Dossier Story Excerpts — Proposal (NOT APPLIED)

Builder: Subagent B2 · Content Hardening · completion-run Date: 2026-08-08 Status: Proposal only — approval-gated diff for the owner. No dossier page has been touched with these blocks. Source of copy: # MIZ OKI 3.5/docs/marketing/signal-story-bank.md (v1.1) ONLY — Stories 1, 2, 3, 4, 7. Numbers, metaphors, and beats are the bank's own; nothing invented. All five rules applied: composite/illustrative labels, one ledger grammar (question → quiet test → ledger → change), translate-don't-define, preview stays preview (Story 7), never a guaranteed outcome. Visual idiom: every target dossier already defines the .story card CSS (.q / dl dt dd / .verdict) and the .mark label line in its inline stylesheet — these blocks render with zero CSS changes. All insertions land INSIDE existing §-sections, so the content_qa §-sequence (check D) is unaffected. Each insertion carries its composite/illustrative label in-block, so content_qa's section-label rule (check C) stays green after application; the Story 7 block carries the exact Preview · in development string content_qa's check B requires wherever net yield / net contribution is mentioned.

Line numbers are against the current completion-run working tree (post-B2 edits; only signal-measurement.html was edited by B2, and that edit changed no line numbers).

1 · The Doorman (Story 1) → # MIZ OKI 3.5/signal-audiences.html

Anchor: inside §01 · The four quadrants — insert immediately AFTER line 241 (<p class="lede" style="font-size:.95rem">The quadrants are estimated per segment…</p>), before the closing </div> of .wrap. Selector: section:has(> .wrap > p.sec-mark:contains("§01")) .wrap — append as last children. Rationale: the doorman is §01's "sure things" quadrant told as a customer moment — the quadrant model priced in dollars.

    <p class="mark" style="margin-top:30px">Field note — composite scenario · illustrative numbers</p>
    <div class="story">
      <p class="q">Retargeting is our best channel. The dashboard says 8x. So why does growth feel stuck?</p>
      <dl>
        <dt>The test</dt>
        <dd>Signal quietly kept the ads away from a small, identical slice of shoppers and watched what they did anyway.</dd>
        <dt>The ledger</dt>
        <dd>Most of that slice bought regardless — sure things, not persuadables. The real return was closer to 1.4x.</dd>
        <dt>The change</dt>
        <dd>Retargeting cut by a third; the budget moved to net-new customers. Revenue held.</dd>
      </dl>
      <p class="verdict">You were paying a doorman to open a door people were already walking through.</p>
    </div>

2 · Paying For Your Own Name (Story 2) → # MIZ OKI 3.5/signal-budget.html

Anchor: inside §01 · The gate — insert immediately AFTER line 233 (<p class="lede">Negative uplift is clipped to zero by construction…</p>), before the closing </div> of .wrap. Selector: section:has(> .wrap > p.sec-mark:contains("§01")) .wrap — append as last children. Rationale: brand search is the canonical channel that correlates without causing — exactly the spend §01's gate exists to zero out.

    <p class="mark" style="margin-top:30px">Field note — composite scenario · illustrative numbers</p>
    <div class="story">
      <p class="q">The CFO wants to know why we pay Google every time someone searches our own brand name.</p>
      <dl>
        <dt>The test</dt>
        <dd>Brand-search ads switched off in a handful of matched cities, left on in their statistical twins. Six weeks. No opinions, just registers.</dd>
        <dt>The ledger</dt>
        <dd>Roughly four in five of those clicks slid down to the free listing one inch below — the paid ad caused a sliver, not the number on the report.</dd>
        <dt>The change</dt>
        <dd>Brand-search spend rightsized to what it actually causes. The CFO got a confidence interval instead of an opinion.</dd>
      </dl>
      <p class="verdict">A channel that merely correlates with revenue never clears the gate — this is what that rule looks like in a budget meeting.</p>
    </div>

3 · The Unneeded Coupon (Story 3) → # MIZ OKI 3.5/signal-budget.html

Anchor: inside §03 · Portfolio — insert immediately AFTER line 272 (the </div> closing the §03 .grid.g3 that ends with the "Funnel-position bids" cell), before the closing </div> of .wrap. Selector: section:has(> .wrap > p.sec-mark:contains("§03")) .wrap — append after the .grid.g3. Rationale: §03 prices every channel at its margin, not its average — the coupon story is that pricing applied to the "cheapest" channel.

    <p class="mark" style="margin-top:30px">Field note — composite scenario · illustrative numbers</p>
    <div class="story">
      <p class="q">Our email discounts drive half our orders. But margins keep shrinking and unsubscribes keep climbing.</p>
      <dl>
        <dt>The test</dt>
        <dd>Signal sorted customers by what the send actually changed — not who converted, but who converted because of it.</dd>
        <dt>The ledger</dt>
        <dd>About a fifth of discount sends went to sure-buyers — pure margin given away — and a smaller group was being pushed toward unsubscribe.</dd>
        <dt>The change</dt>
        <dd>Sure-buyers stopped getting coupons for things already in the cart; the annoyed went quiet. Margin recovered without one extra send.</dd>
      </dl>
      <p class="verdict">Cheap to send, expensive in margin — priced at its margin instead of its average, email changed shape.</p>
    </div>

4 · Prove It Or Lose It (Story 4) → # MIZ OKI 3.5/signal-measurement.html

Anchor: inside §03 · The incrementality engine — insert immediately AFTER line 266 (<p class="lede">Incremental return — not platform-reported return…</p>) and BEFORE the Stated plainly note (line 267), which stays the section's closing word. Inserted together with Story 7 below as ONE block (shared label mark, one .grid.g2). Selector: section:has(> .wrap > p.sec-mark:contains("§03")) .wrap — insert before p.note. Rationale: the story is §03's machinery end-to-end — registered geo design, lift with intervals, and a budget decision made from receipts.

5 · The Bundle That Lost Money (Story 7 — PREVIEW) → # MIZ OKI 3.5/signal-measurement.html

Anchor: same block as Story 4 (second card in the same .grid.g2). Rationale: the bundle ledger is what the measurement page's "one ledger" means for a Shopify operator — and the net-yield bidding step is told future-tense with the preview tag, per the claim ledger.

Combined block for anchors 4 + 5:

    <p class="mark" style="margin-top:30px">Field notes — composite scenarios · illustrative numbers</p>
    <div class="grid g2">
      <div class="story">
        <p class="q">Streaming TV is our biggest bet and our least measurable one. Finance wants to cut it. Prove it works — or lose it.</p>
        <dl>
          <dt>The test</dt>
          <dd>The campaign ran in some markets and deliberately not in their statistical twins — the design registered before the first impression.</dd>
          <dt>The ledger</dt>
          <dd>Real lift in nine of fourteen markets, nothing defensible in the other five — each with its interval printed.</dd>
          <dt>The change</dt>
          <dd>Doubled down on nine, cut five. The budget came out smaller and unkillable, because every remaining dollar had a receipt.</dd>
        </dl>
        <p class="verdict">"Prove it or lose it" stopped being a threat and became the operating procedure.</p>
      </div>
      <div class="story">
        <p class="mark">Net yield · Preview · in development</p>
        <p class="q">Our $180 bundle is the best seller. Why is cash down while revenue is up?</p>
        <dl>
          <dt>The test</dt>
          <dd>Every order priced at what it truly nets — components, packing, dimensional shipping, payment fees, and the returns coming back.</dd>
          <dt>The ledger</dt>
          <dd>The hero bundle netted between −$2 and $16 an order once a third of them boomeranged. The boring single SKU netted $31 and almost never returned.</dd>
          <dt>The change</dt>
          <dd>When the net-yield capability ships, the bidding signal switches from checkout revenue to net contribution — so the platforms hunt buyers who keep what they buy.</dd>
        </dl>
        <p class="verdict">The dashboard applauded the bundle; the bank account preferred the boring one.</p>
      </div>
    </div>

Post-application checklist (for whoever applies this diff)

1. Owner approval recorded for the specific insertions (canon-adjacent surfaces; signal dossiers are not canon-pinned, but website-governance still applies).
2. python3 scripts/content_qa.py — must stay green (checks B/C/D reasoned through above; Story 7's block carries the Preview · in development string).
3. python3 -m unittest tests.test_content_qa tests.test_signal_capability_site from the site dir.
4. Line-number anchors re-verified at apply time if other diffs land first (the selectors above are the durable anchors).