, then to warm repliers.
  • Update the landing page headline to the post-sale framing.

Deliverable: 150 post-sale messages sent and the page reframed.

Day 17Agencies

Four-Store Pack pipeline

  • Build a 30-strong list of agencies and fractional CMOs publicly listing Shopify clients.
  • Send each the pack offer with a consolidated cross-store summary described.
  • Offer one free single audit as a sample for a multi-client agency.

Deliverable: 30 agency contacts pitched and one sample offered.

Day 18Delivery quality

Refund prevention

  • Review every delivered ledger for a number that could read as implausible.
  • Tighten the projection language and confidence band in the PDF template.
  • Add the re-run checklist prompt to the delivery email.

Deliverable: Revised PDF template shipped to all future jobs.

Day 19Referrals

Warm introductions

  • Ask all buyers so far for one named referral with a one-line forwarded note.
  • Send the referral note on the buyer's behalf.
  • Track which buyer segment refers most.

Deliverable: Referral requests sent and one forwarded introduction.

Day 20Re-runs

Second purchase from existing buyers

  • Offer every buyer their free re-run after they apply fixes.
  • Record which codes moved from KILL to CAP in the re-run.
  • Turn the best re-run into an anonymised case note.

Deliverable: At least one re-run delivered and one case note drafted.

Day 21Week three review

Read the numbers honestly

  • Count DMs sent, replies, teasers run, offers made and audits paid.
  • Cut the worst-performing opener and keep the best.
  • Decide whether agency or single-store outreach gets the next nine days.

Deliverable: Written channel decision with the underlying counts.

Week 4 · Scale what works · 7 days
Day 22Scale winner

Double down on one channel only

  • Send 40 messages using only the winning opener and segment.
  • Publish the anonymised case note publicly.
  • Add a second worked example to the sample ledger.

Deliverable: 40 messages sent and a stronger sample asset.

Day 23Pack conversion

Raise order value

  • Pitch the Four-Store Pack to every agency that replied.
  • Offer existing single buyers an upgrade credit toward a pack.
  • Add the consolidated cross-store summary example to the page.

Deliverable: Pack pitches sent and the upgrade credit offered.

Day 24Automation patch

Remove manual steps

  • Fix the top two parse failures observed across all jobs.
  • Add a pre-upload validation message that names the expected columns.
  • Reduce the manual parse backlog to zero.

Deliverable: Fewer manual parses and a validating upload screen.

Day 25Outreach volume

Keep the funnel fed

  • Send 40 more messages to the remaining prospect sheet rows.
  • Re-contact every non-responder once with a new finding.
  • Track reply rate by day to find the best send window.

Deliverable: Full prospect sheet contacted at least once.

Day 26Season-two framing

Sell the January re-run

  • Add a January re-run offer to the delivery email for every buyer.
  • Test whether buyers will pre-pay a $99 re-run; take no money until the contest window allows it.
  • Write the January promo angle.

Deliverable: Pre-pay interest recorded and January angle written.

Day 27Trust assets

Reduce hesitation

  • Publish the delivery-time commitment and the confidence-band explanation on the page.
  • Add a clear statement that exposure flags are directional, never proven attribution.
  • Shorten the checkout to one screen.

Deliverable: Updated page with trust copy and single-screen checkout.

Day 28Referral push

Borrow other people's audiences

  • Ask buyers to forward the referral note to one named peer.
  • Offer a free re-run for any introduction that becomes a paid audit.
  • Follow up on all introductions sent in the previous ten days.

Deliverable: Referral loop running with a defined reward.

Days 29-30 · Consolidate · 2 days
Day 29Final outreach

Last full sending day

  • Send 50 messages to any untouched prospect rows.
  • Re-contact all warm but unpaid repliers with a hard deadline.
  • Confirm delivery capacity through the closing date.

Deliverable: Every prospect contacted and warm leads re-engaged.

Day 30Close and reconcile

Bank what was earned

  • Deliver every outstanding audit before the window closes.
  • Reconcile Stripe to TrustMRR and confirm each sale sits inside the contest window.
  • Write a one-page post-mortem on reply rate, conversion and delivery time.

Deliverable: All deliveries complete, revenue reconciled, post-mortem written.

Section 6 of 8

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

  • Base case assumes a $249 blended order value from roughly 90% $199 Single Store Audits and 10% $699 Four-Store Packs.
  • Payment fees assume Stripe at about 2.9% plus $0.30 per transaction, blended to roughly 3.5% on mixed order sizes.
  • Conservative assumes 3 audits from the first outreach wave with no refunds and no paid marketing.
  • Base assumes 12 audits across the window with one $199 refund.
  • Upside assumes 30 audits plus one agency pack chain, with two $199 refunds and $120 of paid marketing on a sponsored newsletter placement.
  • Cold DM reply rate of 8-12% and one paid audit per three teaser audits are hypotheses, not measurements.
  • Refunds are modelled at 0-5% of collected revenue, contingent on the ledger stating confidence levels and never claiming proven coupon-site attribution.
  • Operating cost covers Supabase, Vercel, domain, Resend and LLM inference to serve the modelled customers; the founder's time is not costed.
  • All figures are projections. No customers, revenue or partnerships exist at the time of writing.

Conservative

3 customers
$747
collected revenue · avg payment $249
Refunds
$0
Chargebacks
$0
Payment fees
$22
Paid marketing
$0
Hackathon profit
$725
Operating costs
$25
Business profit
$700

Base case

12 customers
$2,988
collected revenue · avg payment $249
Refunds
$199
Chargebacks
$0
Payment fees
$105
Paid marketing
$0
Hackathon profit
$2,684
Operating costs
$70
Business profit
$2,614

Optimistic

30 customers
$7,470
collected revenue · avg payment $249
Refunds
$398
Chargebacks
$0
Payment fees
$262
Paid marketing
$120
Hackathon profit
$6,690
Operating costs
$160
Business profit
$6,530

Model your own numbers

Runs in your browser · nothing is saved
Collected revenue
$2,388
Payment fees (3.9%)
$93
Paid marketing
$0
Hackathon profit
$2,295
Business profit
$2,225

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.

Scenario models from stated assumptions, not verified revenue. No customers, payments or traction exist; every figure is a projection.

Section 7 of 8

Main risks and validation experiments

Risk 1Merchants cannot pull the right two exports, so the upload flow stalls after payment.

Mitigation

Column mapper with preview, plain fallback template, and a same-day manual parse by the founder as backup.

Validation experiment

Watch three merchants pull their own exports on a 15-minute screen share and record every wrong click and missing field.

Kill criteria

If fewer than two of three merchants can produce a usable export pair in 15 minutes, replace CSV intake with a guided copy-paste form before scaling outreach.

Risk 2Cold DMs from an unknown solo founder are ignored or land in spam, so reply rate is the binding constraint, not product quality.

Mitigation

Lead every message with a specific verified flaw in the recipient's own public codes plus a screenshot, never a generic pitch.

Validation experiment

Send 25 hand-checked DMs in week one and measure reply rate, teaser acceptance rate and paid conversion separately.

Kill criteria

If reply rate is under 3% across 75 messages with the specific-finding opener, switch to agency and fractional-CMO outreach as the primary channel.

Risk 3The public-exposure heuristic overstates coupon-site attribution and triggers refunds or public criticism.

Mitigation

Label every exposure flag as directional in the PDF and the CSV, and never name a coupon site as the proven cause.

Validation experiment

Show the flagged section of a sample ledger to three merchants and ask whether it reads as a directional flag or as a factual claim.

Kill criteria

If two of three readers interpret it as a proven claim, remove the heuristic from the paid artifact and sell only the margin verdicts.

Risk 4Urgency collapses after December 1 and the post-sale message does not land, stranding the remaining pipeline.

Mitigation

Prepare the post-sale message in advance and send it on December 1 to the full list before interest decays.

Validation experiment

Send the post-sale angle to ten non-responders on December 1 and compare reply rate with the pre-sale wave.

Kill criteria

If the post-sale angle replies at under half the pre-sale rate, stop new outreach and focus the remaining days on re-runs and referrals from existing buyers.

Risk 5Shopify export column formats vary and break the parser for a buyer who has already paid.

Mitigation

Validate columns before taking payment where possible, and keep a manual parse path that can deliver within the 10-minute promise.

Validation experiment

Parse the exports of three different stores before the first paid sale and log every column-name mismatch.

Kill criteria

If more than one of three stores needs a code change to parse, add a required column-confirmation step to the upload flow before accepting further payments.

Risk 6One founder cannot run concierge delivery and cold outreach past roughly 15 sales.

Mitigation

Automate the webhook-to-PDF-to-email path by day 5 and reserve manual work only for failed parses.

Validation experiment

Time ten consecutive deliveries and track minutes per audit, including any manual intervention.

Kill criteria

If average delivery time exceeds 25 minutes of founder attention per audit at 15 sales, pause outreach and fix automation before taking more orders.

Section 8 of 8

Hour-by-hour: the first 24 hours

Hours 1-2

  • Pull a real discount-code export and an order-level discount usage export from a volunteer store.
  • Record every column name and note which fields are optional.

Output: A written column contract and a screenshot of both exports.

Hours 3-4

  • Register the domain and open the Stripe account.
  • Connect Stripe to TrustMRR and confirm the connection before any charge is possible.
  • Verify a test-mode payment lands and the webhook fires.

Output: Live payment account wired to TrustMRR, one test webhook received.

Hours 5-7

  • Build the deterministic engine: per-code redemptions, discount dollars, effective AOV versus baseline, new-vs-returning split, margin after discount.
  • Check the output against a hand-computed table for one synthetic store.

Output: Engine module with a passing hand-checked test.

Hours 8-10

  • Build the two-CSV upload form and the column mapper preview.
  • Create the job record schema in Supabase.

Output: An upload flow that stores a job and previews parsed rows.

Hours 11-12

  • Publish a one-section landing page with the $199 price and a placeholder sample slot.
  • Write the fallback two-column template.

Output: A live page and a downloadable fallback template.

Hours 13-16

  • Start the 150-store prospect sheet, recording store URL and three public codes per row.
  • Hand-check the first 20 stores for stacking, missing caps and storewide application.

Output: A prospect sheet with 150 rows started and 20 stores hand-checked.

Complete report

Full 30-day launch report

Read the full generated report

Discount Leak Ledger - 30-Day Launch Report

1. Product opportunity and positioning

Discount Leak Ledger sells a single per-code answer to the question every DTC operator asks after a major sale: which of my discount codes actually cost me money? Two Shopify CSV exports go in. A ledger PDF and a fix-pack CSV come out inside 10 minutes.

The buyer is deliberately narrow. One person running a Shopify store between roughly $30K and $1M a year, maintaining 15 or more discount codes, in the US, UK, Canada or Australia, triggered by Black Friday-Cyber Monday or a big seasonal promotion. That person holds the company card, has no finance team and no procurement step. The decision takes an afternoon.

The problem is structural. Codes get created ad hoc, scraped by coupon aggregators within hours, stacked with other offers, and applied to the lowest-margin SKUs. At month end the merchant sees one line item called discounts. They cannot see that one code ran 412 times mostly on a thin-margin product while another code gave 15% to repeat customers who would have paid full price. So the same codes come back next season and the leak compounds.

Alternatives are weak where it matters. Shopify analytics gives an aggregate. Agency discount audits sit near the $2,000 mark with a multi-week turnaround, which is a non-starter for a one-person store in November. Rebuilding the join in a spreadsheet is possible and nobody does it, because it is an hour of miserable work with a real chance of getting the join wrong.

The wedge is the deliverable, not the software. The output ends in KILL, CAP or KEEP mapped to one concrete settings change per code, exported as a CSV the merchant can work through in an afternoon. The PDF carries a summary page, a top-10 leak table, a season projection with a confidence band and a next-season checklist. Because the deliverable is a document, the first ten audits can be run by hand before automation exists. That is what makes the 30-day window survivable for one person.

One limit is stated up front: the engine flags codes that look publicly exposed based on single-use share, low repeat rate and bursty redemption timing. It never claims proven coupon-site attribution. That framing is a product requirement, because overstating attribution is the fastest route to a refund and a public complaint.

2. Smallest sellable MVP

The MVP is a paid upload form with a working ledger behind it: two-CSV intake with a column mapper and a fallback template, a deterministic engine computing per-code redemptions, discount dollars, effective AOV against store baseline, new-vs-returning split and margin after discount, verdict rules mapped to concrete setting changes, a React-PDF ledger, a fix-pack CSV, and a Stripe webhook that creates the job, emails both artifacts and leaves a re-download page. Out of scope: Shopify OAuth, automatic code disabling, scheduled re-runs, a coupon-site crawler, multi-store consolidation, seats and any agency portal.

3. Technical implementation plan

Next.js on Vercel serves the page, upload flow and job page. Supabase holds jobs, uploads and artifacts. A Node serverless function runs the engine and verdict rules with no model involvement in any number. An LLM writes only the narrative around numbers it is given. React-PDF renders the ledger, Resend delivers it, Stripe and TrustMRR are connected before the first charge.

Build order is column contract, engine with hand-checked tests, verdicts and CSV, PDF, then webhook-to-email, then one live $199 purchase timed end to end on day 5. The hardest technical risk is not the mathematics, it is export variance: column names differ by plan and by which fields the merchant selects, and a buyer who uploads the wrong file has already paid. Mitigation is a mapper preview, a fallback template and a same-day manual parse.

4. Pricing and monetization

Two one-time charges: the $199 Single Store Audit, covering the ledger, verdicts, fix-pack and one free re-run, and the $699 Four-Store Pack for agencies and fractional CMOs with a consolidated summary. Blended expected order value is $249, assuming roughly 90% singles and 10% packs. One-time pricing matches a dated trigger, clears a company card without subscription approval and avoids any 24/7 support promise. The upsell path is the free re-run becoming next season's audit, and single buyers at agencies escalating to a pack.

5. First-customer acquisition

One channel: personalised X and LinkedIn DMs to DTC founders publicly promoting a BFCM or December code. Build a 150-store list, record three public codes per store, hand-check them for stacking, missing caps or storewide application to already-discounted items, and open with that specific finding plus a screenshot and the redacted sample. The offer is the $199 audit, no demo required. After Cyber Monday the same list gets the post-sale message: you just ran 40 codes through Cyber Monday, here is what they cost you. Week-one goal is 150 prospect rows and 25 sent DMs, each with a real finding.

6. 30-day execution roadmap

Days 1-5 build and ship the paid product, ending with a live $199 purchase timed from upload to inbox. Days 6-7 produce a conservative redacted sample and the 150-row prospect sheet. Days 8-11 run the first outreach waves of 25-30 DMs per day with hand-checked teasers; day 10 is the first-sale target, delivered live on a screen share with a referral ask. Days 12-15 ride Black Friday and Cyber Monday with concierge delivery and pre-sale urgency, then queue the post-sale message for December 1. Days 16-21 send that message to the entire list, open agency outreach for the Four-Store Pack, tighten the PDF for refund prevention and convert free re-runs. Days 22-28 double down on the winning opener, fix the top parse failures and run a referral loop. Days 29-30 push the last outreach wave, deliver every outstanding audit and reconcile Stripe against TrustMRR before the window closes.

7. Revenue and profit scenarios

All figures are projections, not verified revenue. Conservative: 3 audits at a $249 blended price, $747 collected, no refunds, $22 of payment fees, no paid marketing, $25 of operating cost. Base: 12 audits, $2,988 collected, $199 of refunds, $105 of payment fees, no paid marketing, $70 of operating cost. Upside: 30 audits, $7,470 collected, $398 of refunds, $262 of payment fees, $120 of paid marketing and $160 of operating cost.

Fees assume Stripe at roughly 2.9% plus $0.30, blended to about 3.5% across mixed order sizes. Operating cost covers Supabase, Vercel, domain, Resend and LLM inference; the founder's time is not costed. The $100K hackathon target is not projected here: it would need roughly 400 audits at $249 plus a working agency resale channel, which is a stretch path rather than the base case.

8. Main risks and validation experiments

Export friction comes first: every extra upload step cuts conversion. Watch three merchants pull their own exports on a screen share in week one and record each wrong click. Cold DM reply rate is the biggest swing factor, assumed at 8-12% and entirely unvalidated; send 25 hand-checked DMs and measure reply, teaser acceptance and paid conversion separately. The public-exposure heuristic is the reputational risk; show the flagged section to three merchants and ask whether it reads as a directional flag or a factual claim. Timing is the fourth risk, since urgency collapses after December 1; test the post-sale message on ten non-responders on December 1 and compare reply rates. Fifth is export variance; parse three different stores before the first paid sale and log every column mismatch.

9. First 24-hour action plan

Hours 1-2 pull real exports from a volunteer store and write the column contract. Hours 3-4 register the domain, open Stripe, connect it to TrustMRR and verify a test webhook. Hours 5-7 build the deterministic engine and check it against a hand-computed table. Hours 8-10 build the upload form, column mapper and job schema. Hours 11-12 publish the one-section landing page at $199 with the fallback template. Hours 13-16 start the 150-store prospect sheet and hand-check the first 20 stores. No payment is accepted until the day 5 end-to-end test passes.

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 →