Catalog Scout: business plan · 1.0.3
Buyer and offer
Buyer: US home-goods Shopify merchants who add supplier collections several times a year, and the freelancers and VAs who upload those products for them.
The job: a new batch of 20–50 products arrives. The vendor's line sheet has the sizes, materials and care. The product pages often don't, because someone has to find and type each fact.
The offer: a new-arrivals check.
- For each batch, Catalog Scout checks every new product for missing size, material, care and assembly, by product family.
- It drafts each missing fact word for word from the vendor's line sheet, citing file, row and column.
- Anything the sheet doesn't cover is marked "ask vendor".
- A person verifies the drafts before anything is published.
Price hypothesis:
- $49 per batch of up to 50 new products, about $1 per product. That sits at the low end of published per-SKU data-entry rates (see RESEARCH.md).
- An optional one-time baseline check of up to 250 existing products for $99.
- These are unvalidated hypotheses, not current revenue.
Why this and not the old offer: the previous candidate proposed a $49/month weekly completeness report. The commercial reviewer showed that a $1.99/month app already covers generic audits, and that re-checking an unchanged catalog weekly finds little. The new offer is priced per batch, is only needed when new products arrive, and does something the cheap tools don't: it matches the page to the vendor's own facts and cites them.
What you get that a chatbot or a $2 app doesn't
| Catalog Scout | $1.99/month audit app | Shopify Magic / a chatbot | |
|---|---|---|---|
| Home-goods facts (size, material, care, assembly) by family | Yes | Not listed | Writes copy; doesn't check |
| Reads the vendor's line sheet and matches by SKU | Yes | No | Only if you paste it, product by product |
| Every drafted fact cites file, row and column | Yes | No | No |
| Refuses to guess ("ask vendor") | Yes | n/a | Shopify says it "can occasionally invent details" |
| Knows what changed since the last batch (by handle + SKU) | Yes | No | No |
| Runs locally on the merchant's export; source is yours | Yes | App-hosted | Hosted |
The comparison is based on the listings checked on 2026-09-28 (RESEARCH.md). It claims no better accuracy than any tool; it lists capabilities.
Evidence and uncertainty
- There are no interviews, pilots, paying customers or sales. Merchant demand and kit demand are both UNKNOWN.
- Line sheets vary. Some vendors send PDFs, and some sheets lack care or assembly, so the "ask vendor" rate is itself a thing to measure.
- Detection accuracy on a builder-written blind set: 96.1% of advisories correct and 98% of gaps caught. This is probably optimistic (ACCURACY.md).
Fulfilment
- The client (or you) drops the batch export and line sheet into their
inbox/. Pulling via the merchant's own Shopify app is tested against a local mock only. intake.mjschecks, drafts, compares and writes the report. This is automated and tested.- You verify every draft against the cited row and label false alarms. This is a person's job: estimated 15–30 minutes per batch, not yet measured.
- You send the report and, if agreed, paste the verified drafts into the store. This is a person's job.
The AI operator can run steps 2–3 preparation and draft messages. It cannot invent facts, send without authorization, edit a store or spend money.
Acquisition
- Primary: Shopify freelancers and VAs who upload supplier collections. One freelancer can bring several stores, and they already pay for data entry.
- Secondary: merchants with visible new collections whose pages lack sizes or materials that the brand's wholesale page shows.
- Channel: a useful public sample (the demo plus a fictional line sheet) and personal, permission-based outreach. No paid marketing.
- Cost: acquisition is expensive in time. ECONOMICS.csv assumes 6–12 owner hours per new customer, and that decides whether this works.
Economics and workload
See ECONOMICS.md for the tables: churn, acquisition hours, review minutes, CAC, LTV:CAC and payback across Pilot fails / Low / Base / High.
- Losing scenarios: the Pilot-fails and Low scenarios lose money after valuing the owner's time.
- Base: breaks even after labor around month 7, under optimistic-but-plausible assumptions.
- Not proven: none of this is measured.
Risks and remedies
- No usable line sheet: the check still flags gaps, but the "draft" value disappears. Qualify prospects on whether their vendors send CSV or Excel sheets.
- Review takes longer than typing: measure it in pilots. If drafts don't save time, stop.
- False alarms and misses: run the rule-edit gate on every change and publish honest numbers.
- Cheap tools improve: keep the moat in cited, vendor-sourced facts and the home-goods rules, not in generic checks.
- Private data: use authorized files only, private folders and no public logs. The hosted demo processes everything in the browser.
- Branding: name, domain and trademark are not checked or transferred.
Milestones and kill tests
- Days 1–2: run the demo and dry run, and own-brand the kit.
- Days 3–14: qualify 10 freelancers and 10 merchants, and get 5 permissioned pilot batches with their line sheets.
- After 5 pilots, continue only if all three hold:
- at least 3 say the drafts saved time;
- at least 2 agree to pay $49 for their next batch;
- the measured review time is 30 minutes or less per batch.
Otherwise re-scope or stop. Never report gross revenue as profit, or kit downloads as demand.