Lead Utility Master Plan

1 The model shift

• Old model: sell a contractor first, then generate their leads. New model: own the demand first — generate B2C leads in a market, then sell results to a contractor at 10% of attributed revenue, SP covers everything up front.
• The sales call changes from "buy my service" to "I already have qualified homeowners in your city — want them?" A trust close, not a pitch close.
• The unit of the business is a cell = one niche × one market (e.g. CL × Dallas). ~10 niches × ~50 valuable markets ≈ 500 cells.
• The sellable asset = 500 revenue-producing cells + a 500-contractor roster + the attribution/billing rail that proves the 10%.

2 What already exists — the four tracks

• ABM / Dream-900. The 1,004-account tracker (5,583-account universe, 9 niches) becomes the buyer-side map: which contractor gets offered each cell's leads. JIT enrichment waves (25–50/wk, enriched the Thursday before dialing) = the fill-the-cell motion. Surround-sound 1-mile pins warm the exact contractor before the "I have leads" call.
• Market-by-market rollout. Wedge proven: CL Dallas did $184.53 → 10 qualified ≈ $18/QL vs the $150 ceiling (best week ever, Jul 20–26). Top-10 CL market scoring done Jul-27. The WS4 preview-lead pilot (7 TX/OK metros, $10/day, +10mi residents-only) was this model in embryo. Geo standard stands: every market +10mi.
• Outbound. Field guide, 7:30 call sheet + cockpit, demo-bot rail, show-rate playbook. Under the new model outbound stops selling a retainer and starts giving away proof: preview leads as the opener (3 leads + the market stat line). Dream-900 accounts in live metros = the dial list; zero-lead metros deprioritized.
• Economics. Tickets: CL $3K · CF $2K · CS $10K · FR $15K; CPL target = 5% of ticket. One closed job pays SP 10%: CL $300 · CS $1,000 · FR $1,500. At a 25% close on qualified leads, expected revenue per QL (CL $75 / CS $250 / FR $375) clears QL cost ($18–50) by 3–8x. Margin survives down to ~10–15% close.

3 The physics — constraints that shape the plan

• Page factory: ~2 FB Pages/day, serialized. 500 market pages ≈ 9 months of minting. Start day 1, never stop, mint in ranked-cell order. Interim: 10 niche-brand pages let ads run before a market's page exists. Per-Page ad-volume limits favor per-market pages long-term.
• Campaign architecture: campaign = niche (10 CBO campaigns), ad set = market (~50 per campaign). CBO allocates budget to the best markets inside each niche; cross-niche allocation is Akash's weekly call.
• Cash: 500 cells × $10/day ≈ $150K/mo fronted by SP at full scale. Every phase gets a hard budget; runway = THE number; kill criteria per cell.
• Billing: 10% of attributed revenue is uncollectable without job-level attribution + weekly reporting + invoicing. Click/lead attribution exists; the revenue loop does not. #1 build.
• Account state today: both ad accounts dark, unknown-pauser unresolved, B2B write block (1815066) until the system user is assigned, WP-1 lead contamination (privacy) fixed first.

4 Gantt — five phases

Phase 0 — Stabilize + contract (Week 0: Jul 28–Aug 3)

• 1WP-1: quarantine contaminated leads + ingest guard (privacy — do first)
• 2Answer who paused CL Dallas; write single-pause-authority rule; relight CL Dallas (Akash/Sarah gate)
• 3Rewire 06:10 Meta sync token so dashboards read live
• 410%-attributed-revenue contract v1 (one page) + weekly closed-jobs reporting cadence; sign Dan as contract #1
• 5Page factory rail v1: naming standard, serialized 2/day cadence starts
• 6Extend CL market scoring to all 10 niches → ranked 500-cell list; freeze the Top-100

Phase 1 — Wedge proof: CL × 10 markets (Weeks 1–4)

• 1Stand up 10 CL metro cells (WS4 layout): consumer page + form, $10/day ad sets, +10mi residents-only
• 2Transfer rail per market incl. a "lead bank" for metros with no contractor yet; 24–48h delivery SLA, zero ghosted homeowners
• 3B2B ads flip to market-level threshold copy, geo-fenced to live metros only + surround-sound pins
• 4Outbound waves dial Dream-900 CL accounts in live metros, armed with preview leads
• 5Attribution/billing v1 live: weekly closed-jobs report → first 10% invoice collected
• Exit criteria: ≥5 contractors signed · ≥1 invoice paid · CPL ≤ ceiling in ≥7/10 metros

Phase 2 — Top-100 cells (Weeks 5–12)

• 1Page factory → ~100 highest-ranked cells live by ~Week 10 (the 2/day cap sets this slope)
• 2All 10 niche CBO campaigns stood up as each niche's first markets go live
• 3Roster 5 → 60+: outbound + B2B ads fill cells; onboarding wizard; Collins engines (still gated on explicit go)
• 4Automate: n8n lead routing, weekly reports, invoicing; per-cell dashboard cards
• Exit criteria: 100 cells live · ≥60 filled · attributed-rev run-rate $25–50K/mo · collections >90%

Phase 3 — Full 500 + roster 250 (Months 4–9)

• 1Cells 100 → 500 in ranked order; niche-brand pages cover the tail until market pages mint
• 2Roster 60 → 250; setters recruited (Akash owns recruiting); CSM function stood up
• 3Per-cell P&L + kill criteria enforced weekly; zero-lead metros culled, budget recycled by CBO
• 4Retention systems: lead-quality scoring, contractor scorecards, churn <10%/mo

Phase 4 — Roster 500 + exit (Months 10–18)

• 1Roster 250 → 500 (≈1 contractor per cell; double-fill only the biggest metros)
• 212 months of clean attributed-revenue financials; contracts standardized on paper + ACH (handshake deals depress the multiple)
• 3Data room: per-cell P&L, cohort retention, the automation stack as the story
• 4Broker/buyer process; sell the book

5 Weekly operating cadence

• Daily: 2 pages minted · lead SLA check · dial block in live metros
• Thursday: enrich next week's outbound wave (JIT — never mass-enrich)
• Friday: cell review — CPL vs ceiling, fill status, kill/scale calls; scorecard predictions logged
• Weekly: closed-jobs reports in → invoices out → collections tracked

6 Three functions — human vs automated

• Generate pipeline — fully automated. Homeowner ads prove the market · B2B ads + surround-sound pins warm its contractors · Thursday JIT enrichment · pre-call SMS/email sequences · preview-lead packet built per dial.
• Close deals — ONE human step: the sales call (Akash). Everything around it is automated: 7:30 call sheet + pre-call packet, confirmations, no-show chases, post-call follow-up, contract e-sign + ACH on file, onboarding.
• Generate jobs — fully automated, the only money maker. Ads per cell (CBO) · bot qualifies every lead · transfer rail delivers in 24–48h · nurture books the inspection · job closes → revenue attributed → 10% invoiced + collected by ACH (attribution + billing = the two pieces still to build).
• The loop: homeowner lead → contractor closes a $3K job → $300 to SP → funds the next market's ads.

7 Niche-of-the-month season calendar

Principle: launch a niche the month before its demand peak, ride the peak, never launch into a slow season. Peak windows are informed estimates — validate each with Google Trends before its build month.

8 CL page-creation calendar — Jul 28 to Aug 31

2 pages/day, serialized, never parallel. 71 targetable CL markets live in brain.niche_market_fit; Dallas, San Antonio, Austin, Houston already made = 67 to mint. At 2/day the entire CL market list is done by Aug 31. Order = TX/OK wedge first, then clay-soil / freeze-thaw big metros, then the tail.

9 Monthly niche deployment standard — the 4-week template

Repeatable playbook for every niche month (CL Aug, FR Sep, Windows Oct, …). W0 runs during the prior month's week 4, so niches chain without a gap.

10 Risks + kill criteria

• Collection risk on trust-based 10%: weekly reporting ritual + ACH/card on file from client #6 onward
• Meta account health: two unexplained pause events + past compromised-account flag → single pause authority, ledgered writes, preflight on every launch
• Cash front-loading: phase budgets are hard gates; a phase that misses exit criteria pauses spend, not hope
• Ops load at 500 clients: 500 weekly reports/invoices is an automation problem from day 1, not a hiring problem later
• Cell kill rule: a cell that misses CPL ceiling or gets zero leads for 2 weeks is paused, budget recycled — loss ≠ ban, revisit quarterly