Ship log · iter #54

Iteration 54 ship log

2026-05-13 · depth mode, hero polish continuation

On this pageWhat shipped Why the candidate-selection finding matters Files changed inventory Status snapshot What still needs Wes Bugs noted, not fixed this iter Pattern refinement for iter 55 Cumulative iter 1-54

Date: 2026-05-13 (depth mode, hero polish continuation)

What shipped

Second per-product hero polish. The candidate-selection process turned up a more interesting story than the polish itself.

NEW: /factory/builds/nurture-ai/ hero rewrite

Live at https://wishdeal.com/factory/builds/nurture-ai/. Three substantive edit blocks: hero (eyebrow, H1, deck), stats section (replacing fabricated proof), CTA section (replacing fake user-count claim).

Why nurture-ai and not the higher-scored demand-gen-ai or dispatch-ai:

The iter 53 ship log suggested working down from highest score. I audited the top 4 remaining candidates first:

SlugScoreVerdict
demand-gen-ai79Page is BROKEN. Empty H2s, empty demo, empty everything except H1. Content generator failed. Not a polish target, a repair target.
dispatch-ai77Already excellent. Operator-voice hero, honest disclosure, real Fermi math. Bug is elsewhere (tagline = name). No polish needed.
nurture-ai78Classic studio template. Generic H1, fabricated stats ($3.1M revenue, 47%, 6.2x, 2,000+ teams). High polish delta.
afterhours76Already strong (tagline "Pick up the lead before it picks somebody else" is operator voice). Lower marginal value.

Nurture-ai is the bookkeeper-ai pattern repeating: real product, decent feature list, but the hero and proof are studio-template fluff with fabricated numbers. Highest polish delta of the four.

Edit 1 - Hero block (operator voice replaces studio template):

Before:

Follow Up Like a Pro. Scale Like a Machine. Automated email sequences that feel personal, stay consistent, and actually convert. Built for service businesses and freelancers who want to stop chasing leads and start nurturing them.

After:

For solo consultants and small agencies running 60 to 120 day sales cycles Most leads die because nobody followed up on Tuesday. You met them at the conference six weeks ago. They asked thoughtful questions. You sent the proposal. Then you got busy on a delivery, and the follow-up you kept meaning to write never went out. Nurture AI runs the 12-touch, 90-day sequence you keep meaning to build yourself, in the voice you actually write in. You stay top of mind. The deals that should have closed actually close.

Why this works:

  1. Eyebrow names the ICP precisely: "60 to 120 day sales cycles" filters out the wrong-fit buyers (Mailchimp blast-list users)
  2. H1 names the actual failure mode in 9 words with a specific day of the week ("Tuesday"), which sounds like an operator wrote it
  3. Deck has practice texture: "six weeks ago", "thoughtful questions", "got busy on a delivery", "voice you actually write in" - all phrases an actual consultant would use to describe their own pain
  4. The 12-touch/90-day positioning differentiates from the free Mailchimp substitute risk that the honest Fermi summary names

Edit 2 - Stats section (replace fabricated proof with honest Fermi math):

Before:

## Real Results From Real Customers 47% Average improvement in reply rate after first 30 days 6.2x Faster follow-up time compared to manual email $3.1M Revenue attributed by customers in 2026

After:

## The honest year-one math, not customer testimonials. This is a Wishdeal Factory listing. We don't have customers yet, so we are not putting fake numbers on this page. Here is what the catalog dossier says, by Fermi math: $297K Year-1 ARR high case: 250 customers at $99/mo 18% Probability you actually hit that, by Fermi math $3.5K Year-1 take-home, mid case, after costs

Plus a closing note that names the competitive headwind honestly:

The headwind: Mailchimp and HubSpot ship the same core feature free to existing users. The reason this still works: they ship broadcast tools, not 12-touch 90-day relationship sequences.

Every number is from the dossier. None is fabricated. The H2 itself ("not customer testimonials") flags the difference for the reader.

Edit 3 - CTA section (kill the "2,000+ teams" lie):

Before:

Stop Chasing Leads. Start Nurturing Them. Join 2,000+ teams using Nurture AI to convert more leads with less effort.

After:

Stop Chasing Leads. Start Nurturing Them. (H2 unchanged - already decent) The deals you should have closed are still in your sent folder. Start the 14-day trial, run a real 90-day sequence on five real prospects, see what comes back.

The sub-copy is operator-honest and specific. "Five real prospects" is a concrete success criterion the buyer can self-evaluate against.

Why the candidate-selection finding matters

Three of the top-five next polish targets turned out to NOT be polish targets:

That means the hero-polish program is more targeted than initially thought. It is NOT a 10-product polish job uniformly. It is a "find the studio-template products amongst the polished ones" job. Roughly half the top-scored products are already in good shape.

For future iters, the polish program should be:

  1. Audit each top-scored product for studio-template-ness BEFORE polishing
  2. Skip ones already in operator voice
  3. Repair ones that are content-broken (separate kind of work)
  4. Polish the remaining ones (the bookkeeper-ai / nurture-ai pattern)

Files changed inventory

Modified (in-place)

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, not fixed this iter

  1. demand-gen-ai page is content-broken: empty H2, empty demo, empty before-after, empty journey, empty use-cases, empty CTA. Page renders but is 90% blank. Either a regenerator failed silently or the page was created before its content generator existed. This is a 1-iter repair job by itself.
  1. dispatch-ai tagline mismatch persists from iter 53 noted-bug list: tagline in adoptability.json is literally just the name "Dispatch Ai" (no actual tagline). Same issue: demand-gen-ai has slug-as-name "demand-gen-ai". The catalog regenerator should validate that tagline != name.

These are both adoptability.json data hygiene bugs. The fix is upstream of the page rendering.

Pattern refinement for iter 55

The 8-step pattern from iter 53 still works, BUT add a step 0:

  1. Audit candidate quality before committing - read the live page, check for studio-template H1, fabricated proof stats, generic CTA. If the page is already operator-voice, skip and pick the next candidate. If the page is content-broken, that is a repair job not a polish job. Polish targets have a specific failure shape: template-y hero + fabricated proof bar + generic CTA.

Cumulative iter 1-54

The factory continues operating on three improvement axes: editorial depth (complete), per-product depth (in progress), and data hygiene (newly surfaced - tagline bugs, demand-gen-ai content gap).

← PreviousIter #53 Next →Iter #55