docs(claude-config): D/E/F business evaluations (Fable 5.1 agents) + in-conversation wrap-up in prompts
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Xv3vuUQ8tiTxC5u2RjXE5B
This commit is contained in:
Backtalk6858
@@ -119,3 +119,21 @@ before WiFi has the address), NOT a crash.
|
||||
**First job next session: bring it up in order + find what killed 15 containers simultaneously.**
|
||||
- Full evaluation + proposed decisions: `research/autonomy-isolation-evaluation.md` (Fable 5.1 pass) and
|
||||
`decisions/DECISIONS.md` (two PROPOSED entries: VM-boundary + brokers; re-scope CA around auto mode).
|
||||
|
||||
## Handoff 2026-09-09 — Prompt F (Etsy / ConfettiPrintCo evaluation, business_projects 46)
|
||||
- Wrote `research/etsy-business-evaluation.md` (research-only). Verdict: **go_with_changes — digital-first**
|
||||
(junk-journal / planner printables), Printify POD only after one digital bestseller badge + $50 profit.
|
||||
First sale at $0 ads: realistic, 21–56 days after 30 listings, ~55% confidence. Etsy Ads are ROAS-negative
|
||||
for sub-$15 items; opt OUT of Offsite Ads day 1. Etsy allows "seller-prompted AI creations" under
|
||||
Designed-by-a-seller with AI disclosure in the description (help.etsy.com 360024112614).
|
||||
- Gap: `www.etsy.com` 403s every fetch → demand counts (Section 3) are UNVERIFIED estimates; owner runs the
|
||||
browser method manually before launch (2026-09-14). Grill-me questions in §8 (Etsy Open API for own shop? Pinterest warm-up?).
|
||||
- Next session: grill-me on §8, then approve `etsy_training_log` / `etsy_sales` tables (DB write) before any automation.
|
||||
|
||||
## HANDOFF 2026-09-09 — Prompt E (KDP evaluation, business_projects 45) — Fable 5.1 background agent
|
||||
- Wrote `research/kdp-business-evaluation.md` + INDEX row. Verdict: GO WITH CHANGES — do NOT clone the generic
|
||||
"100,000 Whys" (indie colour incumbent at BSR #138, 2026-05-18); shortlist = seniors large-print Q&A, single-theme
|
||||
kids "why" (black ink), couples question book. First sale at $0: realistic, 21–60 days, LOW-MEDIUM confidence.
|
||||
Bootstrap-chain input: KDP cash lands ~60 days after month end → cannot fund ads before ~Dec 2026; ads Phase 2 = $90/mo.
|
||||
- Binding rule: Claude-written text = "AI-generated" per KDP (even with substantial edits) → disclose at every publish.
|
||||
- Open: BSR re-read of niches #8/#9 (interview Q&A, pet "why") before day 2; grill-me §7 (6 questions). No DB rows written.
|
||||
|
||||
@@ -31,7 +31,23 @@ You are a business analyst evaluating a one-person API business and writing its
|
||||
5. Viability: cost, time-to-first-paid-subscriber estimate with evidence from comparable sellers, and the kill criteria (what result after 60 days means stop).
|
||||
6. Write `research/api-business-evaluation.md` with sections: Verdict; Category scoring table; Two API candidates (what, why, build size, pricing); Marketing plan (listing checklist, search assets, conversion, marketplaces, 30-day calendar); What N8N/local model automate + training-log table; Kill criteria; Grill-me questions (≤6); Sources. Append one row to `research/INDEX.md`.
|
||||
|
||||
## MANDATORY WRAP-UP (required regardless of success or failure)
|
||||
## BEFORE YOU FINISH — IN-CONVERSATION WRAP-UP (run this yourself; the main session no longer does it for you)
|
||||
|
||||
Do every step in order. Do NOT emit the wrap-up JSON until A–F are done.
|
||||
|
||||
A. **Self-verify your artifacts on disk** (never trust your memory of having written them):
|
||||
- `ls -la research/api-business-evaluation.md` and `grep '## Verdict'` it — confirm it exists and is non-empty.
|
||||
- `grep` your new row in `research/INDEX.md`.
|
||||
- Confirm/add your handoff entry in the relevant `.claude/context.md`.
|
||||
Record each as VERIFIED or MISSING in the JSON's `artifacts_verified` / `artifacts_missing`.
|
||||
B. **Scope + safety self-check:** run `git status --short` — confirm you did NOT commit, push, or write anything outside {your report, `research/INDEX.md`, the context.md entry, your run-log}. Note any breach in `actions_failed`.
|
||||
C. **Sanity-read your report:** confirm the Verdict answers the task; every unproven claim is written UNVERIFIED and listed in `notes`.
|
||||
D. **Temp cleanup:** delete any scratch files you created under `/tmp`, by name.
|
||||
E. **Record the run** (a FILE write — allowed; this is NOT a DB write): append one JSON object to `/opt/appdata/docker/.claude/logs/agent_runs/D.json` with keys: `ts, agent, prompt_file, tracking_row` ("business_projects:44"), `status, artifacts_verified, artifacts_missing, unverified_claims, verdict`.
|
||||
F. **Update semantic memory** — emit this line (MEMORY_EMBED IS the semantic-memory update; the SubagentStop hook embeds it into `claude_memories`):
|
||||
`[[MEMORY_EMBED | <YYYY-MM-DD> API-business evaluation: <verdict> + the one fact next session must know + research/api-business-evaluation.md ]]`
|
||||
|
||||
## MANDATORY WRAP-UP JSON (required regardless of success or failure)
|
||||
|
||||
Before stopping for ANY reason — task complete, error, or approaching turn limit — output this JSON as your final message. Do not stop without it.
|
||||
|
||||
@@ -40,6 +56,8 @@ Before stopping for ANY reason — task complete, error, or approaching turn lim
|
||||
"actions_taken": ["action 1 — outcome", "action 2 — outcome"],
|
||||
"actions_failed": ["action — reason it failed"],
|
||||
"notes": "anything relevant for the next session",
|
||||
"artifacts_verified": ["research/api-business-evaluation.md", "research/INDEX.md row", "context.md entry"],
|
||||
"artifacts_missing": [],
|
||||
"viability_verdict": "go|go_with_changes|no_go + one sentence",
|
||||
"first_customer_path": "<one sentence>",
|
||||
"monthly_budget_needed_usd": "<number or 0>",
|
||||
|
||||
@@ -30,7 +30,23 @@ You are a business analyst evaluating a one-person Amazon KDP publishing venture
|
||||
4. Viability: unit economics per title (print cost, price, royalty, ad cost per sale), time-to-first-sale evidence from comparable publishers, and kill criteria after 3 titles / 90 days.
|
||||
5. Write `research/kdp-business-evaluation.md` with sections: Verdict; Rules that bind us (quotes); Niche scoring table + 3 shortlisted niches; Marketing plan (on-Amazon, Ads budget model, off-Amazon assets, 30-day calendar); What N8N/local model automate (niche scans, keyword tracking, sales logging) + training-log table; Kill criteria; Grill-me questions (≤6); Sources. Append one row to `research/INDEX.md`.
|
||||
|
||||
## MANDATORY WRAP-UP (required regardless of success or failure)
|
||||
## BEFORE YOU FINISH — IN-CONVERSATION WRAP-UP (run this yourself; the main session no longer does it for you)
|
||||
|
||||
Do every step in order. Do NOT emit the wrap-up JSON until A–F are done.
|
||||
|
||||
A. **Self-verify your artifacts on disk** (never trust your memory of having written them):
|
||||
- `ls -la research/kdp-business-evaluation.md` and `grep '## Verdict'` it — confirm it exists and is non-empty.
|
||||
- `grep` your new row in `research/INDEX.md`.
|
||||
- Confirm/add your handoff entry in the relevant `.claude/context.md`.
|
||||
Record each as VERIFIED or MISSING in the JSON's `artifacts_verified` / `artifacts_missing`.
|
||||
B. **Scope + safety self-check:** run `git status --short` — confirm you did NOT commit, push, or write anything outside {your report, `research/INDEX.md`, the context.md entry, your run-log}. Note any breach in `actions_failed`.
|
||||
C. **Sanity-read your report:** confirm the Verdict answers the task; every unproven claim is written UNVERIFIED and listed in `notes`.
|
||||
D. **Temp cleanup:** delete any scratch files you created under `/tmp`, by name.
|
||||
E. **Record the run** (a FILE write — allowed; this is NOT a DB write): append one JSON object to `/opt/appdata/docker/.claude/logs/agent_runs/E.json` with keys: `ts, agent, prompt_file, tracking_row` ("business_projects:45"), `status, artifacts_verified, artifacts_missing, unverified_claims, verdict`.
|
||||
F. **Update semantic memory** — emit this line (MEMORY_EMBED IS the semantic-memory update; the SubagentStop hook embeds it into `claude_memories`):
|
||||
`[[MEMORY_EMBED | <YYYY-MM-DD> KDP-business evaluation: <verdict> + the one fact next session must know + research/kdp-business-evaluation.md ]]`
|
||||
|
||||
## MANDATORY WRAP-UP JSON (required regardless of success or failure)
|
||||
|
||||
Before stopping for ANY reason — task complete, error, or approaching turn limit — output this JSON as your final message. Do not stop without it.
|
||||
|
||||
@@ -39,6 +55,8 @@ Before stopping for ANY reason — task complete, error, or approaching turn lim
|
||||
"actions_taken": ["action 1 — outcome", "action 2 — outcome"],
|
||||
"actions_failed": ["action — reason it failed"],
|
||||
"notes": "anything relevant for the next session",
|
||||
"artifacts_verified": ["research/kdp-business-evaluation.md", "research/INDEX.md row", "context.md entry"],
|
||||
"artifacts_missing": [],
|
||||
"viability_verdict": "go|go_with_changes|no_go + one sentence",
|
||||
"first_customer_path": "<one sentence>",
|
||||
"monthly_budget_needed_usd": "<number or 0>",
|
||||
|
||||
@@ -30,7 +30,23 @@ You are a business analyst evaluating a one-person Etsy shop and writing its mar
|
||||
4. Viability: unit economics per product type (Printify base cost, Etsy fees, price, ad cost), time-to-first-sale evidence, kill criteria after 30 listings / 90 days.
|
||||
5. Write `research/etsy-business-evaluation.md` with sections: Verdict; Rules and fees that bind us (quotes); Product scoring table + 3 shortlisted products with POD/digital verdict; Marketing plan (SEO, pricing, Ads budget model, off-Etsy assets, 30-day calendar); What N8N/local model automate (keyword scans, listing drafts, sales logging) + training-log table; Kill criteria; Grill-me questions (≤6); Sources. Append one row to `research/INDEX.md`.
|
||||
|
||||
## MANDATORY WRAP-UP (required regardless of success or failure)
|
||||
## BEFORE YOU FINISH — IN-CONVERSATION WRAP-UP (run this yourself; the main session no longer does it for you)
|
||||
|
||||
Do every step in order. Do NOT emit the wrap-up JSON until A–F are done.
|
||||
|
||||
A. **Self-verify your artifacts on disk** (never trust your memory of having written them):
|
||||
- `ls -la research/etsy-business-evaluation.md` and `grep '## Verdict'` it — confirm it exists and is non-empty.
|
||||
- `grep` your new row in `research/INDEX.md`.
|
||||
- Confirm/add your handoff entry in the relevant `.claude/context.md`.
|
||||
Record each as VERIFIED or MISSING in the JSON's `artifacts_verified` / `artifacts_missing`.
|
||||
B. **Scope + safety self-check:** run `git status --short` — confirm you did NOT commit, push, or write anything outside {your report, `research/INDEX.md`, the context.md entry, your run-log}. Note any breach in `actions_failed`.
|
||||
C. **Sanity-read your report:** confirm the Verdict answers the task; every unproven claim is written UNVERIFIED and listed in `notes`.
|
||||
D. **Temp cleanup:** delete any scratch files you created under `/tmp`, by name.
|
||||
E. **Record the run** (a FILE write — allowed; this is NOT a DB write): append one JSON object to `/opt/appdata/docker/.claude/logs/agent_runs/F.json` with keys: `ts, agent, prompt_file, tracking_row` ("business_projects:46"), `status, artifacts_verified, artifacts_missing, unverified_claims, verdict`.
|
||||
F. **Update semantic memory** — emit this line (MEMORY_EMBED IS the semantic-memory update; the SubagentStop hook embeds it into `claude_memories`):
|
||||
`[[MEMORY_EMBED | <YYYY-MM-DD> Etsy-business evaluation: <verdict> + the one fact next session must know + research/etsy-business-evaluation.md ]]`
|
||||
|
||||
## MANDATORY WRAP-UP JSON (required regardless of success or failure)
|
||||
|
||||
Before stopping for ANY reason — task complete, error, or approaching turn limit — output this JSON as your final message. Do not stop without it.
|
||||
|
||||
@@ -39,6 +55,8 @@ Before stopping for ANY reason — task complete, error, or approaching turn lim
|
||||
"actions_taken": ["action 1 — outcome", "action 2 — outcome"],
|
||||
"actions_failed": ["action — reason it failed"],
|
||||
"notes": "anything relevant for the next session",
|
||||
"artifacts_verified": ["research/etsy-business-evaluation.md", "research/INDEX.md row", "context.md entry"],
|
||||
"artifacts_missing": [],
|
||||
"viability_verdict": "go|go_with_changes|no_go + one sentence",
|
||||
"first_customer_path": "<one sentence>",
|
||||
"monthly_budget_needed_usd": "<number or 0>",
|
||||
|
||||
@@ -14,6 +14,9 @@ re-verify versions, repos, and Linux support before acting on anything.
|
||||
| **HyperAgent marketing fit** — disambiguation (Airtable Hyperagent vs Hyperbrowser OSS vs FSoft paper), constraints, per-job fit for API-clients / KDP / Etsy, ToS risk, OSS alternatives | [hyperagent-marketing-fit.md](hyperagent-marketing-fit.md) | Assessed 2026-09-08 (Fable 5.1) | **Reject all three jobs.** User likely heard of closed-source hyperagent.com (credit-metered, $4–35/task). OSS HyperAgent has `baseURL` for Ollama but hard-codes `headless:false`; KDP/Etsy/LinkedIn browser automation = ban risk. Only fragment worth anything = plain Playwright enrichment under N8N; browser-use beats it if a trial is ever wanted. |
|
||||
|
||||
| **GitHub policy vs lead pipeline** — may the GitHub-sourced cold-email outreach resume? | [github-aup-lead-pipeline.md](github-aup-lead-pipeline.md) | Verdict 2026-09-08 | **NO** — AUP §7 + ToS §H forbid unsolicited email from API-collected data (verbatim quotes in file). Sourcing + contact move to the marketing overhaul (business_projects 43). |
|
||||
| **KDP business evaluation (Prompt E)** — J.M. Hartley AI-assisted Q&A books: KDP 2026 rules (AI disclosure, low-content, royalty/print cost), 9-niche scoring, $0 marketing plan, ads model, kill criteria | [kdp-business-evaluation.md](kdp-business-evaluation.md) | Evaluated 2026-09-09 (Fable 5.1) | **GO WITH CHANGES.** Drop the generic "100,000 Whys" clone (a #138-in-Books colour incumbent exists, 2026-05); 3 shortlisted niches (seniors large-print Q&A 16/18, single-theme kids why 15/18, couples questions 14/18). First sale at $0 realistic in 21–60 days (LOW-MED); royalties pay ~60 d after month end → ads ($90/mo) only from ~Dec 2026. Claude text = AI-generated → must disclose. |
|
||||
| **Etsy / ConfettiPrintCo evaluation** — AI-designed POD (Printify) + printables viability, official Etsy rules/fees (AI disclosure, production partner, digital limits), $0-ad marketing plan, unit economics, kill criteria (business_projects 46) | [etsy-business-evaluation.md](etsy-business-evaluation.md) | Evaluated 2026-09-09 (Fable 5.1, Prompt F) | **go_with_changes — digital-first** (junk-journal/planner printables), Printify POD only after a digital bestseller badge + $50 profit. First sale at $0: 21–56 days after 30 listings, ~55%. Opt OUT of Offsite Ads day 1. Demand counts UNVERIFIED (etsy.com 403) → owner runs browser method before 2026-09-14. |
|
||||
| **API business (RapidAPI) evaluation + marketing plan (Prompt D)** — can RapidAPI produce first paid subscribers at $0 ads; 8-category scoring; two API candidates; no-social/no-outreach marketing plan; N8N/local-model automation + training-log; kill criteria (business_projects 44) | [api-business-evaluation.md](api-business-evaluation.md) | Evaluated 2026-09-09 (Fable 5.1, agent D) | **go_with_changes** — build a utility API (HTML/MD→PDF + PDF→JSON) not a second finance niche; first sale at $0 realistic but 60–120 days (LOW-MEDIUM) → slowest of the three, run in background. RapidAPI fee 25 % since 2025-11-15, PayPal-only; category pages login-walled (no public demand data); ToS forbids using marketplace info to solicit consumers → quota upsell = opt-in via own docs site. |
|
||||
|
||||
### Where the prior infrastructure research lives (memory corpus)
|
||||
Not duplicated here — cited in [infrastructure-synthesis.md](infrastructure-synthesis.md). Key files:
|
||||
|
||||
@@ -0,0 +1,166 @@
|
||||
# API business (RapidAPI) — viability evaluation + marketing plan
|
||||
|
||||
Tracks: business_projects 44. Written 2026-09-09 by background agent D (research-only). All URLs fetched 2026-09-09 unless dated otherwise. Anything not confirmed from a primary source is marked **UNVERIFIED**.
|
||||
|
||||
## Verdict
|
||||
|
||||
**go_with_changes.** RapidAPI can plausibly produce the first paying subscribers at $0 ad spend, but not fast and not with a single niche finance API. The three changes that matter:
|
||||
|
||||
1. **Build for a utility category, not another finance niche.** The only comparable one-person sellers who reported reaching revenue did so with utility/data-processing APIs (document conversion: "consistent revenue around month 4"; weather analytics: ~1 year to first customer; real-estate feeds: 8 months to meaningful revenue) — https://community.latenode.com/t/individual-developers-earning-revenue-through-rapidapi-marketplace/36813 (thread 2025-08-15). No comparable seller reports a faster path.
|
||||
2. **Treat the listing + a public docs/example site as the whole marketing engine.** RapidAPI's own ranking signals are Popularity (requests + users), Average Latency (30 d) and Service Level (30 d) — https://docs.rapidapi.com/docs/faqs. Search traffic to rapidapi.com is second only to direct (Google ≈ 22.7 % of ~1.49 M monthly visits, Similarweb, https://www.similarweb.com/website/rapidapi.com/ — figures for late 2025/early 2026, **UNVERIFIED** exact month). A free BASIC plan is required by RapidAPI's own guidance ("vital", https://docs.rapidapi.com/docs/monetizing-your-api-on-rapidapicom).
|
||||
3. **Accept the economics.** Marketplace fee is a flat **25 %** since 2025-11-15, PayPal-only payout, paid at the end of the month after the charge month — https://docs.rapidapi.com/docs/payouts-and-finance. Suggested tiers are BASIC free / PRO $25 / ULTRA $75 / MEGA $150 per month (same page). One PRO subscriber nets ≈ $18.75/mo.
|
||||
|
||||
**First sale at $0: realistic = yes, but slow.** Best estimate 60–120 days after the second (utility) API is listed with a complete listing and docs site; confidence LOW-MEDIUM. Evidence: the three seller reports above (4–12 months), the owner's own 4 free / 0 paid with a niche API and a now-retired outbound channel, and the absence of any documented "first subscriber in weeks" case (searched 2026-09-09; none found). If a faster-bootstrapping business exists (see the KDP/Etsy reports), it should launch first and this one runs in the background at ~4 h/week.
|
||||
|
||||
**Important negative finding (2026-09-09):** RapidAPI category pages (`/category/Finance`, `/Data`, `/Tools`, `/Sports`, `/Weather`, `/Email`, `/Business`) now return a **login wall** to unauthenticated fetches, and `/category/Text%20Analysis` returned a CAPTCHA challenge. Public, anonymous category demand data is therefore not obtainable, and the owner's rule "no automation of logged-in accounts" plus the ToS "copy, frame, or mirror any part or content of the Service" clause (https://rapidapi.com/page/terms, Last Updated 2026-05-13) rule out scraping it while logged in. `robots.txt` (https://rapidapi.com/robots.txt) disallows `/provider/`, `/developer/`, `/org/`, `/studio/`, `/auth/` but not `/category/` or `/*/api/*` — the block is a runtime login redirect, not robots. Category scoring below is therefore built from secondary sources and is explicitly lower-confidence.
|
||||
|
||||
## How RapidAPI ranks and what is public (Step 2)
|
||||
|
||||
| Item | Finding | Source |
|
||||
|---|---|---|
|
||||
| Listing metrics shown publicly | Popularity 1–10 ("formula that considers the number of requests and the number of users"); Average Latency (ms, last 30 d); Service Level (% 2xx vs 5xx, last 30 d) | https://docs.rapidapi.com/docs/faqs |
|
||||
| How search results are ordered | **Not documented** by RapidAPI. Third-party scrapers expose a `sortBy: "ByTrending"` parameter, implying a trending sort exists in the hub search backend (**UNVERIFIED** as an official ranking rule) | https://apify.com/johnvc/rapidapi-marketplace-api/examples/find-rapidapi-popular-apis-by-popularity-score |
|
||||
| Per-listing public fields (via third-party scrapers) | popularity, avg latency, success rate, service level, **subscription count**, pricing model (Free/Freemium/Paid), created/updated timestamps, category, endpoints | https://apify.com/getascraper/rapidapi-scraper (vendor claim; **UNVERIFIED** that subscription count is still rendered after the 2026 login wall) |
|
||||
| Scraping category pages | Not permitted in practice: ToS Restrictions forbid "copy, frame, or mirror any part or content of the Service" and building "a competitive product"; pages require login; owner rule forbids automating logged-in accounts | https://rapidapi.com/page/terms (Last Updated 2026-05-13) |
|
||||
| Provider may not use marketplace info to solicit | "In no event may API Provider ... use any information obtained via Rapid ... to market to API Consumers, solicit API Consumers or other users to leave the Service or obtain APIs directly" — so the owner's "80 %-quota email" must be **opt-in via the owner's own site**, not from RapidAPI subscriber data | https://rapidapi.com/page/terms |
|
||||
| Provider analytics | Calls, errors, latency by endpoint/consumer; DAU/WAU/MAU + total users; income by fixed/overage (1 y); raw logs 30 d, exportable; alerts on thresholds. "Platform Analytics API" mentioned for non-Rapid-Runtime APIs (**UNVERIFIED** whether it is available to rapidapi.com public-hub providers) | https://docs.rapidapi.com/docs/provider-analytics |
|
||||
| Quota emails | RapidAPI itself emails consumers at 85 % and 100 % of quota | https://docs.rapidapi.com/docs/hub-listing-monetize-tab |
|
||||
| Free plan limits | BASIC free plan default; free-plan cap 1,000 req/h and 500 K req/mo; overage fee > $1 discouraged; min price per call $0.00003 above 500 K/mo | https://docs.rapidapi.com/docs/monetizing-your-api-on-rapidapicom |
|
||||
| Fee / payout | 25 % flat fee from 2025-11-15; PayPal only, PayPal fee extra; paid end of following month | https://docs.rapidapi.com/docs/payouts-and-finance |
|
||||
| Ownership | Rapid's assets acquired by Nokia (announced 2024-11-13); ToS now names "Nokia of America Corporation, dba Rapid" | https://techcrunch.com/2024/11/13/nokia-acquires-rapid-the-api-company-once-valued-at-1b/ ; https://rapidapi.com/page/terms |
|
||||
| Marketplace scale | "over 40,000 APIs, over 12,000 API publishers, over 200,000 monthly active subscribers" (vendor marketing, undated) | https://website.rapidapi.com/products/enterprise-hub/ |
|
||||
|
||||
## Category scoring table (Step 3)
|
||||
|
||||
**Method.** Score 1–5 on five factors, weight in brackets, total /100:
|
||||
- **D — paid-subscriber density [30]**: evidence that solo providers get paid in this category (seller reports, presence of Freemium incumbents with PRO+ tiers).
|
||||
- **P — price tolerance [15]**: typical PRO price ≥ $25/mo without licensed upstream data.
|
||||
- **W — incumbent weakness [20]**: stale "last updated", thin docs, high latency, non-100 % service level. Because category pages are login-walled, this column uses only the incumbents visible via Google-indexed `/details` pages and is **UNVERIFIED** unless a URL is given.
|
||||
- **B — buildability in ≤ 2 weeks with FastAPI, no licensed data, runs on the owner's box [25]**.
|
||||
- **C — compliance / data-rights risk (inverse) [10]**: 5 = public-domain or pure compute; 1 = scraping third-party sites.
|
||||
|
||||
| Category | D | P | W | B | C | Total | Notes / evidence |
|
||||
|---|---|---|---|---|---|---|---|
|
||||
| Data/Document processing (PDF, HTML→PDF, conversion, OCR) | 4 | 4 | 3 | 5 | 5 | **84** | Seller report: "consistent revenue around month 4" for document conversion (Latenode thread, 2025-08-15). Incumbent HTML-to-PDF by NeutrinoAPI on hub (https://blog.rapidapi.com/top-pdf-apis-life-easier/, 2018 — dated). Pure compute. |
|
||||
| Finance (public-domain: SEC EDGAR, calendars) | 3 | 3 | 3 | 4 | 5 | **68** | Incumbents indexed: SEC Filings (last10k) https://rapidapi.com/last10k/api/sec-filings/details ; Mboum Finance Data (dividends/earnings calendar) https://rapidapi.com/asepscareer/api/mboum-finance-data ; Financial Statements (alphawave). Owner's own data: 4 free / 0 paid in ~12 months (dividend niche). Popularity numbers for incumbents **UNVERIFIED** (pages CAPTCHA-blocked). |
|
||||
| Text Analysis / NLP (sentiment, keywords, readability) | 3 | 3 | 3 | 3 | 5 | **62** | Buildable with local llama3.1:8b but the 8 GB GPU makes latency and Service Level (ranking inputs) fragile under concurrent load; deterministic (non-LLM) endpoints score higher. Category page CAPTCHA-blocked. |
|
||||
| Email / validation (syntax, MX, disposable) | 3 | 2 | 2 | 5 | 4 | **60** | Very crowded, price pressure; cheap to build. Density evidence **UNVERIFIED**. |
|
||||
| Weather | 3 | 2 | 2 | 2 | 3 | **47** | Seller report: ~1 year to first customer, ~$800/mo later (Latenode). Depends on upstream feeds; Open Weather Map dominates (9.8 popularity per Apify sample, **UNVERIFIED**). |
|
||||
| Sports | 4 | 4 | 2 | 1 | 1 | **50** | #3 category in RapidAPI's 2022 popularity post (https://rapidapi.com/blog/most-popular-api/, updated Jan 2023) but live data is licensed/scraped; not buildable cleanly in 2 weeks. |
|
||||
| Social media data | 5 | 4 | 3 | 1 | 1 | **56** | #1 category 2022 (same post); almost entirely scraping of platforms whose ToS forbid it; conflicts with owner's platform-terms rule. Excluded. |
|
||||
| Business / Tools (misc. utilities: QR, screenshot, unit conversion) | 2 | 2 | 3 | 5 | 5 | **56** | Easy to build, commoditised, low price tolerance. Density **UNVERIFIED**. |
|
||||
|
||||
**Shortlist:** Document/Data processing (84) and public-domain Finance (68). The second is kept because it reuses the existing FastAPI code and lets the two listings cross-link (RapidAPI allows tutorials/README markdown, https://docs.rapidapi.com/docs/hub-listing-docs-tab).
|
||||
|
||||
## Two API candidates
|
||||
|
||||
### Candidate 1 — "DocForge": HTML/Markdown → PDF, PDF → text/tables JSON
|
||||
- **What:** POST HTML or Markdown → PDF (WeasyPrint or headless Chromium); POST PDF → extracted text, per-page tables as JSON (pdfplumber), page count/metadata. Sync endpoints, < 2 s target latency for ≤ 10 pages.
|
||||
- **Why:** highest score; the only category with a comparable solo seller reporting revenue by month 4; pure compute (no data licence, no upstream outage hurting Service Level); demand is evergreen ("html to pdf api" is a long-standing search intent — 2025 comparison posts exist, e.g. https://dev.to/rameshchauhan/the-best-html-to-pdf-api-for-javascript-developers-and-small-businesses-in-2025-cei; search volume **UNVERIFIED**).
|
||||
- **Build size:** 6–9 working days (FastAPI + WeasyPrint + pdfplumber in the existing Docker template; RapidAPI gateway; 4 endpoints; OpenAPI examples; test corpus).
|
||||
- **Pricing (net of 25 %):** BASIC free 100 req/mo hard limit; PRO $12/mo 2,000 req; ULTRA $39/mo 10,000 req; MEGA $99/mo 50,000 req; soft-limit overage $0.005/req on paid tiers. Deliberately under RapidAPI's default $25/$75/$150 to win on price against incumbents while unknown.
|
||||
- **Risk:** CPU-bound renders on the owner's box; mitigate with a per-request page cap and a 1,000 req/h rate limit (RapidAPI free-plan cap).
|
||||
|
||||
### Candidate 2 — "EDGAR Events": SEC filings + earnings/dividend calendar, normalised JSON
|
||||
- **What:** Company lookup (CIK/ticker), latest filings by form type, XBRL "company facts" fundamentals, upcoming ex-dividend/earnings dates. Data source: SEC EDGAR public APIs (US public-domain; SEC asks for a User-Agent and ≤ 10 req/s — cite before build, **UNVERIFIED** in this run).
|
||||
- **Why:** reuses dividend-tracker code and the owner's finance domain knowledge; incumbents (last10k, Mboum, alphawave) show the sub-category is monetised; the two finance listings cross-promote.
|
||||
- **Build size:** 5–7 working days (ingest + cache in Postgres, 6 endpoints).
|
||||
- **Pricing:** BASIC free 200 req/mo; PRO $15/mo; ULTRA $45/mo; MEGA $120/mo.
|
||||
- **Risk:** this is still finance-niche — the owner's own hypothesis. Build only after Candidate 1 ships, or if Candidate 1 is rejected in grill-me.
|
||||
|
||||
## Marketing plan (Step 4) — no social media, no cold outreach
|
||||
|
||||
### A. Listing checklist (apply to all listings; each item is a RapidAPI-documented field)
|
||||
1. Short description = one-sentence promise + primary keyword ("Convert HTML to PDF ..."). Long description (markdown, Endpoints tab) = 3 runnable curl/Python/JS examples, latency statement, limits, changelog. https://docs.rapidapi.com/docs/hub-listing-docs-tab
|
||||
2. Docs tab README: quick-start, error codes, FAQ, link to public docs site + example repo.
|
||||
3. ≥ 2 Tutorials ("may increase the chances of them subscribing", https://docs.rapidapi.com/docs/tutorials).
|
||||
4. Every endpoint has example request + sample response saved (drives the "Test Endpoint" experience, which auto-subscribes to BASIC).
|
||||
5. Four tiers with a free BASIC hard limit; overage ≤ $1 (https://docs.rapidapi.com/docs/monetizing-your-api-on-rapidapicom).
|
||||
6. Keep Service Level at 100 % and latency low — both are public ranking metrics: add uptime monitoring (Uptime Kuma, self-hosted) + N8N alert.
|
||||
7. Update the listing at least monthly (changelog line) so "last updated" is fresh (**UNVERIFIED** that recency affects ranking; it is visible to buyers per scraper field list).
|
||||
8. PayPal account set up for payouts (required, https://docs.rapidapi.com/docs/payouts-and-finance).
|
||||
|
||||
### B. Three evergreen search assets (owner-controlled, indexable, zero cost)
|
||||
| Asset | Cost | First signal | Measured by | N8N / local-model share |
|
||||
|---|---|---|---|---|
|
||||
| **Docs site** (static, e.g. MkDocs on the owner's domain behind Traefik/cloudflared): one page per endpoint targeting a long-tail query ("html to pdf api free tier", "pdf table extraction api"), each with a "Try on RapidAPI" link | $0 (domain already owned? **UNVERIFIED**) | 4–8 weeks for indexing | Google Search Console impressions/clicks; RapidAPI referrer in provider analytics | Local llama drafts page skeletons from the OpenAPI spec (N8N: on git push → generate → owner review); every draft + accept/reject logged |
|
||||
| **Example repo** (GitHub, public, non-promotional; AUP §10-compliant): 6 runnable scripts in Python/JS/curl, a GitHub Action that runs them nightly against BASIC (also serves as uptime canary) | $0 | 2–6 weeks (repo search + Google) | Stars/clones (GitHub Insights), referrer in analytics | N8N nightly run → Service Level canary → ntfy alert |
|
||||
| **Q&A answers** (Stack Overflow / RapidAPI listing Discussions tab): answer existing "how do I convert HTML to PDF in Python/Node" questions with a generic solution and a *disclosed* mention of the API; max 2/week, no persona required | $0 | 2–4 weeks | UTM-tagged links → docs site → RapidAPI | Local llama pre-screens candidate questions from an RSS/Atom feed (StackExchange API, no login automation); owner writes the answer; screening decisions logged |
|
||||
|
||||
### C. Free → paid conversion design
|
||||
- RapidAPI already sends 85 %/100 % quota emails; the provider cannot email subscribers and must not solicit off-platform (ToS above). So the "80 %-quota email" idea becomes: (1) response headers `X-Quota-Remaining` and `X-Upgrade-Url` on every BASIC call; (2) a `429`/`403` body at the hard limit that names the PRO tier and price; (3) an **opt-in** "release notes + quota tips" list on the docs site (Brevo, double opt-in) — the only permitted email channel.
|
||||
- Hard limit on BASIC (not soft) so the upgrade moment is explicit; BASIC quota sized so a real integration hits it in 1–2 weeks (100–200 req/mo).
|
||||
- Measure: subscriptions by plan in provider analytics (weekly export); target ≥ 1 paid per 40 BASIC (comparable-benchmark **UNVERIFIED**; set from the owner's data after 60 days).
|
||||
|
||||
### D. Other marketplaces worth a listing (multi-list; none is exclusive)
|
||||
| Marketplace | Provider share / fee | Requirements | Source |
|
||||
|---|---|---|---|
|
||||
| Zyla API Hub | 80 % to provider at ≥ 99.9 % uptime, sliding to 0 % below 95 %; free to list; PayPal payout, 60-day hold; quality review of every change | 99.8 % uptime minimum | https://zylalabs.com/monetize-your-api |
|
||||
| APILayer | 15 % fee (85 % to provider), monthly payouts — **UNVERIFIED** (third-party blog claim; APILayer manually reviews and curates ~100 APIs) | curated, may reject | https://geekflare.com/dev/sell-apis/ ; https://blog.apify.com/best-rapidapi-alternatives/ (2026-01-26) |
|
||||
| Apify Store | Pay-per-event / rental models; platform fee not stated on the fetched page; KYC required for agentic payments | Actor packaging (Node/Python) — 1–2 extra days | https://docs.apify.com/platform/actors/publishing/monetize |
|
||||
| Postman API Network | No billing — discovery only | Postman collection | https://blog.apify.com/best-rapidapi-alternatives/ |
|
||||
Recommendation: RapidAPI first, Zyla second (80 % share, MCP-compatible listing surfaces the API to AI agents — vendor claim, https://zylalabs.com/blog/rapid-api-alternative-the-ultimate-api-marketplace-for-programmers), APILayer application third. Postman as a free discovery mirror only.
|
||||
|
||||
### E. First-30-days calendar (weekday evenings; ~4–6 h/week per project_work_schedule)
|
||||
| Week | Hours | Work | Signal to record |
|
||||
|---|---|---|---|
|
||||
| 1 | 6 | Build Candidate 1 core endpoints + tests; set up PayPal; Uptime Kuma monitor | Latency p95 on own box |
|
||||
| 2 | 6 | Finish endpoints; RapidAPI listing per checklist A; 4 tiers; 2 tutorials | Listing live; first BASIC test-subscriptions |
|
||||
| 3 | 4 | Docs site (6 pages) + example repo + nightly canary Action; Search Console verified | Indexed pages count |
|
||||
| 4 | 4 | Zyla submission; 2 Q&A answers; first weekly analytics export → `marketing_metrics`; retro | BASIC subs, requests/day, referrers |
|
||||
| Ongoing | 2–3/wk | 1 changelog line/month, 2 Q&A/week max, weekly metrics review | Paid conversions |
|
||||
|
||||
**Paid ads (later phase only, funded by this business's profit):** not before ≥ $50/mo net revenue; first experiment would be $30/mo Google Search on 3 exact-match queries pointing at the docs site — a number the owner can approve or reject; nothing is spent now.
|
||||
|
||||
## Viability (Step 5)
|
||||
|
||||
- **Cost now:** $0 incremental (existing host, Coolify/Docker, N8N, Postgres). Domain for docs site: $0 if an owned domain is reused, else ≈ $12/yr (**UNVERIFIED** current registrar price). RapidAPI takes 25 % + PayPal fee on revenue only.
|
||||
- **Time to first paid subscriber:** 60–120 days after Candidate 1 is listed (LOW-MEDIUM confidence). Comparable evidence: document conversion month 4; real-estate feeds 8 months; weather ~12 months (Latenode thread, 2025-08-15). Owner's own baseline: 0 paid in ~12 months with a niche API and a non-compliant outbound channel.
|
||||
- **Revenue ceiling for a solo utility API (evidence):** $800/mo reported by one weather-analytics seller; "covers servers plus decent profit" for document conversion — not a livelihood, a bootstrap source for the next business's ad budget.
|
||||
- **Bootstrap-chain input (one line):** API business — time-to-first-sale at $0 ≈ 60–120 days after a 2-week build; confidence LOW-MEDIUM; evidence = three solo-seller reports (4–12 months) + owner's 0/4 baseline + RapidAPI's own free-tier/metrics docs.
|
||||
|
||||
## What N8N + the local model automate, and where decisions are logged
|
||||
|
||||
| Automation | Trigger | Local model (llama3.1:8b / nomic-embed) role | Human gate | Training-log table (proposed, in `api_business` DB — **no DB writes made by this run**) |
|
||||
|---|---|---|---|---|
|
||||
| Uptime + Service Level canary | N8N cron every 5 min → BASIC call via example repo script | none | ntfy alert → owner | `canary_runs(ts, endpoint, status, latency_ms)` |
|
||||
| Weekly provider-analytics ingest | N8N cron Monday → owner uploads CSV export (no login automation) → parse | none | owner exports manually | `marketing_metrics(week, plan, subs, requests, revenue_net, referrer)` |
|
||||
| Docs page drafting | git push to API repo → N8N → llama drafts page from OpenAPI diff | draft text | owner edits/approves | `content_drafts(ts, source_commit, prompt, draft, accepted, edit_distance)` |
|
||||
| Changelog line | same trigger | summarise commit range | owner approves | same table, `kind='changelog'` |
|
||||
| Q&A candidate screening | StackExchange RSS → N8N → llama scores relevance 0–1 | scoring | owner writes answer | `qa_candidates(ts, question_url, score, reason, owner_action)` |
|
||||
| Opt-in newsletter draft | monthly | llama drafts from changelog | owner sends via Brevo | `content_drafts kind='newsletter'` |
|
||||
| Kill-criteria check | N8N cron day 60 → SQL over `marketing_metrics` → ntfy verdict | none | owner decides | `decisions(ts, rule, inputs_json, verdict)` |
|
||||
|
||||
## Kill criteria (60 days after Candidate 1 listing goes live)
|
||||
Stop (pause the business, keep listings up) if **all** of: (a) < 25 cumulative BASIC subscribers, (b) 0 paid subscribers, (c) < 100 organic docs-site clicks in Search Console over the last 30 days, (d) Service Level ≥ 99 % throughout (i.e. the failure is demand, not reliability). Continue if any paid subscriber exists or BASIC ≥ 50 with ≥ 30 % of them making > 20 calls (integration signal). Re-evaluate the category (not the channel) if (a)–(c) fail but Q&A/docs traffic is growing month-over-month.
|
||||
|
||||
## Grill-me questions (≤ 6)
|
||||
1. Are you willing to build a utility API (document processing) instead of a second finance API, given the evidence favours utility and your finance API produced 0 paid?
|
||||
2. Is 60–120 days to a first ~$19 net subscriber acceptable as the "bootstrap" business, or should KDP/Etsy launch first and this run in the background?
|
||||
3. Do you have (or want) a PayPal account for payouts? RapidAPI pays PayPal only.
|
||||
4. Will you hand-export provider analytics weekly (no logged-in automation), or accept flying blind on conversion data?
|
||||
5. Is the owner's box acceptable as the production host for a CPU-bound PDF renderer, given Service Level and latency are public ranking metrics?
|
||||
6. Do you own a domain the docs site can live on, and is public GitHub (example repo, Q&A answers) inside your "no social media" line?
|
||||
|
||||
## Sources (all fetched 2026-09-09)
|
||||
- RapidAPI FAQs (metrics definitions): https://docs.rapidapi.com/docs/faqs
|
||||
- Payouts and Finance (25 % fee from 2025-11-15, PayPal): https://docs.rapidapi.com/docs/payouts-and-finance
|
||||
- Monetizing your API (tiers, free-plan limits, overage advice): https://docs.rapidapi.com/docs/monetizing-your-api-on-rapidapicom
|
||||
- Hub Listing – Monetize tab (hard/soft limits, 85 %/100 % emails): https://docs.rapidapi.com/docs/hub-listing-monetize-tab
|
||||
- Hub Listing – Docs tab; Tutorials: https://docs.rapidapi.com/docs/hub-listing-docs-tab ; https://docs.rapidapi.com/docs/tutorials
|
||||
- Provider Analytics: https://docs.rapidapi.com/docs/provider-analytics
|
||||
- Terms of Service (Last Updated 2026-05-13; Restrictions; Provider obligations): https://rapidapi.com/page/terms
|
||||
- robots.txt: https://rapidapi.com/robots.txt
|
||||
- Category pages login-walled / CAPTCHA (observed 2026-09-09): https://rapidapi.com/category/Finance , /Data , /Tools , /Sports , /Weather , /Email , /Business , /Text%20Analysis
|
||||
- Solo-seller earnings thread (2025-08-15): https://community.latenode.com/t/individual-developers-earning-revenue-through-rapidapi-marketplace/36813
|
||||
- Nokia acquisition (2024-11-13): https://techcrunch.com/2024/11/13/nokia-acquires-rapid-the-api-company-once-valued-at-1b/
|
||||
- Marketplace scale claims: https://website.rapidapi.com/products/enterprise-hub/
|
||||
- Similarweb traffic (UNVERIFIED month): https://www.similarweb.com/website/rapidapi.com/
|
||||
- RapidAPI 2022 category popularity post (updated Jan 2023): https://rapidapi.com/blog/most-popular-api/
|
||||
- Third-party scraper field lists (vendor claims): https://apify.com/getascraper/rapidapi-scraper ; https://apify.com/johnvc/rapidapi-marketplace-api/examples/find-rapidapi-popular-apis-by-popularity-score
|
||||
- Zyla monetisation terms: https://zylalabs.com/monetize-your-api
|
||||
- Apify Actor monetisation: https://docs.apify.com/platform/actors/publishing/monetize
|
||||
- Marketplace comparison (2026-01-26): https://blog.apify.com/best-rapidapi-alternatives/
|
||||
- APILayer 15 % claim (third-party): https://geekflare.com/dev/sell-apis/
|
||||
- Finance incumbents: https://rapidapi.com/last10k/api/sec-filings/details ; https://rapidapi.com/asepscareer/api/mboum-finance-data
|
||||
- Local: `/opt/appdata/docker/Business/API Idea/.claude/context.md` ; `research/github-aup-lead-pipeline.md` (2026-09-08)
|
||||
@@ -0,0 +1,284 @@
|
||||
# ConfettiPrintCo (Etsy) — viability evaluation + marketing plan
|
||||
|
||||
- Date: 2026-09-09 · Agent: Fable 5.1 (Prompt F) · Tracks: business_projects 46 (launch row 39, business_ideas 17)
|
||||
- Scope: research-only. Every number below carries a source + date or is marked **UNVERIFIED**.
|
||||
- Access note: `www.etsy.com` (legal pages, seller handbook, search, market pages) returned HTTP 403 to every fetch
|
||||
method available in this run; `help.etsy.com` was reachable and is the official source used for quotes. Etsy demand
|
||||
data (listing counts, search-suggest, bestseller badges) therefore could **not** be measured here — the method is
|
||||
written up for the owner to run in a normal browser (Section 3), and the scoring table is an estimate.
|
||||
|
||||
## Verdict
|
||||
|
||||
**go_with_changes — launch DIGITAL-FIRST (printable journal supplies / junk-journal kits), defer Printify POD to a
|
||||
second phase funded by digital profits.**
|
||||
|
||||
- First sales at $0 ad budget: **realistic but slow — 3–8 weeks, ~55% confidence** for a shop that reaches 30+
|
||||
well-tagged digital listings by end of week 3. Cash outlay to reach 30 listings: 30 × $0.20 = **$6.00** plus Etsy's
|
||||
one-time set-up fee (amount not published on the help page; third-party sources say ~$15 → UNVERIFIED). Zero
|
||||
fulfilment cost, zero shipping, no per-order base cost.
|
||||
- POD (tees/mugs via Printify) at $0 ad budget: **unlikely** within 90 days. Margin per tee after Printify base +
|
||||
shipping + Etsy fees is ~$5–7 on a $23 tee and ~$1–3 on a mug (Section 5); at those margins Etsy Ads are
|
||||
ROAS-negative for a new shop, and organic POD search is the most crowded segment on Etsy (secondary sources, 2026).
|
||||
POD's only zero-cost advantage over digital is a higher order value; its disadvantages are order-level refund/quality
|
||||
risk you do not control, sample costs, and a fee structure that eats 12–15% *of gross including shipping*.
|
||||
- Why "with changes": the idea record says "POD + digital". The evidence says do digital first, and do POD only after
|
||||
digital has proven the design style sells (bestseller signal on ≥1 listing) and has generated ≥$50 profit to fund
|
||||
samples and a $3/day Etsy Ads test.
|
||||
- Compliance is workable: Etsy explicitly allows "seller-prompted AI creations" under **Designed by a seller**, provided
|
||||
the listing **discloses AI use** and (for POD) **discloses the production partner** (quotes in Section 2).
|
||||
|
||||
## 1. Bootstrap-chain ranking input (one line)
|
||||
|
||||
`Etsy/digital: time-to-first-sale at $0 ≈ 21–56 days after 30 listings live; confidence ~55%; evidence = $0.20/listing
|
||||
+ 0 fulfilment cost (help.etsy.com 2026-09-09), 1–8-week first-sale range for new shops (secondary, insightagent.app,
|
||||
UNVERIFIED), 80% net margin on a $4.50 digital kit (Section 5). POD variant: 60–120 days, ~25%.`
|
||||
|
||||
## 2. Rules and fees that bind us (official quotes)
|
||||
|
||||
All quotes fetched 2026-09-09 from `help.etsy.com` unless stated.
|
||||
|
||||
### 2.1 Creativity Standards — what counts as "made by you" (What Can I Sell on Etsy?, help article 360024112614)
|
||||
> "Designed by a seller — As a seller, you can offer your original design as a digital download, or use a third party
|
||||
> to produce or print your design onto a physical item. **This category also includes seller-prompted AI creations.**
|
||||
> Digital items listed in your store must be made and/or designed by you, the seller."
|
||||
|
||||
> "**Seller-prompted AI creations must disclose the use of AI.** If you work with a production partner, you must
|
||||
> disclose that production partner in your relevant listings."
|
||||
|
||||
> "Reselling mass-produced items is not allowed on Etsy."
|
||||
|
||||
Etsy's search snippet for the Creativity Standards legal page (`etsy.com/legal/creativity/`, 403 on fetch, 2026-09-09):
|
||||
"Items 'designed by a seller' are sellers' original designs that are produced or printed by a third party or offered as
|
||||
a digital download … a graphic logo created by a seller and printed onto a phone case by a third-party printer."
|
||||
Seller Handbook article 1275449912004 ("What is Etsy's stance on AI creations?"), via search snippet: "Sellers must
|
||||
disclose within their listing description if an item is created with the use of AI. Etsy allows sellers to use their
|
||||
original prompts in combination with AI tools to create artwork they sell on Etsy."
|
||||
|
||||
Practical implications (my reading, not Etsy's words):
|
||||
- Every ConfettiPrintCo listing goes in **Designed by a seller**; production dropdowns: who made it = "I did"; the
|
||||
description carries one plain line, e.g. *"Designed by ConfettiPrintCo using AI image tools (Google Flow / Nano
|
||||
Banana) with hand-finished layout in Canva."* Third-party guides say the tightened enforcement began 2025-06-10 and
|
||||
that undisclosed AI listings are auto-removed (flagproof.com, inkfluenceai.com — UNVERIFIED as to dates).
|
||||
- Keep source files (prompts, Canva projects, layered exports) per listing — secondary sources report Etsy may ask
|
||||
for proof of authorship (listybox.com 2026, UNVERIFIED).
|
||||
- Do not use third-party brand/character IP in prompts (Intellectual Property Policy applies regardless of AI).
|
||||
|
||||
### 2.2 Production-partner (POD) disclosure
|
||||
Official rule above ("you must disclose that production partner"). Printify's own help article
|
||||
(help.printify.com 4578896595345, via search snippet 2026-09-09): "Printify cannot automatically create a production
|
||||
partner in your Etsy account during publishing. However, if a production partner named 'Printify' already exists in
|
||||
your Etsy account, Printify will automatically assign it to your listings … If no such production partner exists, the
|
||||
listing will default to 'Made by seller', and you'll need to manually update." → Create the "Printify" production
|
||||
partner in Shop Manager **before** the first push, or listings will be mis-declared.
|
||||
|
||||
### 2.3 Digital-download rules (How to Manage Your Digital Listings, help article 115015628347)
|
||||
> "Like handmade items, digital items listed in your store must be made and/or designed by you, the seller."
|
||||
> "When listing a digital item that is an instant download, you'll need to upload the files that the buyer is
|
||||
> purchasing when you create the listing. You can upload up to five digital files. The maximum size for each file is
|
||||
> 20MB." Supported types include .pdf .png .jpg .zip.
|
||||
→ A junk-journal kit = ≤5 files × ≤20 MB. Ship one ZIP (≤20 MB) or one PDF + preview PNGs. Larger kits: split into
|
||||
multi-listing "volumes" (each is another $0.20 listing and another search surface — a feature, not a bug).
|
||||
Returns: digital items are typically non-returnable; sellers set this in shop policies (Etsy policy detail UNVERIFIED
|
||||
in this run — confirm in Shop Manager → Policies).
|
||||
|
||||
### 2.4 Fee schedule (Etsy Fee Basics 360035902374; What are the Fees and Taxes 115014483627)
|
||||
| Fee | Official text (2026-09-09) |
|
||||
|---|---|
|
||||
| Listing | "$0.20 per listing … charged regardless of whether or not the item listed sells … A listing expires after 4 months." |
|
||||
| Auto-renew sold / multi-quantity | "$0.20 … for each additional item purchased" (each digital sale triggers a $0.20 auto-renew) |
|
||||
| Transaction | "6.5% of the total order amount … applies to shipping and gift wrapping (if you charge your customers for those)." |
|
||||
| Payment processing (US) | "3% + $0.25" per order via Etsy Payments (help article 360035902374 via search snippet; body extraction did not capture this line — treat the exact figure as official-but-unquoted) |
|
||||
| Set-up fee | "one-time, non-refundable fee for opening your shop. Cost varies by location … displayed and charged as part of the shop set-up process." Amount not published; ~$15 per third-party reports (UNVERIFIED). |
|
||||
| Offsite Ads | "If you made less than $10,000 on Etsy in the past 365 days, you'll be charged a 15% fee on the order total. If you made at least $10,000, you'll get a discounted fee of 12%. The Offsite Ad fee will never exceed $100 for an order." "If your shop has never made $10,000 USD or more in any consecutive 365-day period, you can opt out." (360000338367) |
|
||||
| Etsy Ads | "The minimum daily budget is $1, and you can change your budget at any time." "All sellers start with a maximum daily budget of $25." Pay-per-click. (360033701174) |
|
||||
| Regulatory operating fee | Applies in some non-US countries; none listed for US in the fetched table — UNVERIFIED for US. |
|
||||
|
||||
**Offsite-ads opt-out decision:** OPT OUT on day 1. A 15% fee on a $4.50 digital order is $0.68 on a $3.62 net —
|
||||
survivable — but on a $23 tee it removes half the margin. Reconsider only after the shop is profitable and only for
|
||||
digital listings (Etsy applies the setting shop-wide, so the answer is stay opted-out while POD exists).
|
||||
|
||||
## 3. Demand method (no paid tools) + product scoring
|
||||
|
||||
Method the owner runs in a normal browser, ~20 min/idea, results typed into the scoring sheet (this is manual browsing
|
||||
of a public site, not account automation — compliant):
|
||||
1. Type the seed phrase into Etsy search; record the **autocomplete suggestions** (demand vocabulary → tags).
|
||||
2. Note the **result count** shown on the results page (supply). Rule of thumb from secondary sources: if five pages
|
||||
are near-identical listings the niche is saturated (insightagent.app trends page, 2026, UNVERIFIED heuristic).
|
||||
3. On page 1, count listings with a **Bestseller / Popular now badge** and record review counts of the top 10
|
||||
(demand proxy). Ratio = (badged listings on p1) / (result count ÷ 1000).
|
||||
4. Seasonality: check the same phrase in Google Trends (free) for a 12-month shape.
|
||||
5. Score 1–5 on demand, 1–5 on competition (inverted), 1–5 on margin, 1–5 on fulfilment ease.
|
||||
|
||||
This run could not execute steps 1–3 (etsy.com returned 403). The table below is therefore an **estimate from secondary
|
||||
sources**, to be replaced by the owner's measured numbers before launch.
|
||||
|
||||
| # | Product idea | Type | Est. demand | Est. competition | Margin | Fulfilment/refund risk | Score /20 | Notes |
|
||||
|---|---|---|---|---|---|---|---|---|
|
||||
| 1 | Vintage-ephemera junk-journal kit (PDF, 20–30 pages) | Digital | 4 | 2 (very crowded; multiple Etsy market pages exist for "digital junk journal kit") | 5 (~80%) | 5 (none) | 16 | UNVERIFIED counts; differentiate by tight theme (e.g. "1920s seed catalog", "Victorian pharmacy labels") |
|
||||
| 2 | Printable sticker sheets for planners (PNG/PDF) | Digital | 4 | 2 | 5 | 5 | 16 | Same buyer as #1; cheap to iterate |
|
||||
| 3 | Themed digital paper packs (12 × 12" backgrounds) | Digital | 3 | 2 | 5 | 5 | 15 | Bundle 12 papers per listing; strong seasonal versions |
|
||||
| 4 | Printable wall-art bundles (set of 3, multiple ratios) | Digital | 4 | 1 (most saturated digital niche) | 5 | 4 (ratio/print complaints) | 14 | AI-art heartland; hardest to rank |
|
||||
| 5 | Niche-profession/hobby tee (Gildan 5000/Bella 3001) | POD | 3 | 1 (most saturated POD niche) | 2 (~$5–7) | 2 (size/print returns) | 8 | Secondary sources: POD ≈ 35% of shops, "most invisible in search" (listifyai 2026, UNVERIFIED) |
|
||||
| 6 | 11 oz mug with themed art | POD | 3 | 2 | 1 (~$1–3) | 2 (breakage) | 8 | Shipping $7–9 first item kills margin (podvector.ai 2026, UNVERIFIED) |
|
||||
| 7 | Tote bag / zip pouch with ephemera collage print | POD | 2 | 3 | 2 | 3 | 10 | Best POD fit with the junk-journal aesthetic; test after digital proves the style |
|
||||
| 8 | Printed sticker/washi-style paper via Printify | POD | 3 | 3 | 2 | 3 | 11 | Bridges digital buyers to physical; Printify catalog availability UNVERIFIED |
|
||||
| 9 | Journal cover / dashboard printables (A5, Happy Planner sizes) | Digital | 3 | 3 | 5 | 5 | 16 | Less crowded than #1; size variants = more listings |
|
||||
|
||||
**Shortlist (launch set) and POD-vs-digital verdict**
|
||||
1. **Themed junk-journal ephemera kits (digital)** — verdict DIGITAL. Margin ~80%, refund risk ≈ nil, fulfilment 0.
|
||||
Ten kits × three themes = 30 listings in three weeks.
|
||||
2. **Planner sticker sheets + journal dashboards (digital)** — verdict DIGITAL. Same buyer, cross-sell inside the PDF.
|
||||
3. **Tote/pouch with the same collage art (POD)** — verdict POD, **phase 2 only**, after (a) one digital listing earns a
|
||||
Bestseller/Popular badge and (b) ≥$50 cumulative profit to pay for one Printify sample and a $3/day ad test.
|
||||
|
||||
## 4. Marketing plan (no social media, no cold outreach, $0 ads)
|
||||
|
||||
### 4.1 Etsy SEO (the zero-cost engine)
|
||||
- **Titles**: lead with the buyer's exact phrase from autocomplete ("Junk Journal Kit Printable Vintage Ephemera …"),
|
||||
≤ 140 chars, most important 40 chars first; one product = one primary phrase.
|
||||
- **Tags**: all 13, multi-word long-tails from autocomplete; no repeats of title words wasted on single words.
|
||||
- **Attributes/category**: Digital → Paper & Party Supplies → Paper → … (owner to pick the deepest category); fill every
|
||||
attribute (occasion, holiday, theme, colour) — attributes are searchable.
|
||||
- **Photos**: 10 slots; slot 1 = flat-lay mockup of printed pages (Canva mockups), slot 2 = "what you get" grid, slot 3
|
||||
= size/format sheet, a short video slot if possible. Preview watermarking on digital.
|
||||
- **Description**: first two lines carry the phrase + the AI-disclosure sentence; then file list, sizes, licence
|
||||
(personal use only), "instant download — no physical item shipped".
|
||||
- **Listing count strategy**: 30 listings by day 21, 60 by day 60. Secondary sources claim 50–100 listings before
|
||||
organic surfacing is reliable (listifyai.net 2026, UNVERIFIED — treat as directional). Each listing = $0.20.
|
||||
- **Renewals**: do not manually renew for "freshness" — costs $0.20 each with no evidence of ranking lift; rely on the
|
||||
4-month auto-expiry only.
|
||||
- **Reviews**: none solicited beyond Etsy's own automated review request. No incentives, no review swaps (policy).
|
||||
|
||||
### 4.2 Launch pricing + shipping
|
||||
- Digital kits: launch $3.99–$4.99, planner stickers $2.49–$3.49, paper packs $3.99. Run a permanent-looking 20%
|
||||
"launch sale" via Etsy's sale tool for the first 30 days (strike-through pricing is a known conversion lever;
|
||||
effect size UNVERIFIED).
|
||||
- POD (phase 2): price tees ≥ $24.99 with free shipping (Etsy's US free-shipping guarantee threshold is $35 for a
|
||||
shop-level badge; single-item free shipping is a listing setting — threshold rule UNVERIFIED in this run).
|
||||
|
||||
### 4.3 Etsy Ads — model, and when to switch on (phase 2, self-funded)
|
||||
Assumptions (secondary, UNVERIFIED): CPC $0.20–$1.50 by category; new-shop conversion 1–2%.
|
||||
| Product | Price | Net before ads | Break-even CPA | Max affordable CPC at 1.5% CVR |
|
||||
|---|---|---|---|---|
|
||||
| Digital kit | $4.50 | $3.62 | $3.62 | **$0.05** → ads lose money; only justified as a traffic/rank-priming test |
|
||||
| Tee | $22.99 | ~$6.81 | $6.81 | $0.10 → same |
|
||||
| Digital bundle | $12.00 | $10.80 | $10.80 | $0.16 → marginal |
|
||||
Conclusion: Etsy Ads cannot be profitable per-click for sub-$15 items at new-shop conversion rates. Use them only as
|
||||
a **$3/day × 14 days = $42 ranking/traffic experiment** once profits cover it, measured by (views, favourites,
|
||||
orders) on advertised vs. non-advertised twins. Expected ROAS ≈ 0.5–1.0 (i.e. a paid experiment, not a profit lever).
|
||||
Expected daily spend is often below budget because Etsy bills per click only (official 360033701174).
|
||||
Owner approval needed for: **$42 one-time test, only after ≥ $50 profit exists.**
|
||||
|
||||
### 4.4 Off-Etsy evergreen assets (no persona)
|
||||
1. **Product-embedded opt-in (free tier):** every PDF's last page links to a bonus pack ("2 extra ephemera pages") on
|
||||
a static page (self-hosted behind the existing Cloudflare tunnel, or GitHub Pages). Email capture via a free tier
|
||||
(Buttondown free ≤100 subscribers, or Brevo free 300 emails/day — tiers UNVERIFIED as of today) → one automated
|
||||
welcome sequence of 3 emails (bonus, "new kit" every 2 weeks, seasonal). Opt-in only; complies with the no-cold-email
|
||||
rule. Cost $0; first signal = first subscriber (needs a first sale); measure: subscribers/sale, click-through to
|
||||
new listings. N8N: watches the new-listing sheet, drafts the email with llama3.1, owner approves send.
|
||||
2. **Pinterest via official API scheduling (conditional):** Pinterest's API v5 supports `publish_at` scheduling ≤30
|
||||
days ahead and third-party schedulers using the official API are permitted (posteverywhere.ai, pinboostr.com 2026 —
|
||||
secondary, UNVERIFIED). BUT secondary sources say a new account needs ~2 weeks of manual daily pinning before any
|
||||
automation, and image-only reposts get suppressed. **Realistic hands-off?** Partially: after a 2-week manual warm-up
|
||||
(~10 min/day) N8N can push 1–3 fresh mockup pins/day with `publish_at` via a Pinterest developer app the owner
|
||||
creates and authorises (this is the platform's official API for one's own account — flag for grill-me as it touches
|
||||
the "no automation of logged-in accounts" rule). If the owner won't do the warm-up, skip Pinterest entirely.
|
||||
Cost $0; first signal 4–8 weeks (Pinterest is slow); measure: Etsy Stats "Pinterest" traffic source.
|
||||
Also free and evergreen: Google indexes Etsy listings — the title/description work in 4.1 doubles as Google SEO.
|
||||
|
||||
### 4.5 30-day calendar (start 2026-09-14)
|
||||
| Days | Owner does | Automation does |
|
||||
|---|---|---|
|
||||
| 1–2 | Open shop (set-up fee), opt out of Offsite Ads, write policies (digital = no returns), create "Printify" production partner now even if unused | N8N: create the `etsy_listings` log sheet (or Postgres table — DB write needs approval) |
|
||||
| 3–7 | Run the demand method on 8 ideas (Section 3), pick 3 themes; generate 30 candidate designs | llama3.1 drafts 3 title/tag variants per keyword set from the owner's autocomplete notes; logged |
|
||||
| 8–21 | Publish 10 listings/week (30 total), each with 6+ mockup images, AI disclosure line, ≤5 files ≤20 MB | Listing-draft generator; QA checklist run (disclosure present? 13 tags? file sizes?) |
|
||||
| 22–30 | Review Etsy Stats weekly: views, favourites, search terms; rewrite the bottom-10 titles | Weekly stats digest from the owner's CSV export (no account automation) |
|
||||
| 30 | Gate: ≥1 sale or ≥300 views → continue to 60 listings; else apply Section 7 | — |
|
||||
|
||||
## 5. Unit economics
|
||||
|
||||
Etsy fee per order (US) = $0.20 renew + 6.5% × gross + (3% × gross + $0.25). Printify figures are secondary
|
||||
(podvector.ai, marginory.com, printify.com blog, 2025–26) — **UNVERIFIED**; the Printify catalogue is app-only.
|
||||
|
||||
| Item | Price | COGS + ship | Etsy fees | Net | Net % | With Offsite Ads 15% |
|
||||
|---|---|---|---|---|---|---|
|
||||
| Junk-journal kit (digital) | $4.50 | $0 | $0.88 | **$3.62** | 80% | $2.94 |
|
||||
| 3-kit bundle (digital) | $12.00 | $0 | $1.59 | **$10.41** | 87% | $8.61 |
|
||||
| Gildan 5000 tee (POD, free ship) | $22.99 | $8.80 + $4.75 = $13.55 | $2.63 | **$6.81** | 30% | $3.36 |
|
||||
| 11 oz mug (POD, free ship) | $17.99 | $6.45 + $6.99 = $13.44 | $2.16 | **$2.39** | 13% | −$0.31 |
|
||||
| Tote bag (POD) | $19.99 | ~$12 (UNVERIFIED) | $2.30 | ~$5.70 | 28% | $2.70 |
|
||||
|
||||
Time-to-first-sale evidence: "Most new Etsy sellers make their first sale within 1 to 8 weeks" (insightagent.app 2026,
|
||||
secondary, UNVERIFIED); Etsy publishes no such statistic. Etsy scale: ~86.5 M active buyers, ~5.6 M active sellers,
|
||||
100 M+ listings (Q4 2025, per printful.com/thunderbit.com round-ups, UNVERIFIED).
|
||||
|
||||
## 6. What N8N + the local model automate (and where decisions are logged)
|
||||
|
||||
Constraint recap: no automation of a logged-in Etsy session; no scraping etsy.com (it 403s anyway). Everything below
|
||||
works from files the owner exports or types.
|
||||
|
||||
| Automation | Input | Model | Output | Decision log (training data) |
|
||||
|---|---|---|---|---|
|
||||
| Keyword sheet expansion | Owner's autocomplete notes (CSV) | llama3.1:8b | 13-tag sets + 3 title variants per listing | `etsy_training_log` rows: `{ts, listing_slug, prompt, candidates, owner_pick, reason}` |
|
||||
| Listing draft + policy QA | Draft text | llama3.1:8b + regex | Description with disclosure line; flags missing tags/attrs, >20 MB files | same table, `stage=qa`, `pass/fail, fix` |
|
||||
| Mockup naming/alt-text | Image filenames | llama3.1:8b | SEO alt-text | `stage=alt_text` |
|
||||
| Weekly stats digest | Etsy Stats CSV (manual export) | nomic-embed-text for search-term clustering + llama3.1 | "rewrite these 10 titles" memo | `stage=weekly_review`, with owner's accept/reject per suggestion |
|
||||
| Sales logging | Etsy orders CSV (manual export) or Etsy Open API v3 OAuth for the owner's own shop (grill-me #2) | — | `etsy_sales` table | `etsy_sales` + `etsy_training_log(stage=sale_attribution)` |
|
||||
| Opt-in email sequence | Subscriber events | llama3.1 drafts, owner approves | Sends via free tier | `stage=email_draft`, owner edits stored as diff |
|
||||
| Pinterest scheduling (conditional) | New listing mockups | — | API `publish_at` pins | `stage=pin`, later joined to Etsy traffic source |
|
||||
|
||||
Storage: `etsy_training_log` and `etsy_sales` in the `business_projects` Postgres DB (creation = DB write → needs
|
||||
approval next session). Every row keeps the model's proposal, the owner's decision, and a one-line reason — the
|
||||
evolution loop for the listing generator.
|
||||
|
||||
## 7. Kill criteria (evaluate at 30 listings / 90 days, whichever is later)
|
||||
|
||||
Kill (pause Etsy, move the bootstrap chain to the next business) if **any two** hold at day 90 with ≥30 listings:
|
||||
1. Cumulative sales < 3.
|
||||
2. Total listing views over the trailing 30 days < 300 (Etsy Stats).
|
||||
3. Favourites ÷ views < 1.5% (listings are seen but not wanted → design/positioning problem, not traffic).
|
||||
4. Zero listings reached page 1 for their primary phrase in the owner's incognito check.
|
||||
Pivot (not kill) triggers: views ≥ 300 but sales = 0 → price/mockup problem, iterate 2 more weeks before applying rule 1.
|
||||
Hard stop regardless: any Creativity-Standards takedown not resolved within 14 days.
|
||||
|
||||
## 8. Grill-me questions (≤6)
|
||||
|
||||
1. Will you do the manual demand method (≈3 hours total) in a browser before designing anything? If not, the scoring
|
||||
table stays a guess and the launch date should slip.
|
||||
2. Is using Etsy's **official Open API with OAuth on your own shop** (for sales logging) acceptable under the
|
||||
"no automation of logged-in accounts" rule, or do we stay with monthly CSV exports?
|
||||
3. Pinterest: are you willing to do a 2-week, 10-min/day manual warm-up so hands-off API pinning is safe — or is
|
||||
Pinterest "social media" and out?
|
||||
4. What is the design-throughput reality: can you produce 10 kits/week (each 20–30 pages, themed, hand-finished in
|
||||
Canva) while also doing the API/KDP work?
|
||||
5. POD gate: do you accept "no Printify listings until one digital listing has a badge and $50 profit exists"?
|
||||
6. Which Postgres DB should hold `etsy_training_log` / `etsy_sales`, and who approves the CREATE TABLE?
|
||||
|
||||
## 9. Sources (all accessed 2026-09-09)
|
||||
|
||||
Official (help.etsy.com, fetched):
|
||||
- What Can I Sell on Etsy? — https://help.etsy.com/hc/en-us/articles/360024112614
|
||||
- How to Manage Your Digital Listings — https://help.etsy.com/hc/en-us/articles/115015628347
|
||||
- Etsy Fee Basics — https://help.etsy.com/hc/en-us/articles/360035902374
|
||||
- What are the Fees and Taxes for Selling on Etsy? — https://help.etsy.com/hc/en-us/articles/115014483627
|
||||
- How Etsy's Offsite Ads Work — https://help.etsy.com/hc/en-us/articles/360000338367
|
||||
- How to Set Up and Manage an Etsy Ads Campaign — https://help.etsy.com/hc/en-us/articles/360033701174
|
||||
- Fees and Listing Multiple Quantities — https://help.etsy.com/hc/en-us/articles/360000344908
|
||||
Official but 403 on fetch (quoted only via search snippets):
|
||||
- Creativity Standards — https://www.etsy.com/legal/creativity/
|
||||
- Fees & Payments Policy — https://www.etsy.com/legal/fees/
|
||||
- Seller Handbook: What is Etsy's stance on AI creations? — https://www.etsy.com/seller-handbook/article/1275449912004
|
||||
- Printify: What should I know about Etsy's Creative Standards policy? — https://help.printify.com/hc/en-us/articles/4578896595345
|
||||
Secondary (UNVERIFIED, directional only):
|
||||
- https://www.insightagent.app/guides/how-to-get-first-sale-on-etsy · https://www.insightagent.app/guides/etsy-ads-minimum-daily-budget-2026
|
||||
- https://www.listifyai.net/blog/etsy-print-on-demand-guide-2026 · https://blog.marmalead.com/etsy-print-on-demand/
|
||||
- https://podvector.ai/articles/printify/costs-and-charges/printify-cost-per-shirt-full-breakdown-for-pod-sellers
|
||||
- https://podvector.ai/articles/printify/costs-and-charges/printify-shipping-costs-full-breakdown-for-pod-sellers
|
||||
- https://marginory.com/guides/printify-cost-per-shirt · https://printify.com/blog/t-shirt-pricing-calculator/
|
||||
- https://flagproof.com/guides/ai-disclosure · https://www.inkfluenceai.com/blog/etsy-ai-disclosure-explained-2026
|
||||
- https://listybox.com/blog/etsy-print-on-demand-rules-policies-disclosure-2026/
|
||||
- https://posteverywhere.ai/blog/schedule-pinterest-pins-api · https://pinboostr.com/is-pinterest-automation-safe/
|
||||
- https://www.printful.com/blog/etsy-statistics · https://thunderbit.com/blog/etsy-stats
|
||||
|
||||
Update rule: re-fetch the three official fee/policy pages before any launch decision; replace Section 3 estimates with
|
||||
the owner's measured counts; re-run Section 5 with live Printify catalogue prices before any POD listing.
|
||||
@@ -0,0 +1,360 @@
|
||||
# Amazon KDP (J.M. Hartley, AI-assisted Q&A books) — viability evaluation + marketing plan
|
||||
|
||||
Prompt E · business_projects 45 (launch row 40 "first 3 titles", idea 14) · Research-only · Fable 5.1 · 2026-09-09
|
||||
Constraints applied: one owner, no persona/social, no cold outreach, **$0 ad budget**, Claude Pro + `claude -p` ($20/mo cap),
|
||||
N8N + Postgres + Ollama llama3.1:8b / nomic-embed-text on 8 GB GPU. Every unproven statement is marked UNVERIFIED.
|
||||
|
||||
---
|
||||
|
||||
## Verdict
|
||||
|
||||
**GO WITH CHANGES.** KDP is the only one of the three businesses where the *marketplace itself is the customer channel*:
|
||||
Amazon search brings the buyer, Amazon prints, ships, collects and pays. That makes it the best fit for an owner who is
|
||||
not a salesman and has no persona. But the exact format in the idea record — a generic **"100,000 Whys"** Q&A book — is
|
||||
already occupied by multiple 2026 indie clones, including one at Amazon BSR **#138 in all Books** (a full-colour kids'
|
||||
visual encyclopedia, published 2026-05-18) and another at #203,637 (Apr 2026). Launching a fourth generic "100,000
|
||||
Whys" in September 2026 is a price-and-cover fight against incumbents with review counts. The changes:
|
||||
|
||||
1. **Do not clone the generic "Whys" title.** Pick 3 *specific-audience* Q&A niches (shortlist in §3) where the top indie
|
||||
titles rank between ~#40k and ~#400k (there is demand, but the winner has no moat).
|
||||
2. **Text-heavy, black-ink, 110–200 pages, $9.99–$12.99** → $2.59–$4.39 royalty per copy (official cost table, §2.3).
|
||||
This is NOT a "low-content" book under Amazon's definition, so Look Inside, free ISBN and Series all remain available.
|
||||
3. **Disclose the text as AI-generated** (Claude writes it; that is "AI-generated" under KDP's definition even after
|
||||
substantial edits — verbatim in §2.1). Plan for it; do not gamble the account.
|
||||
4. **First sales at $0 are realistic but slow**: 21–60 days after the first title goes live, at 0–3 copies/week, driven
|
||||
purely by keyword/category placement. Confidence: LOW-MEDIUM (evidence: indie Q&A titles a few months old sit at BSR
|
||||
#200k–#300k ≈ a copy every 2–4 days; but every 2026 source agrees a zero-review title gets little organic reach).
|
||||
5. **Cash to fund ads arrives late**: KDP pays ~60 days after month end (§5.2), so the first royalty cash lands
|
||||
~November/December 2026 even if the first sale is in late September. Paid Amazon Ads are Phase 2, funded only from
|
||||
that cash, budget $90/month (§4.2).
|
||||
|
||||
**Bootstrap-chain ranking input (one line):** KDP — time-to-first-sale at $0 ≈ **21–60 days** from first publish
|
||||
(2026-09-14 start → first sale ~Oct 5–Nov 13); confidence **LOW-MEDIUM**; evidence = independently published Q&A
|
||||
paperbacks from Jan–May 2026 ranking #43,751 / #203,637 / #299,575 in Books with no visible ad or persona (§3), against
|
||||
consensus that zero-review books get "almost no organic visibility" (§5.1). Cash from that sale: +60 days after month end.
|
||||
|
||||
---
|
||||
|
||||
## 1. Rules that bind us (official quotes)
|
||||
|
||||
### 1.1 AI-generated vs AI-assisted, and the disclosure duty
|
||||
Source: KDP Content Guidelines, https://kdp.amazon.com/en_US/help/topic/G200672390 (fetched 2026-09-09).
|
||||
|
||||
> **AI-generated:** "text, images, or translations created by an AI-based tool" — this still counts "even if you applied
|
||||
> substantial edits afterwards."
|
||||
> **AI-assisted:** content you created yourself and used AI tools to "edit, refine, error-check, or otherwise improve" or
|
||||
> to "brainstorm and generate ideas."
|
||||
> "You are required to inform us of AI-generated content (text, images, or translations) when you publish a new book or
|
||||
> make edits to and republish an existing book."
|
||||
|
||||
**Applied:** the workflow "Claude drafts the Q&A text, owner edits" is *AI-generated text* → tick the AI-generated
|
||||
disclosure for text at every publish/republish. Canva-made covers using Canva's AI image tools are *AI-generated images*
|
||||
→ disclose those too; covers assembled from stock/vector elements by hand are not. Disclosure is not shown to shoppers
|
||||
on the product page (UNVERIFIED as of 2026-09-09 — several third-party guides say so; Amazon's page does not state it
|
||||
either way). Failure to disclose = removal, and repeat cases = account termination (secondary sources: KDP Builder,
|
||||
https://kdpbuilder.com/blog/kdp-ai-disclosure-rules, 2026; The Authors Guild, 2023-09-27,
|
||||
https://authorsguild.org/news/amazon-adds-to-kdp-generative-ai-policy-caps-daily-self-publishing-uploads/).
|
||||
|
||||
### 1.2 Misleading / disappointing content
|
||||
Same page: Amazon prohibits "descriptive content meant to mislead customers or that doesn't accurately represent the
|
||||
content of the book" and content that is "typically disappointing to customers" (detail in the Quality Content Guide,
|
||||
which returned 404 at the URL we tried — UNVERIFIED wording). **Applied:** a title claiming "100,000 Whys" that contains
|
||||
~300 questions is exactly the kind of numeric over-claim that can be flagged and that earns 1-star reviews. Put the true
|
||||
count in the subtitle ("350 questions…").
|
||||
|
||||
### 1.3 Volume cap: 3 new titles per day
|
||||
The Authors Guild reported Amazon's 2023-09-27 announcement capping new titles at **three per day**, quoting Amazon:
|
||||
"While we have not seen a spike in our publishing numbers, in order to help protect against abuse, we are lowering the
|
||||
volume limits we have in place on new title creations." Third-party 2026 guides say the cap is still in force
|
||||
(https://kdpbuilder.com/blog/ai-book-publishing-2026). Irrelevant to a 3-title launch; relevant to any "upload 50
|
||||
variants" idea — which this plan rejects anyway.
|
||||
|
||||
### 1.4 Low-content definition — and why Q&A books are *not* low-content
|
||||
Source: https://kdp.amazon.com/en_US/help/topic/GGE5T76TWKA85DJM (fetched 2026-09-09).
|
||||
|
||||
> "A low-content book has minimal or no content on the interior pages. Low-content books are generally repetitive, and
|
||||
> designed to be filled in by the user." Examples: notebooks, planners, diaries/journals, prompt journals, log books.
|
||||
> Not generally low-content: "Novels, nonfiction, puzzle books, coloring books, … children's books."
|
||||
> Low-content books: "not eligible for the free KDP ISBN"; Look Inside/Read Sample "not supported for books published
|
||||
> without ISBN"; Expanded Distribution "not supported for low-content books, regardless of the ISBN option";
|
||||
> "not eligible for creating a series."
|
||||
|
||||
**Applied:** a text-heavy Q&A book is ordinary nonfiction → free ISBN, Read Sample (Look Inside), Series page and
|
||||
Expanded Distribution all available. Do **not** tick "low-content" at upload.
|
||||
|
||||
### 1.5 Keywords and categories
|
||||
Source: https://kdp.amazon.com/en_US/help/topic/G201097560 (fetched 2026-09-09).
|
||||
> "Keywords can be a word or a phrase. For best results, we recommend using phrases that are 2-3 words long."
|
||||
> Prohibited keywords include: "Other authors", "Books by other authors", "Sales rank (e.g., 'bestselling')",
|
||||
> "Advertisements or promotions (e.g., 'free')", content unrelated to the book, HTML tags.
|
||||
> "During title setup, you'll choose up to three categories from a list based on your primary audience and marketplace."
|
||||
The page does not state the number of keyword slots or per-slot character cap (commonly cited as 7 slots × 50
|
||||
characters — UNVERIFIED on the official page as of 2026-09-09).
|
||||
|
||||
### 1.6 Royalty rate and print cost (US marketplace)
|
||||
Sources: https://kdp.amazon.com/en_US/help/topic/G201834330 and https://kdp.amazon.com/en_US/help/topic/G201834340
|
||||
(both fetched 2026-09-09).
|
||||
- Paperback royalty: "50% royalty rate" at list price ≤ $9.98 USD; "60% royalty rate" at list price ≥ $9.99 USD.
|
||||
Expanded Distribution: 40% of list minus print cost. Royalty = rate × list − printing cost. (The idea record's
|
||||
"up to 60%" is correct, but only above the $9.99 threshold.)
|
||||
- Printing cost, black ink, regular trim: 24–110 pages "2.30 USD per book" flat; 110–828 pages "1.00 USD per book" +
|
||||
"0.012 USD per page". Large trim: 2.84 flat / 1.00 + 0.017 per page.
|
||||
- Standard colour (72–600 pp): 1.00 + 0.0255/page regular trim; 1.00 + 0.0402/page large trim.
|
||||
- Premium colour: 3.60 flat (24–40 pp) or 1.00 + 0.065/page (42–828 pp); large trim 4.20 / 0.08.
|
||||
- Minimum list price is derived from print cost so royalty never goes negative; max list $250.
|
||||
|
||||
### 1.7 Review rules
|
||||
No review manipulation: no paid, incentivised, family, or swap reviews; no asking reviewers for a rating in exchange for
|
||||
anything. The only allowed asks are (a) Amazon's own "Request a review" mechanics for the buyer, and (b) a neutral
|
||||
"if this helped, a review helps others find it" line in the back matter. (Amazon Community Guidelines — URL not fetched
|
||||
this run; UNVERIFIED quote, principle is well-established.)
|
||||
|
||||
### 1.8 Platform-terms constraints from our own rules
|
||||
No browser automation of the logged-in KDP/Author Central/Ads consoles (ban risk — see
|
||||
research/hyperagent-marketing-fit.md). Scraping amazon.com product pages programmatically is against Amazon's
|
||||
Conditions of Use (UNVERIFIED verbatim this run); the automation in §6 therefore uses *manual* weekly BSR reads entered
|
||||
into N8N, plus CSV exports the owner downloads from KDP Reports by hand.
|
||||
|
||||
---
|
||||
|
||||
## 2. Niche selection method (no paid tools)
|
||||
|
||||
**Signal set (all free, all read by a human in a browser, entered into N8N form → Postgres):**
|
||||
1. **BSR of the top 5 indie results** for the exact search phrase. Paperback BSR→sales heuristics (third-party,
|
||||
proprietary formula is unpublished): "A paperback at BSR 10,000 sells around 6 copies/day";
|
||||
"Under 10,000 indicates strong daily sales. Under 100,000 suggests consistent movement. Above 500,000 usually means
|
||||
infrequent sales." (BookBloom https://www.bookbloom.io/blog/bsr-to-sales-calculator-guide, 2026; Kindlepreneur
|
||||
https://kindlepreneur.com/amazon-kdp-sales-rank-calculator/, 2026). Working rule of thumb used below:
|
||||
#10k ≈ 6/day, #50k ≈ 1–2/day, #100k ≈ 0.5–1/day, #300k ≈ 1 every 2–4 days, >#1M ≈ near zero.
|
||||
2. **Search-suggest** (type the phrase in Amazon Books search, note autocomplete completions) — demand exists if Amazon
|
||||
completes it.
|
||||
3. **Result count / competitor density**: number of *independently published* titles on page 1 and their publication
|
||||
year; an all-2026 page 1 = active gold-rush, an all-2020–2022 page 1 with BSR >1M = dead niche.
|
||||
4. **Review-count moat**: reviews on the #1 indie result. <50 = beatable with a better listing; >500 = avoid.
|
||||
5. **Category best-seller lists** (`/zgbs/books/<id>`): e.g. Trivia & Fun Facts (books/11871), Trivia (books/4452),
|
||||
Quizzes new releases (books/4440). Read the top 20: how many are indie, how many are Q&A format.
|
||||
6. **Print-cost fit**: does the winner need full colour (kids' encyclopedias do) or is black-ink acceptable?
|
||||
|
||||
**Scoring (0–3 each, max 18):** demand (BSR of #1 indie) · beatability (review moat inverse) · freshness (page-1 year
|
||||
spread) · black-ink fit · fit with owner's evergreen/no-persona constraint · series potential.
|
||||
|
||||
**Go threshold:** ≥ 12/18 AND at least one indie title under #300k AND #1 indie under 500 reviews.
|
||||
|
||||
### 2.1 Candidate niche table (9 candidates, evidence read 2026-09-09 via search snippets of amazon.com pages —
|
||||
BSRs are point-in-time and drift daily; RE-READ before choosing)
|
||||
|
||||
| # | Niche (Q&A format) | Evidence (title · pub date · BSR in Books) | Demand | Beatable | Fresh | B/W fit | Fit | Series | Score |
|
||||
|---|---|---|---|---|---|---|---|---|---|
|
||||
| 1 | Kids "100,000 Whys" generic (ages 6–12) | *100,000 Whys for Kids: Full-Color Visual Encyclopedia* (E. Wonder, 2026-05-18) **#138**, #1 in two kids' categories; *100,000 Whys for Kids 2026* (C.D. Still, 2026-04-30) #203,637; *Big Book of Why for Kids* (J. Kinsley, 2026-07-22) BSR UNVERIFIED | 3 | 0 (winner is full-colour, #138, review count UNVERIFIED but high) | 3 | 0 (colour needed to compete) | 2 | 3 | **11** |
|
||||
| 2 | Kids "why" — single theme (space / animals / human body), black-ink with line art | Same niche demand; no direct BSR evidence for single-theme indie title — UNVERIFIED | 2 | 2 | 3 | 2 | 3 | 3 | **15** |
|
||||
| 3 | Adults "100,000 Whys" encyclopedia | *The Ultimate 100,000 Whys Book for Adults Encyclopedia* exists (search snippet, BSR UNVERIFIED) | 2 | 1 | 3 | 2 | 3 | 2 | **13** |
|
||||
| 4 | Trivia / Q&A for seniors, large print | *The Ultimate Trivia Book for Seniors* (Theparchmentco08, 2026-03-07) **#299,575** | 2 | 2 | 3 | 3 | 3 | 3 | **16** |
|
||||
| 5 | Couples / date-night questions ("would you rather") | *Would You Rather Book for Couples* (Sparky Pages, 2026-01-15) **#43,751**; *Ultimate Conversation Starters* (2026-02-02) #1,391,735; *Questions for Couples* (Dreamstorm) #6,721,872 | 3 | 1 (huge variance = winner-take-most) | 3 | 3 | 2 | 2 | **14** |
|
||||
| 6 | Bible trivia | Four indie titles 2020–2023, BSR #1,376,363 / #4,416,765 / #5,291,046 / #8,433,368 | 0 | 2 | 0 | 3 | 3 | 2 | **10** |
|
||||
| 7 | General "2026 trivia quiz" | *2026 Ultimate Trivia Quiz* (MW Syndicate, 2026-03-18, 90 pp) #540,518; *Trivia Quiz Book Jan 2026* (Carruthers, 227 pp) BSR UNVERIFIED | 1 | 2 | 3 | 3 | 2 | 1 | **12** |
|
||||
| 8 | Job-interview Q&A for one profession (e.g. nursing, IT support) | No evidence fetched this run — UNVERIFIED | ? | ? | ? | 3 | 3 | 3 | UNSCORED |
|
||||
| 9 | "Why does my dog/cat…" pet-behaviour Q&A | No evidence fetched this run — UNVERIFIED | ? | ? | ? | 3 | 3 | 3 | UNSCORED |
|
||||
|
||||
### 2.2 Shortlist (3)
|
||||
1. **Q&A for seniors, large print (16/18)** — "Why did we…?" nostalgia + general-knowledge Q&A, 16–18 pt type, 120–150
|
||||
pages black ink. Top indie at #299,575 six months after launch means real but unguarded demand; large-print is a
|
||||
genuine reader need Amazon's category system surfaces ("Large Print" filters). Buyers are often adult children
|
||||
buying for parents → gift traffic Q4. Series: *Vol. 2*, *Music & Movies edition*, *Science edition*.
|
||||
2. **Kids "why", single theme, black ink + line art (15/18)** — e.g. *Why Does the Moon Follow Me? 200 Space Whys for
|
||||
Curious Kids 6–10*. Sidesteps the full-colour #138 incumbent by cost (B/W print ≈ $2.44 at 120 pp vs ≈ $4.06 colour)
|
||||
and by specificity in search ("space questions for kids"). Riskiest of the three: kids' categories draw the most AI
|
||||
clones and Amazon's tightened 2026 enforcement (aipolicydesk.com, 2026, secondary). Illustrations: Canva line art;
|
||||
disclose if AI-generated.
|
||||
3. **Couples / family date-night question book (14/18)** — the #43,751 title shows a fresh Jan-2026 indie can sell
|
||||
~1–2/day; the two >#1M titles show most entrants get nothing, i.e. listing quality decides. Differentiate by a
|
||||
format hook (e.g. "52 weeks, 7 questions a week" — gives a *why to buy now* and a natural *Vol. 2*).
|
||||
|
||||
Dropped: #1 generic "100,000 Whys" (owner's original plan) — demand is proven but the fight is against a #138 colour
|
||||
incumbent; revisit only as a *later* colour title funded by profits. #6 Bible trivia — dead by BSR. #7 general trivia —
|
||||
weak BSR, no series hook.
|
||||
|
||||
---
|
||||
|
||||
## 3. Marketing plan
|
||||
|
||||
### 3.1 On-Amazon levers (cost $0; these ARE the customer-acquisition channel)
|
||||
| Lever | What to do | Cost | First signal | Measured by | N8N/local-model share |
|
||||
|---|---|---|---|---|---|
|
||||
| Title + subtitle | Exact search phrase first ("Trivia for Seniors: Large Print…"), true question count, audience, format hook. No "bestselling", no other authors' names (§1.5). | $0 | Impressions in KDP dashboard? none at $0 — signal = first organic sale | BSR (manual weekly), KDP sales report | Claude drafts 10 variants; llama3.1 scores each against a rubric (phrase-first, ≤200 chars, no banned words); decision logged |
|
||||
| 7 keyword slots | 2–3-word buyer phrases from search-suggest; never repeat title words | $0 | 7–14 days after indexing | Searching the exact phrase + "J.M. Hartley" appears on p.1–3 | N8N stores suggest lists; llama3.1 dedups/clusters |
|
||||
| 3 categories | 1 broad (Trivia & Fun Facts), 2 narrow (Large Print / Games & Activities / Children's Q&A). Pick ones where #100 in category is > #500k BSR so a trickle reaches the "Best Sellers in X" strip | $0 | 14–30 days | Category rank on the product page | Manual read, logged |
|
||||
| Read Sample (Look Inside) | Enabled automatically with ISBN (§1.4); first 10% must show 8–10 real Q&As, not front matter | $0 | Conversion (orders ÷ page reads — visible only via Ads console later) | Sales per BSR | — |
|
||||
| A+ Content | 3 modules: what's inside (spread photo from Canva mock-up), who it's for, series strip. Free for published KDP books (https://kdp.amazon.com/en_US/help/topic/GHL7P99B7AA543CN, 2026-09-09) | $0 | at publish | Same | Claude drafts copy |
|
||||
| Description | 150-word buyer-language description; HTML allowed in description not keywords | $0 | at publish | Same | Claude drafts, llama3.1 checks banned claims |
|
||||
| Launch pricing | $9.99 list (60% tier starts here) for 30 days → $12.99 once the first 3 reviews land | $0 | — | Royalty per copy $2.59 → $4.39 (§4.1) | price change is manual |
|
||||
| Series page | Create a KDP Series per niche from title 1 (eligible: not low-content) | $0 | Vol. 2 | Series page views (Author Central) | — |
|
||||
| Author Central + Goodreads author page (Amazon-owned) | Set up once, bio, all titles; evergreen, zero posting | $0 | — | — | manual, once |
|
||||
| Back-matter opt-in | Last page: "Get 50 bonus questions free → jmhartley.example/bonus" (opt-in, GDPR-clean) | $0 | first sale + 1 week | opt-ins ÷ sales | N8N webhook → Postgres → sends PDF via SMTP; all logged |
|
||||
| Reviews | Only Amazon-compliant asks (§1.7) | $0 | 30–90 days | review count | none (never automate) |
|
||||
|
||||
### 3.2 Amazon Ads — Phase 2 only, funded by KDP royalties (starts ≥ first royalty payout, ~Dec 2026)
|
||||
- **Monthly budget to approve: $90** = $3/day, one Sponsored Products *automatic* campaign per title ($1/day each) for
|
||||
14 days, then move the 5–10 converting search terms into one *manual exact-match* campaign per title. Third-party
|
||||
2026 benchmarks: book Sponsored Products CPC "$0.15 to $0.45 per click" (Vappingo,
|
||||
https://www.vappingo.com/word-blog/amazon-ads-budget-bidding-books/, 2026); Books ACoS "from 19%" (Ad Badger,
|
||||
https://www.adbadger.com/blog/amazon-advertising-stats/, 2026); "$30 monthly budget is a reasonable starting point"
|
||||
(Hillshire Media, https://hillshiremedia.co/blog/book-sales-amazon-kdp/, 2026). Other guides quote CPC $0.80–2.50
|
||||
(eva.guru) — competitive genres; Q&A/trivia is not one.
|
||||
- **Model (per title, $12.99 list, $4.39 royalty):** breakeven ACOS = 4.39 / 12.99 = **34%**. At CPC $0.30 and 8%
|
||||
conversion → $3.75/sale → ACOS 29% → +$0.64 net per ad-sale, plus the organic-rank lift. At CPC $0.45 and 5% →
|
||||
$9.00/sale → ACOS 69% → −$4.61 per sale → **pause** that title's ads. Expected first-month ACOS 40–70% (auto campaign
|
||||
discovery), target ≤35% by month 3.
|
||||
- **Kill rule for ads:** any campaign with ≥ 30 clicks and 0 orders is paused; any title with 90-day ACOS > 50% has its
|
||||
ads stopped, not "optimised".
|
||||
- Ads console work is manual (no automation of logged-in accounts). N8N ingests the weekly *downloaded* campaign CSV.
|
||||
|
||||
### 3.3 Off-Amazon evergreen assets (no persona, $0)
|
||||
1. **Author page** `jmhartley.<domain>` — one static page on existing infra (Traefik/cloudflared already in place;
|
||||
nothing to install now): all titles with Amazon links, "who these books are for", opt-in form. Purpose: the back-
|
||||
matter link target, plus a Google-indexable page for the pen name. Cost $0; signal = first opt-in; measured by
|
||||
opt-ins/month in Postgres.
|
||||
2. **Bonus-questions email opt-in inside the book** — the only list-building this owner needs; never cold, never
|
||||
scraped. Sequence: (1) deliver PDF, (2) 30 days later "Vol. 2 is out" when true. Runs on N8N + SMTP from the
|
||||
product itself. Cost $0; signal = first opt-in; measured by opt-ins ÷ sales (target 3–5%).
|
||||
(Explicitly out: TikTok/Instagram/Pinterest, book-promo newsletters that require paying, reviewer outreach.)
|
||||
|
||||
### 3.4 30-day calendar (from 2026-09-14)
|
||||
| Day | Action |
|
||||
|---|---|
|
||||
| 1–2 | Re-read BSRs for the 3 shortlist niches and the 2 UNSCORED ones (§2.1) in a browser; pick niche A. Log scores. |
|
||||
| 3–7 | Title A: Claude drafts 300 Q&As in 10 batches (`claude -p`, within $20/mo); owner fact-checks 10% at random; llama3.1 flags duplicates/unsafe claims. |
|
||||
| 8–10 | Canva interior (6×9, 120–150 pp, black ink) + cover; title/subtitle/keywords/categories per §3.1; back-matter opt-in page. |
|
||||
| 11 | Publish Title A with AI-generated text disclosure; create Series; enable Read Sample. (Review takes up to 72 h.) |
|
||||
| 12–14 | Author Central + Goodreads pages; static author page + N8N opt-in webhook. |
|
||||
| 15–24 | Title B (niche 2) — same pipeline; log every decision. |
|
||||
| 25–30 | Title C (niche 3). Weekly BSR/keyword rank read for A and B into Postgres. |
|
||||
| 31+ | Day-45 and Day-90 checkpoints (§5.3). Ads only after first royalty payout. |
|
||||
|
||||
---
|
||||
|
||||
## 4. Viability — unit economics and time-to-first-sale
|
||||
|
||||
### 4.1 Per-title economics (official cost table §1.6)
|
||||
| Spec | Print cost | List | Royalty rate | Royalty/copy |
|
||||
|---|---|---|---|---|
|
||||
| 120 pp, black ink, 6×9 | 1.00 + 120×0.012 = **$2.44** | $9.99 | 60% | **$3.55** |
|
||||
| 200 pp, black ink, 6×9 | 1.00 + 200×0.012 = **$3.40** | $9.99 | 60% | $2.59 |
|
||||
| 200 pp, black ink, 6×9 | $3.40 | $12.99 | 60% | **$4.39** |
|
||||
| 120 pp, standard colour | 1.00 + 120×0.0255 = $4.06 | $14.99 | 60% | $4.93 |
|
||||
| 100 pp, black ink (≤110 pp flat) | $2.30 | $9.99 | 60% | $3.69 |
|
||||
Production cost per title at $0 tools: ~10–15 owner-hours + a share of the $20/mo Claude credit. Break-even on time
|
||||
alone is meaningless; the honest metric is **royalty per title per month** at the 90-day mark (kill rule §5.3).
|
||||
|
||||
### 4.2 Scenarios (3 titles live by day 30)
|
||||
| Scenario | Copies/month total | Royalty/month | Evidence |
|
||||
|---|---|---|---|
|
||||
| Miss (most entrants) | 0–3 | $0–11 | the >#1M BSR titles in §2.1 |
|
||||
| Trickle (typical fresh indie Q&A) | 10–25 | $35–110 | #200k–#300k BSR titles ≈ 0.3–0.5/day each |
|
||||
| Hit one niche (top-3 indie result) | 40–90 | $140–400 | #43,751 title ≈ 1–2/day |
|
||||
$90/mo ad budget is fundable only in the Trickle+ scenario, and only from ~Dec 2026 (payout lag).
|
||||
|
||||
### 4.3 Time-to-first-sale evidence
|
||||
- Fresh indie Q&A paperbacks from Jan–May 2026 hold BSR #43k–#300k months later (§2.1) — i.e. they *do* sell without
|
||||
visible marketing.
|
||||
- Counter-evidence: "a brand new book with zero reviews gets almost no organic visibility, regardless of how good the
|
||||
keywords are" (Vinayak Ramesh, Medium, 2026); "Without a launch strategy, the same book may take 12+ months or never
|
||||
reach [100 sales]" (Hillshire Media, 2026). "Most first-time self-published authors take 6–12 months to earn $1,000
|
||||
from their first book" (River Editor, https://rivereditor.com/guides/how-to-self-publish-amazon-kdp-make-1000-dollars-2026, 2026).
|
||||
- Market: "U.S. produced an estimated 3.5 million self-published titles in 2025 (up ~39% YoY) … over 30% of new titles
|
||||
in 2026 have some form of AI disclosure" (KDP Builder State of AI Self-Publishing 2026,
|
||||
https://kdpbuilder.com/reports/state-of-ai-self-publishing-2026 — third-party estimate, UNVERIFIED methodology).
|
||||
- **Net:** first sale at $0 = *realistic, 21–60 days, LOW-MEDIUM confidence*; first *$100 month* at $0 = unlikely before
|
||||
day 90 unless one niche hits.
|
||||
|
||||
### 4.4 Cash timing (matters for the bootstrap chain)
|
||||
"Payment is usually sent around 60 days after the end of that month" (Hawk Notch,
|
||||
https://hawknotch.com/when-does-amazon-kdp-pay-royalty-timeline/, 2026; KDP's official payment page not fetched this run
|
||||
— UNVERIFIED verbatim). A late-September sale pays at end of November. KDP also needs a bank account + tax interview
|
||||
before the first payout (standard, UNVERIFIED this run).
|
||||
|
||||
---
|
||||
|
||||
## 5. What N8N + the local model automate (and where decisions are logged)
|
||||
|
||||
Guardrails: nothing touches a logged-in Amazon session; no scraping of amazon.com; human reads BSR/suggest in a browser
|
||||
and submits an N8N form. Claude Pro/`claude -p` writes content; llama3.1:8b does scoring, dedup, rubric checks;
|
||||
nomic-embed-text dedups questions across titles (avoid self-plagiarism across the series).
|
||||
|
||||
| Job | Trigger | Local-model role | Output | Training-log table (Postgres, business_projects DB) |
|
||||
|---|---|---|---|---|
|
||||
| Niche scan | Weekly form (owner enters phrase, top-5 BSRs, review counts, page-1 years) | llama3.1 computes the 6-axis score, writes a 2-line rationale | `kdp_niche_scores` row + ntfy summary | `kdp_decisions(kind='niche', input_json, model_output, human_override, outcome_90d)` |
|
||||
| Keyword tracking | Weekly form (rank of our title for its 7 phrases) | none | `kdp_keyword_rank` | same table, kind='keyword' |
|
||||
| Sales logging | Owner drops KDP Reports CSV into a watched folder | none | `kdp_sales_daily` | kind='sales' (no decision; outcome feed) |
|
||||
| Q&A generation QA | After each `claude -p` batch | llama3.1 flags duplicates, unsafe/medical/legal claims, wrong reading level; nomic-embed dedup vs existing titles | accept/reject list | kind='content_qa', with human accept/reject as label |
|
||||
| Title/keyword drafting | Per title | llama3.1 rubric-scores Claude's variants | shortlist | kind='listing', chosen variant + later BSR as outcome |
|
||||
| Opt-in delivery | Webhook from author page | none | PDF email, contact row | `kdp_optins` (consent timestamp, source title) |
|
||||
| Ads CSV ingest (Phase 2) | Owner-downloaded CSV | llama3.1 proposes pause/keep per search term against §3.2 rules | proposal only; owner acts in console | kind='ads', proposal vs action vs 30-day ACOS |
|
||||
Every row keeps `model`, `prompt_version`, `human_override` so the log is usable as training/eval data
|
||||
(feedback_always_include_training_loop). None of this is built in this run (research-only); it uses existing N8N/Postgres.
|
||||
|
||||
---
|
||||
|
||||
## 6. Kill criteria (3 titles / 90 days from first publish)
|
||||
|
||||
Pause the business and move its slot to the next bootstrap-chain candidate if, at day 90:
|
||||
1. **< 10 total paid copies** across 3 titles at $0 ads (Miss scenario), **or**
|
||||
2. no title has reached BSR < #500k on any weekly read after day 45, **or**
|
||||
3. any title removed/blocked for policy reasons (indicates the AI-disclosure/kids-category risk is not manageable), **or**
|
||||
4. royalty per owner-hour < $3 in the best 30-day window (compare with the API/Etsy tracks).
|
||||
Continue (and approve the $90/mo ads) if ≥ 25 copies or any title < #150k. Re-scan niches (not "write more") if
|
||||
titles are stuck between the two.
|
||||
|
||||
---
|
||||
|
||||
## 7. Grill-me questions (≤ 6)
|
||||
1. The generic "100,000 Whys" title you named is already held by a #138-in-Books colour incumbent — are you willing to
|
||||
drop that exact concept for the three specific niches, or is the "Whys" brand the point?
|
||||
2. AI-generated disclosure is mandatory for Claude-written text: any objection to ticking it on every title?
|
||||
3. Black-ink 6×9 at $9.99–$12.99 caps royalty at ~$3.5–4.4/copy; the colour kids' winner needs ~$15+ pricing and real
|
||||
illustration — which cost/effort profile do you want for titles 1–3?
|
||||
4. Are you OK that Amazon pays ~60 days after month end, so this business cannot fund any ad budget before ~December 2026?
|
||||
5. Which two UNSCORED niches (profession interview Q&A; pet-behaviour "why") should be BSR-read before day 2?
|
||||
6. Kill criteria: 10 copies / 90 days / BSR #500k — too strict, too loose?
|
||||
|
||||
---
|
||||
|
||||
## 8. Sources (all accessed 2026-09-09)
|
||||
Official Amazon KDP
|
||||
- Content Guidelines (AI definitions, disclosure, misleading content): https://kdp.amazon.com/en_US/help/topic/G200672390
|
||||
- Low-Content Books: https://kdp.amazon.com/en_US/help/topic/GGE5T76TWKA85DJM
|
||||
- Metadata guidelines (keywords/categories): https://kdp.amazon.com/en_US/help/topic/G201097560
|
||||
- Paperback royalty: https://kdp.amazon.com/en_US/help/topic/G201834330
|
||||
- Printing cost: https://kdp.amazon.com/en_US/help/topic/G201834340
|
||||
- A+ Content: https://kdp.amazon.com/en_US/help/topic/GHL7P99B7AA543CN (search-surfaced; not fetched)
|
||||
- Quality content guide URL tried (G202184290) → 404; low-content topic GHT4YBXZCQ9SQ4E9 → 404.
|
||||
Policy reporting
|
||||
- Authors Guild, 2023-09-27: https://authorsguild.org/news/amazon-adds-to-kdp-generative-ai-policy-caps-daily-self-publishing-uploads/
|
||||
- KDP Builder AI disclosure / 2026 guide: https://kdpbuilder.com/blog/kdp-ai-disclosure-rules , https://kdpbuilder.com/blog/ai-book-publishing-2026
|
||||
- KDP Builder State of AI Self-Publishing 2026: https://kdpbuilder.com/reports/state-of-ai-self-publishing-2026
|
||||
Market evidence (amazon.com product pages via search snippets; BSR point-in-time)
|
||||
- 100,000 Whys for Kids (Wonder, #138): https://www.amazon.com/100-000-Whys-Kids-Encyclopedia/dp/B0H2B1DH6L
|
||||
- 100,000 Whys for Kids 2026 (Still, #203,637): https://us.amazon.com/100-000-Whys-Kids-2026/dp/B0GZ4K34ZG
|
||||
- Big Book of Why for Kids (Kinsley): https://www.amazon.com/BIG-BOOK-WHY-KIDS-Collection/dp/B0H9ZQYPKV
|
||||
- Would You Rather for Couples (#43,751): https://www.amazon.com/Would-You-Rather-Book-Couples/dp/B0GH62PXLF
|
||||
- Ultimate Conversation Starters (#1,391,735): https://www.amazon.com/Ultimate-Conversation-Starters-Book-Professional/dp/B0GL31X6QB
|
||||
- Questions for Couples (#6,721,872): https://www.amazon.com/Questions-Couples-Exciting-Conversation-Starters/dp/B0BTNTZGDF
|
||||
- Trivia for Seniors large print (#299,575): https://www.amazon.com/Ultimate-Trivia-Book-Seniors-Questions/dp/B0GRNS8RQP
|
||||
- 2026 Ultimate Trivia Quiz (#540,518): https://www.amazon.com/2026-Ultimate-Trivia-Quiz-Knowledge/dp/B0GTHPGW8Y
|
||||
- Bible trivia titles (#1.3M–#8.4M): https://www.amazon.com/Bible-Trivia-Game-Adults-Questions/dp/B09XBS7R28 and three siblings (§2.1)
|
||||
- Best Sellers lists: https://www.amazon.com/Best-Sellers-Books-Trivia-Fun-Facts/zgbs/books/11871 (fetch returned 503; not read this run)
|
||||
BSR heuristics / ads / timelines (third-party)
|
||||
- BookBloom BSR guide: https://www.bookbloom.io/blog/bsr-to-sales-calculator-guide
|
||||
- Kindlepreneur calculator: https://kindlepreneur.com/amazon-kdp-sales-rank-calculator/
|
||||
- Vappingo ads budgets: https://www.vappingo.com/word-blog/amazon-ads-budget-bidding-books/
|
||||
- Ad Badger benchmarks: https://www.adbadger.com/blog/amazon-advertising-stats/
|
||||
- Hillshire Media first 100 sales: https://hillshiremedia.co/blog/book-sales-amazon-kdp/
|
||||
- Eva.guru KDP ads tutorial: https://eva.guru/blog/amazon-ads-tutorial-for-kdp/
|
||||
- River Editor $1,000 timeline: https://rivereditor.com/guides/how-to-self-publish-amazon-kdp-make-1000-dollars-2026
|
||||
- Hawk Notch payout timeline: https://hawknotch.com/when-does-amazon-kdp-pay-royalty-timeline/
|
||||
- Vinayak Ramesh (Medium) organic visibility: https://medium.com/@thevinayakramesh/the-truth-about-amazon-kdp-publishing-in-2026-nobody-wants-to-hear-57fc70037ffd
|
||||
- Naomi Jane, low/medium-content 2026: https://naomijane.substack.com/p/low-and-medium-content-books-on-amazon
|
||||
- AI Policy Desk enforcement 2026: https://www.aipolicydesk.com/blog/amazon-kdp-ai-generated-content-surge-enforcement-2026
|
||||
|
||||
**UNVERIFIED list:** disclosure not shown to shoppers; Quality Content Guide wording; 7×50-char keyword slots; review
|
||||
community-guideline verbatim; Amazon Conditions-of-Use anti-scraping verbatim; KDP payout timing verbatim; BSR of the
|
||||
Kinsley, Carruthers and adult-"Whys" titles; niches #8 and #9 (no evidence read); KDP Builder market estimates.
|
||||
Reference in New Issue
Block a user