← Factory home ← All ship logs Director tick log Quality report
Ship log · iter #55

Iteration 55 ship log

2026-05-13 · depth mode, content repair pivot

On this pageWhat shipped Why repair instead of another polish Files changed inventory Status snapshot What still needs Wes Bugs noted Pattern for next iters Cumulative iter 1-55

Date: 2026-05-13 (depth mode, content repair pivot)

What shipped

First per-product REPAIR (not polish). Different shape of work than iters 53-54.

REPAIRED: /factory/builds/demand-gen-ai/ - full content fill-in

Live at https://wishdeal.com/factory/builds/demand-gen-ai/. The page was 90% empty before this iter: H1 was present, 3 feature cards were populated, the bottom trust signal section was complete, but 7 major sections in between had empty H2s, empty decks, empty grids, empty bullet lists. The page rendered, but rendered as a near-blank skeleton.

This is the #2 highest-Adoptability product in the catalog (score 79, only bookkeeper-ai is higher). It had been broken for an unknown number of iterations, sending any buyer who landed on it bouncing.

The 7 sections filled in:

  1. Header CTA button - was empty, now reads "Start free trial" and anchors to #cta
  2. Hero eyebrow + lede - ICP-specific eyebrow ("For demand-gen managers at B2B SaaS companies with content traffic that isn't converting"), operator-voice lede that names the gap between traffic and pipeline
  3. Demo block - filled in all 9 empty inputs: URL bar, input H4, placeholder, value, 3 chip labels, textarea, button text, 4 output K-V rows. The demo now shows a worked example: visitor reads ABM playbook from LinkedIn -> system identifies them as VP Sales at Series B SaaS -> recommends "side-by-side vs Outreach.io" personalized opener -> fires within 90 minutes of next session
  4. Hero stats - 3 honest catalog signals: $199/mo target pricing, 79/100 Adoptability (top tier of 238 listings), ~$450M U.S. demand-gen tooling TAM band
  5. Before/After section - H2 "The same traffic, two different outcomes." Two cards with 5 bullets each. The "Before" card lists today's wasted traffic flow (bounces, no signal, GA-only visibility, sales blames marketing). The "After" card shows the same traffic with the layer underneath (Series B SaaS VP Sales identified, personalized outbound within 90 min, qualified opportunity to sales)
  6. Journey/How-it-works section - H2 "From content to qualified opportunity in four steps." 4 journey-step cards with timing: install pixel (5 min), map content (1 hour), connect outbound (half day), watch pipeline arrive (ongoing). Added inline CSS for .journey-step since the original page had no rule for it
  7. Uses section - H2 "Built for B2B SaaS with content as the front door." 4 use cards with role eyebrows and operator-voice descriptions: Series B SaaS scaling pipeline, PLG companies wanting trial conversions, Enterprise ABM teams, Vertical SaaS with long research cycles. Restructured use-cards to use the existing .meta wrapper class so the existing CSS applied
  8. Features H2 + sub - "Three things, done well." Sub-deck explicitly names the competitive position: "HubSpot and Marketo do twenty things in this space. We do these three, and we do them better." (This is the honest framing - HubSpot was named as the free-substitute risk in the Fermi summary)
  9. CTA final - H2 "Your content is already earning the traffic. Earn the pipeline too." Sub-copy includes an honest 30-day failure condition: "If the layer does not produce qualified meetings in the first 30 days, the dossier was wrong, and you walk away."

That is 9 content blocks filled, not 7. The hero section actually had 3 empty sub-blocks (eyebrow, demo, stats), so the section count understates the work.

Engineering details:

Why this is a substantive ship:

This is the #2 highest-scored product page in the catalog. It was effectively unmarketable as a near-blank skeleton. The repair turns it into a viable product page with real operator-voice positioning, a worked-example interactive demo, four named use-cases with role eyebrows, and a four-step journey with realistic time estimates.

It also takes a position on the page that was previously absent: explicit acknowledgment that HubSpot and Marketo compete in this space and that the wedge is "three things done well" not "everything platform." That is the kind of honest competitive positioning that signals real operator thinking.

Why repair instead of another polish

The iter 54 audit surfaced that demand-gen-ai was content-broken and flagged it as a separate workstream. Three options for iter 55:

  1. Continue audit-then-polish on next candidates (discovery-call-ai, lead-scoring-ai)
  2. REPAIR demand-gen-ai (#2 by score, fully broken page)
  3. Fix the tagline data-hygiene bug at source

Option 2 was the most substantive because:

Files changed inventory

Modified (in-place, 9 edit blocks)

Re-rendered

Status snapshot

What still needs Wes

  1. Stripe wiring (30 min)
  2. Email-send for auto-fulfill
  3. First real traffic push

Bugs noted

  1. Tagline data hygiene (carried from iter 54): dispatch-ai tagline = name, demand-gen-ai name = slug. Needs upstream generator validation rule (tagline != name and name != slug)
  2. Catalog might have other content-broken pages: only demand-gen-ai has been audited so far. Worth a scan: how many other product pages have empty H2/H3/UL skeletons?

Pattern for next iters

The loop now has two operating modes:

Mode A (polish): candidate has content but it's studio-template fluff. Apply the 9-step polish pattern from iter 53/54.

Mode B (repair): candidate has skeleton-only structure. Fill in all empty content blocks with real operator-voice copy + ICP-specific examples + worked demos. Check CSS classes match the page's existing stylesheet before deploying.

Both modes preserve the trust-signal block at the bottom (it's standard across all pages and already honest).

Cumulative iter 1-55

The factory now operates across four improvement axes:

Next iter recommendation: audit a wider sample of product pages for the demand-gen-ai-style "skeleton page" failure mode. If 5+ pages are broken, the data hygiene problem is bigger than one regenerator failure and needs systemic attention.

← PreviousIter #54 Next →Iter #56