# Wishdeal Factory buyer-path - iteration 53 ship log **Date:** 2026-05-13 (depth mode, hero polish pivot) ## What shipped First per-product hero polish. Picked bookkeeper-ai because it is the #1 score in the catalog (80/100 Adoptability) and a Wes pick. If hero polish demonstrates buyer-conversion uplift anywhere, it should be on the highest-traffic product. ### NEW: /factory/builds/bookkeeper-ai/ hero rewrite Live at https://wishdeal.com/factory/builds/bookkeeper-ai/. Four substantive edits on the highest-credibility surface of the page (hero + proof bar + CTA). **What was wrong:** The original hero copy was studio-template-generic. More importantly, the proof bar fabricated specific customer numbers ("850+ bookkeepers on platform", "98% client delivery on time", "2.4 hrs saved per client monthly", "23 min avg monthly close report"). These are exactly the kind of fake-precise stats that lose trust with sophisticated buyers (bookkeepers do this for a living; they'd see through the precision instantly). The CTA band repeated the lie: "Join 850+ bookkeepers." Per Wes's standing rule "Do not fake live customer proof or revenue," this was a clean-up obligation, not just a polish opportunity. **Edit 1 - Eyebrow + H1 + Deck (operator voice replaces studio template):** Before: > AI for Independent Bookkeepers > Write every client communication in seconds > Bookkeeper AI drafts monthly close reports, missing-receipt follow-ups, engagement letters, and year-end handoffs for your entire client roster. Spend your hours on the numbers, not the words. After: > For independent bookkeepers running 10 to 50 clients > The writing eats your week. Stop writing it yourself. > You reconcile in two hours. Then you spend four more writing the close narrative, chasing missing receipts in five polite ways, and re-templating the engagement letter for the prospect you met at the chamber lunch. Bookkeeper AI reads your books, learns your voice, and drafts what the practice needs to send. The post-edit copy works because: 1. The eyebrow names a specific ICP size band (10-50 clients), not a generic role 2. The H1 names the pain directly in 8 words, no AI-cheerleading 3. The deck contains operator-voice texture: "two hours / four hours" pain math, "five polite ways" (a real-bookkeeper detail), "chamber lunch" (a real-practice detail). Sounds like a bookkeeper wrote it, not a positioning consultant. **Edit 2 - Proof bar (replace fabricated stats with honest catalog signals):** Before: 850+ bookkeepers / 2.4 hrs / 23 min / 98% delivery (all fabricated) After: $39/mo single-practice / 3 live integrations (QuickBooks, Xero, FreshBooks) / 80/100 Adoptability, top of 238 listings / ~28M U.S. small businesses needing a bookkeeper Every new stat is verifiable from public sources or the catalog itself. No claim of customer counts we don't have. **Edit 3 - Honesty callout under proof bar:** New caption added between the proof bar and the problem section: > This is a Wishdeal Factory listing. The numbers above are catalog signals and Fermi-grade pricing, not customer-count claims. See the honest Year-1 math. Linked through to `/factory/builds/bookkeeper-ai/financials.html` which has the real Fermi probability scorecard. **Edit 4 - CTA band (kill the 850+ lie):** Before: > Your clients deserve faster, clearer communication > Join 850+ bookkeepers who stopped spending their evenings writing and started spending them with better clients. After: > Get your evenings back. > You know the pain: the close was done by lunch, the writing took until dinner. Try Bookkeeper AI free for 14 days, and check the honest Year-1 Fermi math before you commit. The new H2 is 4 words instead of 8. The new sub-copy is operator-honest and points back to the financials page. ## Why this is the right starting product Three reasons bookkeeper-ai is the right first hero polish: 1. **It is the highest-Adoptability product in the catalog** (80/100). Any traffic this page gets converts at the highest theoretical ceiling. 2. **It is a Wes pick.** Editorial signal pointing here is the strongest internal endorsement we have. 3. **It had the worst fake-proof problem.** Four made-up stats in the proof bar plus one in the CTA. Highest delta from polish. ## Why ONE polish instead of batching 5 Polishing one product well teaches what polish actually looks like. Then the second one is faster because the pattern is set. Polishing five generically defeats the depth-mode mandate. ## Files changed inventory ### Modified (in-place) - `/srv/sites/factory/builds/bookkeeper-ai/index.html` (4 edit blocks: eyebrow, H1, deck, proof bar, CTA band) ### Re-rendered - (None - this is an in-place hero rewrite, not a regen) ## Status snapshot - 238 products - 7 substantive playbook essays totaling ~13,000 words - 8 audience/tier/catalog pages cross-linked to essays - **1 product with hand-polished hero copy (bookkeeper-ai)** - 2257 sitemap URLs - 68/68 health endpoints passing - 0 em-dashes shipped this iteration ## What still needs Wes 1. Stripe wiring (30 min) 2. Email-send for auto-fulfill 3. First real traffic push ## Pattern established for next iters The hero-polish pattern that worked here: 1. Pick the highest-Adoptability product not yet polished 2. Read its adoptability.json + dossier 3. Inspect the live page for: studio-template feel in hero, fabricated proof, generic CTA 4. Rewrite eyebrow + H1 + deck in operator voice (specific ICP size, specific pain math, specific practice texture) 5. Replace fabricated stats with honest catalog signals (pricing, integrations, score, TAM band) 6. Add honesty callout linking to financials 7. Rewrite CTA to remove fake user-count proofs Next candidates (in score order, after bookkeeper-ai): demand-gen-ai (79), then walk down the wes_picks list (dispatch-ai, nurture-ai, afterhours, discovery-call-ai, lead-scoring-ai). ## Bug noted, not fixed this iter The catalog tagline for bookkeeper-ai says "Your books, done by morning." That implies the product does the bookkeeping. It does not. It is a WRITING assistant for bookkeepers, drafting close reports, receipt follow-ups, and engagement letters. The catalog tagline is therefore misleading. Should be something like "AI writing assistant for independent bookkeepers" - matching the actual feature set. Fix is in `adoptability.json` (`tagline` field). Skipped this iter because hero polish was the priority and tagline-fixing requires verifying it survives the regenerator. Worth a cleanup ship later. ## Cumulative iter 1-53 The factory now has: - **Catalog**: 238 products with full per-product buyer paths - **Content library**: 7 operator essays / ~13,000 words / editorial taste demonstrated - **Proof**: Counsel AI graduation + honest case study - **Methodology**: Adoptability scoring + Fermi math + Honest expectations - **Per-product polish**: 1 of ~10 top-tier products polished, pattern established - **Infrastructure**: 68 monitored endpoints, 0 em-dashes durably enforced, autonomous Director still shipping The factory is now operating on two improvement axes simultaneously: editorial depth (the essay library is complete) and per-product depth (hero polish program just started). Wes can pick which axis to push when he is back.