The deck is the focused version. This is the depth: every assumption with its source, the full model, the bets in detail, and the systems I'd build. Read it if you want to pressure-test the math. Nothing here is a fact until we validate it against your real data; it is a defensible starting hypothesis.
No internal data yet, so every number is a stated assumption. Two are your own figures (captured during the process); the rest are 2026 benchmarks with sources. If your real numbers differ, the model flexes and I show the math so we can swap inputs live.
| Assumption | Value used | Source |
|---|---|---|
| Landing page: visit → booked demo | 3% base, up to 5% tuned | daydream 2026 (demo pages 1.5-4%); Sam Kuehnle (3.8% median) |
| No-show rate (booked → held) | 25% (75% show) | highticketaisystems (~30% unmanaged); J. Donovan (10-20% managed) |
| Demo held → close | ~50% | Shaun, recalled from the call, to confirm. Market benchmark is 39% (Powered by Search) |
| Qualified rate on paid traffic | 70% | Assumption (bidding on high-intent + qualifying on the form) |
| CPC, AI-category terms | $30-45 | DataForSEO, real US Google Ads data |
| CPC, vertical terms | $100-167 | DataForSEO, real US Google Ads data |
| Customer lifetime (for LTV) | 20-30 mo realistic; 12 mo ultra-conservative floor | userjot, hubifi, Optifai (SMB); Parse Labs, Fractal (vertical SaaS is stickier) |
| Monthly fee (LTV basis) | $500/mo | Captured during the process. Internal use only |
| Average job ticket | ~$600 | Captured during the process ("one job = payback") |
| Email list size | 40,000 | The brief |
| B2B Google Ads CPL (cross-check) | $70-120 | Starr Conspiracy, Flyweel |
North star = qualified booked calls per month. The target is at least 50 (the brief says 50+), going higher with proven metrics and budget to scale the winners.
50 booked → 75% show = ~37 held → ~50% close = ~19 new customers → at $500/mo that is ~$9,400 in new recurring revenue every month, roughly $113k in new annual revenue per month of run-rate. Even at the 39% market close benchmark it is ~14 customers and ~$7.2k new MRR, still healthy.
| Qualified booked calls / month | Month 1 | Month 2 | Month 3 | Logic |
|---|---|---|---|---|
| Referral / organic | 5 | 5 | 5 | The organic growth they already have |
| Bet 1 · List | 8 | 13 | 12 | 40k re-engagement funnel + rep outreach on scored leads |
| Bet 2 · Partners | 4 | 8 | 12 | 3 waves, building |
| Bet 3 · Paid | 3 | 10 | 21 | Spend ÷ cost per demo ($1.5k → $1.15k → $950) |
| Total | ~20 | ~36 | ~50+ |
$35 CPC ÷ 3% page conversion ÷ 70% qualified ≈ $1,670 per qualified booked call early, tuning down to ~$875 once fully optimized (the month-3 forecast still uses ~$950 while the account is learning). A booked call is a much deeper action than a raw lead, which is why it costs more than the $70-120 CPL benchmark. Then CAC = cost per booked call ÷ (75% show × 50% close = 0.375) = ~$2,330 tuned to ~$4,450 early.
LTV = $500/mo × lifetime, then ÷ CAC. 3x is the healthy line. Cold paid looks tight at a very conservative 12-month life, but at the real 20-30 month range it clears 3x, even before tuning. So we scale paid as it proves out and confirm lifetime with your cohort data. Warm demos cost about $0.
| Customer lifetime | LTV | Paid, early ($4,450 CAC) | Paid, tuned ($2,330 CAC) |
|---|---|---|---|
| 12 mo (very conservative) | $6k | 1.4x | 2.6x |
| 18 mo | $9k | 2.0x | 3.9x |
| 24 mo (realistic) | $12k | 2.7x | 5.1x |
| 30 mo | $15k | 3.4x | 6.4x |
Two warm bursts to hit the number now, one engine to keep it after peak season. The thesis: inbound is unbuilt, so the fastest path is the demand you already own (list + partners) at near-zero cost, while paid is built and tuned to become the durable channel.
Funnel: 40k list → email re-engagement + Meta remarketing → lead scoring → booked call. Two paths to convert, so we never depend on self-booking: people book from the page, and the rep does outreach on the highest lead-scoring contacts who engaged but did not book.
Assets activated: the list, the ROI calculator, and the live voice-agent demo (the "call and hear it" proof). Timing: hard weeks 1-6, then light nurture on new entrants. Media: ~$1-2k/mo retargeting. Risk: list may be stale (R1), so I measure the first send before projecting the rest.
Funnel: partner audience → dedicated podcast episode + newsletter drop → co-branded ROI calculator / page that names the partner (message match) → booked call.
Assets activated: the partnerships (Service Business Mastery podcast, Contracting Edge, Blue Collar Success Group) and their newsletters and co-marketing flexibility. I would also repurpose the existing dozen masterclasses as the webinar or gated content for a wave, so an asset you already produced does new work at no cost. The conversion asset stays the ROI calculator / page, which converts to a booked call better than a masterclass on its own. Timing: 3 waves (weeks 2-3, 6-7, 10-11). Media: ~$0, co-marketing.
Funnel: Google Search + Meta → conversion page split by trade → form or calendar → booked call → CRM feedback loop tunes the campaigns. Leads with cheap AI-category terms ("ai receptionist" 5,400/mo, $30-45 CPC), with vertical terms used surgically for message match.
Assets activated: the Phase-1 landing page and the demand research. Timing: day 1 onward, small at first (learning phase), scaling with proof into the fall search peak. Media: the bulk of the budget, ramping. This is the channel built to scale and renew every month.
Lean, mostly on the existing stack. In priority order:
This is where I move fastest, and it matches how you work. The goal is workflows that get smarter, not campaigns rebuilt from scratch.
Claude generates ad, email and page variants per trade in the contractor's own language, so we test many message-match angles cheaply and keep the winners.
HubSpot scoring plus Zapier / n8n flows: opens, clicks, site visits, downloads and masterclass views feed a score, and hot leads route to the rep automatically. No lead goes cold.
Dynamic page content and offers by vertical and by source (list, partner, paid), so the message always matches where the click came from.
"ai receptionist" already triggers Google AI Overviews. I would create structured, citable content so Breezy shows up in ChatGPT, Perplexity and AI Overviews over time. A compounding, low-cost channel for later.
The north-star dashboard pulls from the ad platforms and HubSpot via API and updates itself, so budget calls are made on live cost-per-booked-call, not a monthly spreadsheet.
The AI voice agent is itself the best top-of-funnel asset: "skeptical? call and hear it." It disarms the AI-skeptic buyer better than any ad.
Week 1 is mostly this, because without attribution I would be optimizing blind.
Demand is real and mostly cheap. Real US Google Ads data (DataForSEO). The AI-category terms are large and cheap; the vertical terms are small but pristine and low-competition.
| Keyword | Volume/mo | CPC | Bucket |
|---|---|---|---|
| ai receptionist | 5,400 | $29.89 | AI, solution-aware (cheap) |
| ai answering service | 1,900 | $44.48 | AI, solution-aware |
| ai voice agent | 1,900 | $38.54 | AI, solution-aware |
| after hours answering service | 1,300 | $183.59 | Problem-aware (after-hours) |
| plumber answering service | 590 | $158.13 | Vertical (pristine, low comp) |
| hvac answering service | 480 | $155.79 | Vertical (pristine, low comp) |
| contractor answering service | 170 | $167.63 | Vertical (pristine, low comp) |
Pools: AI core ~11-13k/mo; vertical home-services ~1.4k/mo (small but exact-fit). Seasonality: the operational pain peaks in summer (contractors slammed), but people searching for the fix peaks in fall (Sept-Oct, roughly 2x the summer trough). Our 90 days (Jul to Oct) cover both, which is why warm leads now and paid ramps into the fall.
Breezy is invisible in search today: zero paid keywords and ~27 organic keywords (~110 visits/mo). Greenfield. The category is crowded with horizontals (RingCentral, Nextiva, Zoom, IONOS, Synthflow, and Jobber as the closest vertical threat), plus a lot of "build your own with no-code" content. Breezy's wedge: none of them lead with home-services depth (on-call routing to the on-call tech, transfer triggers, automatic booking, go-live in 1 hour, done-for-you). That is the message-match angle for the pages and ads.
Rule: paid scales only as its cost per booked call proves healthy. Warm carries the early number cheaply, so we can stay lean early and open the throttle only once the economics are proven.
| Media budget | Month 1 | Month 2 | Month 3 | 90-day total |
|---|---|---|---|---|
| Scenario A · Lean (recommended) | $5-8k | $12k | $20k | ~$35-40k |
| Scenario B · Accelerate | $10k | $22k | $35k | ~$65-70k |
Where the media goes (Scenario A): Paid engine ~85%, list retargeting ~10%, partners ~5%. The warm bets carry the early number at almost no media cost. Media only: the first 90 days run on the tools you already have (Mailchimp, HubSpot, Calendly), adding tooling when volume earns it. Not a media cost but a dependency: a second closer as booked calls approach 50/mo.
| Risk | Mitigation |
|---|---|
| R1 · The 40k list is stale (the #1 risk) | Warm it, clean bounces, start with a recent segment, and measure the first send before projecting the rest. |
| R2 · Clicks cost more than planned | Lead with cheap AI terms, strong negatives, qualify on the form, cap spend until CAC is proven. |
| R3 · Conversion rates below benchmark | Fast A/B, the "call and hear the AI" proof CTA to lift booking and cut no-show, measure by channel and reallocate. |
| R4 · Sales capacity | One rep can run ~2.5-3 demos/day comfortably; outbound only on hot scored leads; Shaun backs up; add a 2nd closer as volume nears 50. |
| R5 · Retention / LTV unknown | Validate cohort data early; scale paid only with proven CAC/LTV; conservative base + sensitivity table. |
| R6 · Warm assets are finite | Build paid in parallel from day 1 so it takes the baton; keep nurturing new list entrants. |
| R7 · Seasonal window | Front-load warm now; paid rides the fall search peak. The 90 days cover both. |
| R8 · Tracking gaps in the stack | Week-1 audit and setup; without attribution we optimize the wrong thing. |
| R9 · Partner timing | Multiple partners, plug-and-play assets, dates locked in advance; no single point of failure. |
| R10 · AI-skeptical buyer + commoditization | Outcome-first messaging, "teammate not replacement", live proof, and vertical depth as the moat. |
| R11 · Limited budget / resources | Scenario A, media-only, existing stack; scale only with proof of CAC. |
| R12 · Ad-account learning phase | New accounts need 2-4 weeks of data; month 1 paid is a controlled test. This is the one real risk to a clean 50 exactly on day 90; it could slip a couple of weeks. I still commit to the number, with paid and the list Plan B as the levers. |
Links open in a new tab. Demand and CPC data are our own DataForSEO pull (real US Google Ads data), so there is no public URL for those.
02-research/keyword-research.md.