# [SYSTEM_NAME] — Complete Playbook & Service Specification

> The full venture in one document: **Part 1 Strategy → Part 2 The Vertical → Part 3 The Economics → Part 4 The Service Specification.**
>
> **Placeholders** (swap once, globally): `[AGENCY_NAME]` = *Summit* (example) · `[SYSTEM_NAME]` = *The HVAC Revenue Engine™*. Pricing, timelines, and industry stats are marked **(confirm)** — verify before anything goes client-facing.

## Contents
- **Part 1 — Strategy:** the thesis, why features aren't a moat, the real moats, the fresh capability layers, the technical moat, the sales mechanics.
- **Part 2 — The Vertical:** HVAC, the fit lens, the ranked shortlist, the ROI math, the saturation objection, the validation playbook.
- **Part 3 — The Economics:** the mid-ticket band, build-once-redeploy, unit economics, packaging & pricing, scale math, margin traps.
- **Part 4 — The Service Specification:** the full operational definition — every module, architecture, agents, compliance, delivery, ops, KPIs, guarantee, pitch.

---
---

# PART 1 — STRATEGY

## 1.1 The thesis
Out-position a saturated market not by adding more features, but by becoming a different *category*: a vertical-specialized, outcome-selling **growth engine** with a named mechanism, brutal transparency, a guarantee, and an AI-search capability competitors aren't selling — built on a technical/automation moat you *own* rather than rent.

## 1.2 The competitive reality (why features ≠ moat)
Nearly everything a typical agency lists is now **native to GoHighLevel**: the AI Employee suite (Voice AI, Conversation AI, Reviews AI, Content AI, Funnel AI), near-human voice latency, multi-language, Agent Studio. Every white-label reseller is one toggle away from matching any feature. **A moat made of GHL toggles isn't a moat.** Differentiation must come from things that are structurally hard to copy — and from capability layers most haven't adopted yet.

## 1.3 The three layers of differentiation
1. **Table stakes** (everyone has or can flip on): the GHL feature set. Include it, don't lead with it.
2. **Fresh capability layers** (a 12–18 month first-mover window): AI-search visibility, deep agent configuration, compliance, true lifecycle automation, predictive intelligence.
3. **Structural moats** (genuinely hard to copy): vertical depth, a proprietary named mechanism, proof/transparency, risk reversal, brand. **Spend your energy here.**

## 1.4 The structural moats
- **Own one vertical, brutally deep.** "Marketing for local businesses" is what 10,000 resellers say. "The growth system for residential HVAC" is a different universe — niche the snapshot, copy, agents, objection handling, and case studies. A generalist cannot fake vertical knowledge on a sales call.
- **A named proprietary mechanism.** Wrap the stack into [SYSTEM_NAME] with named stages (Answer → Recover → Get Found → Prove). In a sophisticated market you win by introducing a *new mechanism* that explains why yours works when the last agency didn't. Costs nothing but thinking; highest-leverage move on this list.
- **Radical proof & transparency — weaponized.** ~67% of businesses are dissatisfied with their agency within 12 months; top complaints are lack of transparency (~58%) and failure to deliver (~48%); most cycle through 3–5 agencies. **(confirm.)** Make a live, real-time revenue dashboard the *product*. "You'll see every dollar this generates, live — and if you don't, you don't pay" disqualifies 90% of competitors in one sentence.
- **Risk reversal / performance pricing.** In a market full of burn victims, the party that takes the risk wins: a "live in X days or it's free" setup guarantee; first-N-bookings-free trial; or base + performance. Even a partial guarantee makes you the safe choice.
- **Sell the outcome, not the software.** Reframe from "I'll give you tools" to "I'll give you X more booked jobs." The platform becomes the invisible engine, not the pitch.

## 1.5 The fresh capability layers (start here for the "this is different" reaction)
- **AI-search visibility (GEO/AEO) — the one almost nobody local is selling.** Buyers increasingly ask ChatGPT/Gemini/Perplexity and Google AI Overviews "best [trade] near me." ~72% of SEO-investing brands get *zero* AI citations, Google rank barely predicts AI visibility, AI Overviews appear in ~30–40% of queries, and Gartner projects ~25% decline in traditional search by 2026. **(confirm.)** Offer entity/schema work, `llms.txt`, answer-first content, citation building, and a dashboard tracking whether they get named in AI answers. Lead the free audit with it.
- **The AI front desk done *deep*** — not toggled-on. Train on the client's real top-20 objections and actual call recordings; wire into live booking; smart human escalation; weekly "jobs booked by AI" report. Same feature, 10× the result.
- **Compliance-as-a-feature.** Auto-texting/DBR sit on TCPA (violations $500–$1,500 each) and emerging AI-disclosure laws. **(confirm.)** Bake in consent, opt-out, quiet-hours, and AI identification — and *sell it as protection.* Turns the thing that gets sloppy agencies sued into a reason to choose you.
- **Full-lifecycle autonomous follow-up** (instant speed-to-lead → qualify → book → reminders → review → reactivation → win-back) and **database intelligence** (lead scoring, repeat-due flags, best-dormant-customers to win back).

## 1.6 The technical / automation moat (what you OWN)
The moat lives in what you own, not what you rent. Layered architecture (full detail in §4.5): **GHL** (front office) → **self-hosted n8n** (orchestration / your IP) → **Claude/GPT via API** (agent reasoning) + cheap/local models for cheap tasks → **vector DB / RAG** (per-client knowledge) → **programmatic SEO/GEO engine** → **live dashboard** (proof). The orchestration + custom agents + knowledge layers are bespoke and portable — not a GHL toggle.
- **Automated SEO:** programmatic `[service] × [city]` pages with injected `LocalBusiness`/`Service`/`Review` schema, IndexNow for fast indexing, GBP autopilot, AI-visibility tracking, technical-SEO watchdog.
- **Automation backbone:** self-hosted n8n over Zapier — no per-task fees (protects margin at scale), full data control, white-labelable, **you own the workflows as reusable IP**, runs any model, MCP-capable. Also the "connective tissue" that wires GHL into their existing tools.
- **Custom AI agents beyond voice/text:** see the catalog in §5.12 — customer-facing (omni-channel front desk, quoting, RAG knowledge, booking, win-back) and owner-facing (the "ask your business" agent, job-to-marketing, competitor-watch, report-writer, collections).

## 1.7 The "holy shit" sales mechanics
- **The audit (wedge):** an automated **AI Visibility + Missed-Revenue Audit** — show them they're invisible in ChatGPT, quantify dollars lost to missed calls + dead leads, screenshot a competitor getting recommended by AI. The audit *is* the pitch.
- **The live demo:** spin up an AI front desk trained on *their* business and let them call/chat it on the spot.
- **The reverse demo:** call their line after hours (voicemail = lost job) → ask ChatGPT "best [trade] in [town]" (they're absent) → show the fix.
- **The number:** missed-call + dead-lead + AI-invisibility loss = a dollar figure. People buy to stop bleeding.

## 1.8 What makes it uncopyable
The **combination**: a generalist can't fake the niche knowledge, can't copy your proof history, won't take the risk, and hasn't heard of the AI-search angle. **Decision filter for every idea: "Can a competitor copy this by 5pm tomorrow?"** If yes → table stakes. If no → that's where you invest.

---
---

# PART 2 — THE VERTICAL

## 2.1 The decision
**Residential HVAC** — with **plumbing/electrical** as near-identical cousins and **med spa** as the strong plan B. HVAC is the only vertical that scores high on *every* axis at once, and its single biggest pain (the unanswered call) is exactly what the stack annihilates.

## 2.2 The fit lens (what makes a vertical perfect for this stack)
1. **High ticket or high LTV** — one recovered job/client pays for months; ROI is self-evident.
2. **Phone-driven + speed-to-lead sensitive** — customers call; a missed call is a job handed to a competitor.
3. **High inquiry volume** — enough flow that the system is always earning.
4. **A database to reactivate** — past customers or dead quotes = free DBR fuel.
5. **Already spending on marketing** — your fee is a reallocation, and they're primed to buy ROI.

## 2.3 The shortlist (with 2026 numbers — confirm)

| Vertical | Typical job | Volume | Repeat / reactivation | Missed-call ROI | Affordable | Verdict |
|---|---|---|---|---|---|---|
| **HVAC** ✅ | $340 call → $5k–15k install | Strong | Strong | Strong | Strong | **Launch here** |
| Plumbing / electrical | $150 call → $1.5k–5k | Strong | Medium | Strong | Strong | Near-identical; most visceral demo |
| Roofing | ~$10k ($7.5k–30k+) | Medium | Weak* | Strong | Strong | One job = a year of fees; cyclical, low repeat |
| Med spa | $450–700/visit · ~$9k LTV | Medium | Strong | Strong | Strong | High LTV; light medical compliance |

\*Roofing has strong dead-quote DBR but low repeat.

- **HVAC depth:** ~$182B U.S. market, ~423k techs; ~$340 blended revenue per booked call; repairs ~$1,200; installs $5k–$15k (the majority of revenue); 30–100 inbound calls/day at a mid-size shop, 2–4× in peak season; ~62% of calls unanswered (~$210 each); book rate ~65–75%; service work 2–3× more profitable than installs. **(confirm.)**
- **Plumbing/electrical:** same volume profile; emergencies = pricing power + the most visceral speed-to-lead pitch; single-tech jobs run higher margins.
- **Roofing:** the ticket monster; lead follow-up on aged/unconverted leads is the #1 pain (DBR goldmine); storm-cyclical, low repeat.
- **Med spa:** ~$9k LTV, booming, affluent clientele already spending $6.5k–$10k/mo on marketing; leads die at the front desk; light compliance + heavily courted by specialist agencies.

## 2.4 The ROI math (the close)
Mid-size shop: ~50 calls/week × ~62% unanswered = ~30 missed/week. At a 65% book rate × ~$340 ticket, recovering even **5 missed calls/month** ≈ $1,000–$2,250 — the monthly fee is paid before counting installs, SEO, reactivation, or reviews. A shop missing 20/week bleeds **$300k+/year.** You sell back money already leaking out of the phone line.

## 2.5 The saturation objection, handled
Home services is the most-pitched niche by GHL resellers. It doesn't matter: (1) **scale** — ~423k HVAC techs and tens of thousands of shops; you need 30–50 clients, statistically invisible against the market; (2) **differentiation** — almost all those resellers push the cookie-cutter $297 "we'll set up GHL" package; your GEO layer, custom agents, transparency dashboard, and guarantee are exactly the upgrade a burned owner reaches for. Saturation only kills you if you're identical.

## 2.6 Validation playbook (before you commit)
1. **Count prospects** — Google Maps your metro; want 100+ shops in reach.
2. **Mystery-shop 10** — call after 6pm; watch how many hit voicemail. That becomes your live demo.
3. **Read willingness-to-pay** — running Google LSAs / on Angi / buying leads = they'll pay for a system that closes more.
4. **Niche tighter if needed** — residential-only, or one trade in a Sun Belt growth metro.
5. **Start where you have warmth** — first 3 clients from a connection, not cold outreach.

## 2.7 Expansion path
Clone the engine to plumbing/electrical first (near-identical), then roofing or med spa. Each = build once, redeploy.

---
---

# PART 3 — THE ECONOMICS

## 3.1 The mid-ticket band + buyer psychology
**Band:** ~$997–$1,997/mo + a setup fee — below the $5k "needs a board meeting" line, above the "$297 = junk" line. The market backs it: AI-service packages that actually sell cluster at $1,500–$5,000/mo. **(confirm.)** This is the **"an owner can approve it alone"** zone. Frame the price as *the cost of a part-time employee* that delivers *a whole department.* The setup fee funds delivery labor and filters tire-kickers.

## 3.2 The unlock: build once, redeploy
Everything expensive — custom agents, n8n recipes, the programmatic SEO engine, the snapshot, RAG templates — is a **one-time build per vertical.** Client #1's setup + first months fund it; clients #2–50 reuse the same engine, so marginal cost per client is tiny. **Niching is an economics decision:** 30 clients on one engine ≈ 85% margin; 30 bespoke builds ≈ bankruptcy.

## 3.3 Unit economics (per client / month — confirm)

| Cost item | Est. | Note |
|---|---|---|
| GHL sub-account | low double digits | flat agency plan amortized across clients |
| n8n (self-host VPS) | pennies each | one server serves all clients |
| Vector DB | low single digits | Qdrant self-host / Pinecone |
| AI API tokens | ~$10–30 | the variable to watch; cheap models for cheap tasks, hard caps |
| SMS/voice minutes | $0 to you | **rebilled** to client |
| Human delivery time | ~$0 after build | templated + AI-assisted delivery |
| **True variable cost** | **~$100–250** | → **~85%+ gross margin** once the build is amortized |

## 3.4 Packaging & pricing (confirm all numbers)
- **Setup (one-time):** ~$1,995 — full deployment, AI front desk trained on their business, site/funnel, SEO engine live (week one).
- **Core monthly:** ~$1,497/mo — the running engine (everything in §5). Live dashboard included.
- **Usage:** SMS/voice/AI minutes **rebilled transparently** (cost or small markup) via SaaS mode — never touches your margin.
- **Add-ons:** Paid-ads management +$500/mo · Premium AI-search (GEO) +$300/mo · custom website build (one-time) · extra agents.
- **Performance kicker (optional):** lower base + ~5% bonus on an agreed revenue target — aligns incentives; use selectively.
- **Contract:** month-to-month after an initial term **(confirm)**; annual prepay discount for cash flow.
- **Tiering:** one strong core tier + paid add-ons (avoids building/maintaining many tiers; lets some clients climb to $2k+).
- **Anchor it:** vs a receptionist ($3k+/mo), a standalone SEO agency (~$3,200/mo), a marketing coordinator ($4k+/mo) — you give them an AI front desk + SEO/GEO engine + reactivation + dashboard for ~$1,500, *less than just-SEO-alone doing 10× more.*

## 3.5 Scale math (illustrative)
At ~$1,497/mo and ~$150–250 variable → ~$1,250+ gross profit/client/mo → **30 clients ≈ ~$45k MRR at ~85% margin**, on top of a single one-time engine build. Build-once-redeploy is what makes both "phenomenal margin" and "insane value" true at the same time.

## 3.6 Margin traps to avoid
- **Per-client custom work** — the #1 killer; template everything, bespoke only as a paid premium.
- **Token sprawl** — cheap models for cheap tasks, hard caps, monitor spend.
- **Support load** — AI deflection + async + knowledge base, not a phone that rings.
- **Underpricing setup** — it funds the build; never give it away.
- **Scope creep** — fixed deliverables; everything else is a priced add-on.

---
---

# PART 4 — THE SERVICE SPECIFICATION

> The operational definition of what we install and run. Sales language only in §4.12; deep economics in Part 3.

## 4.1 What it is (recap)
A done-for-you, white-labeled growth system for residential HVAC contractors that (1) answers every call/message and books the job, (2) recovers missed calls and dead leads, (3) gets the contractor found in Google **and** AI search, and (4) proves every dollar on a live dashboard — sold as outcomes on a mid-ticket subscription.

## 4.2 Ideal client profile (operational)
Owner/operator residential HVAC, ~$500k–$8M revenue, 2–25 techs, in a dense-enough competitive metro. **Best-fit signals:** already buying leads; missing calls; a dormant database; weak review cadence; invisible in AI search; burned by a prior agency. **Not a fit:** commercial-only/new-construction with no inbound phone flow; pre-revenue; micromanagers; anyone unwilling to let AI answer calls.

## 4.3 Core promise & dollar logic
"We make sure your phone is never unanswered, your old leads turn back into jobs, and you get found everywhere customers look — and we show you every dollar." (Math in §2.4.)

## 4.4 The service stack (modules)
Each module: **Does · Client gets · Included · Not included · Built with · Ongoing · KPIs.**

### 4.4.1 Foundation — Website & Five-Star Funnel
- **Does:** conversion-ready web presence + high-intent funnel.
- **Client gets:** a fast, mobile, schema-rich site/landing system + a five-star lead funnel (offer → form/booking → confirmation → follow-up).
- **Included:** branded site/funnel, booking calendar, lead capture, confirmation flows, tracking pixels.
- **Not included:** custom multi-page corporate sites beyond template scope (add-on); client photo/content sourcing beyond templatized.
- **Built with:** GHL funnels/sites + HVAC snapshot; optional custom Next.js site as a premium add-on.
- **Ongoing:** conversion monitoring, periodic offer/A-B tweaks.
- **KPIs:** visit→lead, form completion, booking rate.

### 4.4.2 Business Phone Number & Unified Comms
- **Does:** tracked business line + one inbox for all conversations.
- **Client gets:** dedicated/ported number; unified inbox (calls, SMS, web chat, FB/IG/WhatsApp, email).
- **Included:** number provisioning/port, call tracking/recording (where lawful), routing rules.
- **Not included:** carrier/landline contracts; handsets.
- **Built with:** GHL phone (Twilio/LC Phone); SaaS-mode usage rebilling.
- **Ongoing:** number health, A2P 10DLC registration, routing.
- **KPIs:** inbound volume, answer rate, response time, channel mix.

### 4.4.3 AI Front Desk (Voice + Conversation AI) — *centerpiece*
- **Does:** answers calls and messages 24/7 in a natural voice, qualifies, answers FAQs, books into live availability, escalates the rest.
- **Client gets:** an AI receptionist trained on their business that never misses a call, including peak-season overflow.
- **Included:** persona + knowledge base (services, pricing rules, hours, service area), call objectives, escalation triggers, SMS/chat agent across channels, booking integration, post-call workflow triggers.
- **Not included:** complex negotiation/disputes/anything needing human judgment (routes to a human — realistically AI handles ~70–80% autonomously).
- **Built with:** GHL Voice AI + Conversation AI (Agent Studio); deeper behavior via custom agents (n8n + Claude/GPT + RAG, §4.5); trained on their real objections/recordings.
- **Ongoing:** prompt/knowledge tuning, monthly "jobs booked by AI" review, drift monitoring.
- **KPIs:** % calls answered, containment %, transfer rate, calls→bookings, after-hours capture.
- **Compliance:** AI identification on calls (§4.6).

### 4.4.4 Missed-Call Recovery / Speed-to-Lead
- **Does:** the safety net — instant text-back + AI follow-up on anything unanswered.
- **Client gets:** zero silent missed calls.
- **Included:** instant text-back, AI re-engagement, retry cadence, routing.
- **Built with:** GHL workflow triggers + Conversation AI.
- **KPIs:** recovery rate, recovered bookings, time-to-first-touch.

### 4.4.5 Auto Call/Text Follow-up & Nurture
- **Does:** automated multi-touch follow-up across the lifecycle (new lead, post-quote, reminders, post-service).
- **Client gets:** nothing falls through the cracks; quotes chased; no-shows down.
- **Included:** lead-nurture, quote follow-up, reminders/confirmations, post-service sequences — multi-channel (SMS/email/voicemail drop).
- **Built with:** GHL workflows + Conversation AI; templated HVAC sequences.
- **KPIs:** quote→close, no-show rate, reply rate.

### 4.4.6 Database Reactivation (DBR)
- **Does:** mines old customers + unconverted quotes and reactivates them into jobs ("found money").
- **Client gets:** revenue from leads they already paid for, on day one.
- **Included:** import + cleanup, segmentation (recency/value/service), AI-personalized multi-channel outreach, offer A/B tests, opt-out handling, dollar-framed results.
- **Not included:** purchased/cold lists with no prior relationship.
- **Built with:** GHL + Conversation AI + segmentation; optional propensity scoring (§4.4.12).
- **Ongoing:** periodic re-runs (seasonal, maintenance-due).
- **KPIs:** reactivation rate, revenue/campaign, cost per reactivated job.
- **Compliance:** strict consent/quiet-hours/opt-out (§4.6).

### 4.4.7 Review & Reputation Engine
- **Does:** grows 5-star reviews, responds to all in brand voice, recycles wins into content.
- **Client gets:** a climbing review profile and managed reputation, hands-off.
- **Included:** compliant review solicitation (no illegal gating), AI responses in brand voice, GBP posting, reviews→social repurposing, negative-review alerting/routing.
- **Built with:** GHL Reviews AI + GBP + Conversation AI; optional GBP autopilot.
- **KPIs:** review velocity, avg rating, response rate/time.

### 4.4.8 Programmatic Local SEO Engine
- **Does:** generates/maintains localized service pages (service × town) with injected schema, indexed fast; plus GBP optimization.
- **Client gets:** broad local organic visibility + high-intent organic leads.
- **Included:** programmatic pages from a data source, auto-injected `LocalBusiness`/`Service`/`Review` schema, IndexNow, answer-first clusters with internal linking + QA pass, GBP posting/Q&A/attributes, grid-by-grid (heatmap) rank tracking.
- **Not included:** large-scale link-building (add-on); non-local informational content beyond the GEO set.
- **Built with:** data store (Airtable/Sheets) → generation pipeline → CMS/site; schema injector; IndexNow; content scored vs Surfer/Semrush; human/AI QA.
- **Ongoing:** page expansion, refresh, technical-SEO watchdog.
- **KPIs:** indexed pages, local pack/heatmap rank, organic leads, cost per organic lead.

### 4.4.9 AI-Search Visibility (GEO / AEO) — *fresh differentiator*
- **Does:** gets the client named/recommended in ChatGPT, Gemini, Perplexity, Google AI Overviews.
- **Client gets:** visibility in the new front door of search competitors don't know exists.
- **Included:** `llms.txt`, entity/schema optimization, answer-first structured content, third-party citation/consistency work, **AI-citation tracking with sentiment** + gap-filling content.
- **Built with:** shared content/schema pipeline (4.4.8) + AI-visibility tracking.
- **KPIs:** AI citation rate by engine, share-of-voice vs competitors, AI-referred traffic/leads.

### 4.4.10 Job-to-Marketing Pipeline
- **Does:** turns every completed job into marketing automatically.
- **Client gets:** constant real content (case studies, social, GBP posts) + review requests from job photos, effortlessly.
- **Included:** tech uploads photos → agent drafts case study + social + GBP post + fires review request + updates portfolio.
- **Built with:** custom agent (§4.4.12) + GHL Social Planner + Content AI.
- **KPIs:** content output, social engagement, review requests sent.

### 4.4.11 Remarketing Campaigns
- **Does:** re-engages non-converting visitors/leads across ad platforms + owned channels.
- **Client gets:** more conversions from traffic already paid for.
- **Included:** pixel/audience setup (Meta/Google), retargeting audiences, owned-channel remarketing, creative templates.
- **Not included:** primary cold paid-ads management (add-on); ad budgets (client-funded, passed through).
- **Built with:** GHL + ad-platform pixels/audiences; ties to dashboard.
- **KPIs:** remarketing conversion rate, cost per recovered conversion.

### 4.4.12 Custom AI Agents (catalog)
Deploy a fixed subset per tier; build once, redeploy.
- **Customer-facing:** omni-channel front-desk brain (4.4.3); AI quoting/estimating (collects details + photos via vision, returns ballpark, books site visit); RAG knowledge agent (accurate answers from their pricing/policies); booking/reschedule/no-show-recovery agent; DBR/win-back agent.
- **Owner-facing ("wow"):** **"Ask your business"** agent (owner texts "how'd we do this week / hottest leads / no-show rate," gets a plain-English answer from live CRM); job-to-marketing agent (4.4.10); competitor-watch agent (weekly digest); report-writer agent (auto monthly report — also cuts your labor); collections agent (polite invoice chasing).
- **Built with:** n8n/LangGraph + Claude/GPT + RAG (vector DB) + connections to GHL/booking/inbox; cheap/local models for classification/routing.
- **Guardrails:** deterministic + AI mix, human-in-the-loop on money/send actions, escalation, compliance (§4.6).
- **KPIs:** per-agent containment %, hours saved, accuracy/QA pass rate.

### 4.4.13 Live Revenue Dashboard & Attribution — *proof layer*
- **Does:** shows every call, lead, booking, review, and dollar attributed to the engine, in real time.
- **Client gets:** total transparency tied to revenue (answers the #1 reason businesses fire agencies).
- **Included:** live dashboard, source attribution, weekly/monthly auto-reports.
- **Built with:** GHL dashboards + custom BI/reporting agent.
- **KPIs:** reports all other modules' KPIs; internal KPI: client log-in/engagement (retention signal).

## 4.5 Technical architecture
Layers 2–4 are **yours, bespoke, portable** — the moat.

| Layer | Role | Built with |
|---|---|---|
| 1 — Front office & comms | CRM, calendar, SMS/email, base AI | GoHighLevel (white-labeled, SaaS mode) |
| 2 — Orchestration | the nervous system; custom workflows | **self-hosted n8n** (+ Make for quick flows) |
| 3 — Intelligence | agent reasoning | Claude/GPT via API; Ollama/local for cheap tasks |
| 4 — Memory/knowledge | per-client RAG | vector DB (Qdrant / Pinecone) |
| 5 — SEO/GEO engine | programmatic pages, schema, indexing, AI-visibility | Airtable/Sheets → pipeline → IndexNow + tracking |
| 6 — Proof | dashboard tying it to dollars | GHL dashboards + BI/reporting agent |

**Data flow (typical lead):** inbound → GHL captures → n8n orchestrates → AI front desk qualifies + books → workflow triggers follow-up/review/pipeline → dashboard attributes the dollar. **Cost rule:** cheap/local models for routing/classification/embeddings; premium models for genuine reasoning; cap usage; monitor spend.

## 4.6 Compliance & risk layer (sold as a feature)
- **TCPA:** consent, opt-out, quiet-hours on all SMS/voice/DBR (violations $500–$1,500 each; repeat → class actions). **(confirm.)**
- **A2P 10DLC:** brand/campaign registration for deliverability.
- **AI disclosure:** identify AI on calls per applicable law. **(confirm jurisdiction.)**
- **Data handling:** per-client isolation, least-privilege, no secrets in repos.
- **Human-in-the-loop:** required for any action touching money or outbound customer sends.
- **Positioning:** "powerful **and** legally safe."

## 4.7 Packaging (summary)
Setup + monthly + transparently rebilled usage + paid add-ons; one core tier; optional performance kicker; month-to-month after an initial term. **Full pricing, anchoring, and unit economics in Part 3.**

## 4.8 Onboarding & delivery process
Target: **live and answering within days** (templated deployment makes this possible — and it's a selling point).
- **Phase 0 — Sale/audit:** the AI Visibility + Missed-Revenue Audit doubles as discovery.
- **Phase 1 — Intake (Day 0–1):** self-serve form → auto-provision GHL sub-account → install HVAC snapshot → collect business info, services/pricing rules, hours, service area, access, DBR list.
- **Phase 2 — Deploy (Day 1–3):** site/funnel live, number provisioned/ported (start A2P early), AI front desk trained, sequences on, review engine on, SEO engine seeded, dashboard live.
- **Phase 3 — Activate (Day 3–7):** DBR first run (early "found money" win), agents deployed, go-live test (mystery-shop their line), quick-start videos.
- **Phase 4 — Optimize (Week 2–4):** tune AI/sequences, expand SEO/GEO pages, first results review.
- **DFY vs client input:** DFY = everything technical/operational. Client input = business facts, asset access, approving the offer/voice, job photos.

## 4.9 Operations & support model (keep margin phenomenal)
- **Templated everything** — near-zero labor per client after the vertical build; **no per-client custom work** except paid premium.
- **AI delivers the delivery** — report-writer agent does monthly reports; knowledge-base agent + async support deflect tickets; human-in-the-loop only where it matters.
- **Support:** async-first (knowledge base + ticketing); define response SLA **(confirm)**.
- **Cadence:** monthly revenue-framed review per client; weekly internal client-health monitoring.
- **Monitoring:** AI drift/hallucination checks, deliverability, token spend, schema/SEO watchdog.

## 4.10 KPIs, reporting & success metrics
- **Client-success KPIs (dashboard):** % calls answered, containment %, missed-call recovery, leads, call→booking, quote→close, no-show, reactivated/recovered revenue ($), review velocity + rating, organic/heatmap rank, AI citation rate, total attributed revenue.
- **Business KPIs:** MRR, gross margin/client, clients per engine, CAC, churn/retention, setup-to-live time, support hours/client, token cost/client.
- **Monthly review leads with money:** "here's what we made you, here's the plan."

## 4.11 Guarantee / risk reversal
- **Setup guarantee:** "Live and answering within [X] days **(confirm)** or your setup is free."
- **Optional results guarantee** (selective): a defined threshold or partial money-back, tightly scoped.
- **Scope discipline:** define what's guaranteed, exclusions, and client responsibilities in the agreement.

## 4.12 Sales & positioning (the pitch)
- **Wedge:** the AI Visibility + Missed-Revenue Audit (see §1.7).
- **Live demo / reverse demo / the number** (see §1.7).
- **Anti-agency framing:** transparency + guarantee + AI-search + live-in-days, contrasted with the agency they fired.
- **Objection handling:** "different from other GHL guys?" → mechanism + proof + guarantee + AI-search; "robot calls?" → near-human latency + human escalation; "legal?" → compliance layer; "long contract?" → month-to-month; "real cost?" → fee + transparent rebilled usage.

## 4.13 Growth & roadmap
Land-and-expand (enter on DBR "found money" or AI visibility → expand to full engine → sell add-ons off the dashboard). Vertical expansion: plumbing/electrical → roofing/med spa. Capability roadmap: deeper predictive scoring, more owner-facing agents, expanded GEO tracking, partner program, client community/education.

## 4.14 Operating principles (don't screw this up)
Productize ruthlessly (one vertical, fixed agent suite, repeatable deployment). No per-client custom work except paid premium. Control token sprawl. Deflect support with AI + async. Don't underprice setup. Kill scope creep. Own the moat layers (2–4). Guardrails + compliance on anything that sends or touches money. (Margin specifics in §3.6.)

## 4.15 Glossary
- **DBR** — Database Reactivation. **GEO/AEO** — Generative/Answer Engine Optimization (getting cited in AI search). **RAG** — Retrieval-Augmented Generation. **SaaS mode** — GHL white-label reseller mode with usage rebilling. **A2P 10DLC** — carrier registration for app-to-person SMS. **Containment %** — share of conversations AI handles without a human. **Book rate** — % of inbound calls converted to booked jobs.

---

*This is the canonical, all-in-one playbook for [SYSTEM_NAME]. Confirm every **(confirm)** value — pricing, timelines, contract terms, industry stats, and compliance specifics — before anything goes client-facing.*