One verified person, sold once, with a dated reason they need you this week, priced as a premium because nothing else compares. Here is the quality ladder, the credit model, the packages, and three live calculators that prove buyer ROI and the economics, defensibly, every number sourced.
The bottom line: HeatSeak sells the best lead intelligence on the market, exclusive, intent-rich, verified, and prices it as a premium because buyer ROI is the only ceiling. Price to the buyer's closed-job value, gate by credit tier, discount on volume to drive expansion. Because the inputs are so efficient, we can deliver the highest quality and keep a healthy margin, quality leads, margin follows.
CONFIDENTIAL · QUALITY-FIRST PRICING MODEL · JUNE 2026 · BASELINE 2,000 LEADS/DAY = 60,000/MO
Type what you want. HeatSeak fuses public sources, resolves the contact, verifies the mobile, grades the lead, and hands it back priced in credits. Here's the actual surface a buyer touches.
The buyer never sees COGS, they see a prompt, graded leads, and a credit meter. That's the product.
Six stages, fully autonomous. Free public signal goes in one end; a priced, graded, verified lead comes out the other.
Every field is fused, not bought as a finished record, owner, mobile, email, address, and buying signal come from separate public sources matched on time + geo + entity.
Provenance chips label each field's origin (free public / verified dip / model). Green = free, violet = proprietary. The buyer sees confidence; we see a sub-cent cost.
The grade is earned, not assigned, a model trained on real buyer outcomes predicts conversion and stamps Bugatti / Premium / Standard.
Priced in credits, mapped to buyer value (1 / 2 / 3 / 10). This single card costs us <1¢ to manufacture and sells for $0.665-$25+.
Fused from multiple partial sources on shared keys, time + geo + entity, then graded, freshness-dated, and exclusively owned. Drag the lead up the five quality tiers: watch the fields, the buying signal, the grade, and the price climb together. This is the product unit.
Clean name + phone + a carrier-verified mobile + area/category intent. You know they exist and you can dial a real cell. You do not yet know they need you today.
Every field carries provenance. Green chip = free public data · violet = paid enrichment · cyan = verified dip. The buyer sees confidence; we see a sub-cent cost.
One caveat we honor: per the validation gate, every tier above T1 is a candidate until a live call (or AI pre-qual in T5) confirms right-person + real event + interest. The right_person, event, and ai_qualified fields are confidence claims, labeled as such, not delivery proof, until validated.
The ladder moves on three axes at once, and margin widens as you climb, because the highest-value layer (the buying-signal overlay) is FREE. A ~3¢ base climbs to a 10× premium.
The cost → price climb, red is our cost (barely visible, that's the point), the cyan/violet bar is the sell price, normalized to the T5 max:
| Tier | What's in it | Our cost | Conversion | Sell price | Credits* |
|---|---|---|---|---|---|
| T1 · STANDARD | Clean name + phone + carrier-verified mobile + area/category intent. The base callable record. |
~$0.01 | 1-3% | $5-15 | 1 |
| T2 · VERIFIED | T1 + right-person verification (~85%) + DNC-clean + address-bound. The Brooke-gate floor. |
~$0.08 | 3-6% | $15-40 | 2 |
| T3 · INTENT 🔥 | T2 + a specific buying-signal EVENT (DOH violation / hail / pre-foreclosure / citation) with source + date. The biggest jump, for ~free. |
~$0.09 | 8-18% | $20-80 PI $250-600 | 3-6 |
| T4 · ENRICHED | T3 + verified email (email-verification layer) + firmographics (B2B enrichment layer). Multi-channel + deal-sizing. |
~$0.18 | 12-22% | $40-120 | 5-8 |
| T5 · EXCLUSIVE · REALTIME | T4 + grade/score + guaranteed exclusivity + realtime freshness (<72h) + optional AI pre-qualification. Buyer can almost dial blind. |
~$0.11-0.20 | 20-35% | $80-200+ PI $400-600+ | 8-15 |
* Illustrative credit weighting, anchor to the client's per-credit price; the point is the tier multiplier.
Identity is a static fact a million people share ("a homeowner in 11215"). An event is a person who needs the service this week ("the rat-activity inspection at 412 7th Ave FAILED on Tuesday"). Intent is the only variable that moves conversion 10-15×. Every signal below is a free, public, dated event, fused onto the base lead, it's the "why now" no list can carry.
Failed health inspections are a legal + reputational emergency, a forced buyer. Plus the proximity superpower: a rat two doors down means this address is next.
A roof hit by 1.5"+ hail has damage the owner can't see yet. Call before the door-knockers and before they file. Window of urgency = days.
Acute-pain trades, the highest phone-yield wedge in the engine. A pulled permit, a new deed, a no-heat complaint: forced spend with a deadline.
A motivated seller is a dated event: an expired listing, a foreclosure filing, a probate. The person needs out, now.
The highest-value vertical, a single event is worth hundreds because the case value is huge. Fuse crash open-data on time+geo with news + social to name the victim.
Cross-vertical radar: junk for one buyer is gold for another. A storm feeds roofing AND solar; an obituary feeds final-expense AND probate. Re-aim, never discard.
A failed rat inspection across town next month is weaker than one two doors down yesterday. Five factors turn each event into a route-probability score, which doubles as the prediction of "will they want to be routed when we call?"
Exponential decay on freshness. Intent rots fast, hail hottest at 48h, PI at days. exp(−Δd/τ)
Magnitude of the event. 2.5" hail > 0.75". "Failed, critical rodent" > minor citation. Normalized 0-1.
Infected-street decay. The address ON the event = 1.0; a neighbor decays with distance. exp(−m/d₀)
Independent sources agreeing on time+geo+entity. One = a clue; 2+ = trusted. The fusion payoff.
Who the signal attaches to. They reported it = 1.0 (raised their hand); inferred neighbor = ~0.5.
Worked example (Bugatti): a restaurant failed a DOHMH inspection for live rodent activity (S=0.9), 2 days ago (R=0.87), at the address itself (P=1.0), confirmed by both the inspection dataset AND a 311 complaint (C=0.75) the owner filed (W=1.0). intent_raw = (0.40·0.87 + 0.25·0.9 + 0.35·1.0) × 0.75 × 1.0 = 0.69 → ≈69, bumped to Bugatti because self-reported + corroborated + a legal forced-fix this week. The conversion math is the whole argument: web-form leads convert ~1-2%; intent/pay-per-call leads ~30-50%, a ~10-15× gap. At the same $40 price, that's $2,667 vs $114 cost-per-acquired-customer, a 23× better CAC. We don't sell data; we sell the difference between a 1.5% and a 35% close rate.
Angi sells the same lead to 5 of you. Brokers sell a dead row. Apollo sells a list. Google rents you a searcher and taxes the click. HeatSeak manufactures ONE exclusive person, verified cell, address-bound, with a dated reason they need you this week, and never sells them again.
| Dimension | HeatSeak | Angi / HomeAdvisor | Data broker | Apollo / ZoomInfo | Google LSA |
|---|---|---|---|---|---|
| Exclusivity | Exclusive · sold once | Shared 3-5× | Resold infinitely | In every DB | Bid for same searcher |
| Intent / "why now" | Event-level, dated | Real but commoditized | None | Fit, not a buying event | High but generic |
| Freshness | Date-stamped per field | Stale in minutes (shared) | Months-years old | Quarterly, decaying | Real-time |
| Contactability | Verified-mobile + right person | Reachable, but 4th in line | Unknown, high disconnect | Mobile reveal gated | Inbound, they called |
| Conversion | High, exclusive + intent | Low-moderate (split 5×) | Very low (cold) | Low (it's a list) | High but capped + costly |
| Price to buyer | $20-80 · PI $250-600 | $15-100 ÷ 3-5 shared | Pennies (you get what you pay for) | $0.50-2 + $1k-15k/yr seat | $6-300, unpredictable |
| What you actually get | Exclusive, dated, verified, graded person | A shared quote-request | A row in a spreadsheet | A targeting list | A rented inbound contact |
Where HeatSeak is honestly NOT the winner, so the pitch survives scrutiny:
Google LSA's searcher is actively looking right now, the highest raw intent. HeatSeak's edge is exclusivity + enrichment + event-timing + price control + no platform tax, the owned outbound channel that complements rented inbound.
ZoomInfo/Apollo have a wider B2B graph. We win on SMB owner mobile + event triggers + exclusivity, not database breadth, and we use a B2B enrichment + email-verification layer as an input, not a competitor.
A broker sells 10M rows for cheap. We sell far fewer, far better, exclusive leads. Volume is their game; conversion-per-dollar is ours.
Most data businesses get more expensive as they scale. Ours inverts that: more outcomes → sharper targeting → lower cost-per-lead → higher margin. Here's what runs in production right now, and where it compounds next.
A self-optimizing lead fleet. The engine continuously auto-prioritizes the best-performing, lowest-cost avenues and starves the dead ones, cost-per-lead drops on its own as it learns which sources convert. No manual tuning.
Measured real-cost instrumentation. Every lead carries its true acquisition cost from live spend, not estimates. What you see is what it cost, no invented numbers.
Cross-agent QC + corroboration. Every lead is validated across multiple independent sources before it ships. A signal that appears once is held; corroborated across sources, it becomes a lead, the count of agreeing sources is the confidence score.
Multi-modal sourcing + intent overlays + grading. We fuse a live discovery layer, public intent sources (NOAA storm data, 311, DOH/DOB, permits, court records) and identity / line-type verification into one graded record, qualified, intent-rich, contactable, scored.
Any-vertical re-aim. The same engine retargets to a new vertical by swapping the signal set, near-zero rebuild.
Internal dashboard + live demo. Full-circle visibility on sourcing, cost, quality and yield, with a working live demonstration for buyers.
The owned intent-data asset. Every call outcome feeds back in, the outcome flywheel makes the next prediction sharper and the underlying dataset more valuable. We stop renting intelligence and start owning it.
Self-serve SaaS + API. Thousands of buyers pull graded, verified leads on demand, no human in the loop per buyer.
Autonomous vertical expansion. The system explores and validates new sources and verticals on its own (an explore / exploit bandit), opening markets without a manual build each time.
An engine that gets smarter and cheaper over time. More outcomes → better targeting → lower cost-per-lead → higher margin. The flywheel is the moat.
It's not a feature gap, it's a category gap. The whole market has accepted "shared / stale / no-intent" as just how leads work. The demand for the opposite is already there, unmet, in every vertical with a buyer.
The marketplaces, Angi, HomeAdvisor, Thumbtack, sell the same lead to 4-6 competitors at once. The buyer is racing to dial a contact who's already been called five times.
By the time a list-broker or a B2B database (Apollo, ZoomInfo) delivers it, the intent is gone or was never there. You're buying a name, not a moment.
None of them tell you why now. No trigger event, no freshness, no signal that this person is in pain today.
What nobody delivers is the only thing that actually converts: exclusive, intent-rich, verified, dated, and graded leads. A lead that's yours alone, tied to a real trigger event, with a verified live contact, stamped with the date it surfaced, and graded for likelihood to convert. That gap is the entire wedge. We're not creating demand, we're filling a hole the whole market has been working around.
There are two ways to monetize this engine. They look similar from the inside and could not be more different as businesses. The two-path comparison below is the whole strategy in one screen.
Build a custom engine per vertical and run the full circle yourself: source → call → route the call → pay-per-call. You charge more per lead, but you charge a handful of buyers, and you carry the labor of full-circle delivery on every one.
Build the engine once. Re-aim it at any vertical by swapping the signal set. Market it wide and cheap through top-of-funnel distribution (Clippers / content creators, Meta / Google / TikTok ads) instead of hand-selling. Let thousands of self-serve buyers pull leads themselves.
The expensive part (the engine) is built once. Each new vertical is a signal-set swap, not a new company.
Clippers and creators flood the top of funnel at a fraction of the cost, vs hand-selling each vertical one buyer at a time.
Every outcome makes the data asset better → lowers cost, raises quality, wins more buyers. Services revenue adds; platform revenue multiplies.
A dollar of platform / data revenue is worth ~8.5×+; a dollar of services revenue ~0.7×. Same top line, an order-of-magnitude difference in enterprise value.
We build a lead by fusing public sources. Most of it is free; the only real per-lead costs are the discovery layer and the carrier/line-type verification dip. Premium adds a right-person identity-resolution layer. Here is the whole stack, amortized at 60,000 leads/month, costs shown as bands, because the point is the gap, not the line item.
| Capability layer | Cost band | Used per lead | STANDARD | PREMIUM |
|---|---|---|---|---|
| Live discovery layer, discovery + phone backfill | a fraction of a cent | per query | <1¢ | <1¢ |
| Carrier/line-type verification layer, verified-mobile stamp | a fraction of a cent | 1 dip | <1¢ | <1¢ |
| AI extraction/scoring layer, planner / extractor | a fraction of a cent | ~1.8K tokens | <1¢ | <1¢ |
| People-search / public records, B2C base | FREE | 1 lookup | $0.0000 | $0.0000 |
| Open data, 311 / DOH / DOB / NOAA intent overlay | FREE | overlay | $0.0000 | $0.0000 |
| Cloud infrastructure, amortized | a fraction of a cent | ÷ 60,000 | <1¢ | <1¢ |
| Premium identity-resolution layer, "right person ~85%" | a few cents to ~a dime | premium only | - | ~a dime |
| ALL-IN COST PER LEAD | under a cent (basic B2B) | ~$0.20-0.50 (PI/B2C skip-trace) | ||
Not per-lead abstractions, the strict CFO view: what hits the card every month at today's volume, and how each line steps with scale. The discovery layer is a flat plan; carrier verification is usage; the AI layer is free until ~60k/mo; the premium identity-resolution layer only touches the premium slice; infrastructure is flat until a box saturates.
| Line item | Type | Detail | Monthly |
|---|---|---|---|
| Live discovery layer | Flat plan | Starter plan covers ~25k leads; well under today | ~$50 |
| Carrier/line-type verification layer | Usage | ~3,000 verified-mobile dips at a fraction of a cent each | ~$15 |
| AI extraction/scoring layer | Usage | Free tier covers this volume | $0 |
| Cloud infrastructure | Flat | App hosting + database + buffer | ~$90 |
| Premium identity-resolution layer | Usage | premium path only: ~600 records at a few cents each | ~$60 |
| ALL-IN TO OPERATE TODAY (~3,000 leads/mo, premium layer on) | ~$215/mo | ||
Free-path default, identity-resolution OFF, dip only the ~30% of candidates worth stamping. How the engine runs most days.
With the 20% premium identity-resolution layer on. Honest range: ~$140-$215/mo all-in at today's volume.
The discovery plan + infrastructure don't move with volume. You pay for the floor, not the volume, 8× today's leads and these lines wouldn't budge.
The discovery rate halves then halves again across plan tiers. The premium identity-resolution layer collapses from pay-as-you-go to subscription pricing. Infrastructure steps when a box saturates. The carrier-verification layer is the one line that never improves, a flat fraction-of-a-cent per dip the whole way.
| Line | 60k/mo (2k/day) | 500k/mo | 5M/mo | 50M/mo |
|---|---|---|---|---|
| Live discovery layer (plan) | $90 | $500 | $3,000 | $30,000 |
| Carrier verification layer (usage) | $300 | $2,500 | $25,000 | $250,000 |
| AI layer | $0 | $150 | $1,500 | $15,000 |
| Cloud infrastructure | $90 1 box | $190 1 beefy box | $700 2nd box + CDN + queue | $4,500 4-6 box cluster |
| Identity-resolution layer (20% prem) | $1,200 pay-as-you-go | $2,000 subscription | $20,000 subscription | $66,000 enterprise |
| TOTAL BILL | ~$1,680 | ~$5,340 | ~$50,200 | ~$365,500 |
| Effective $/lead | $0.028 | $0.0107 | $0.0100 | $0.0073 |
Total monthly operating bill steps, and per-lead cost falls ~10× as volume rises:
Bars are the TOTAL bill (log-scaled for legibility); the cost per lead falls the whole way, that's the point.
At 50M leads/mo it's $250k of a $365k bill, 68% of total, a flat per-dip rate that never steps down. Action: (a) dip only verified candidates, not every shipped lead → at a 30% dip-rate this line drops ~70%; (b) negotiate a volume rate (the verification layer offers enterprise pricing, held at full rate here, unverified). Fix this one line and the 5M/50M tiers get dramatically cheaper.
~$140/mo is the floor whether you ship 500 or 25,000 leads. You're massively under-utilizing what you already pay for, the discovery Starter plan alone covers 8× today's volume. From a revenue standpoint, pushing volume is nearly free until you cross ~25k leads/mo. The lever is distribution, not cost.
Sell-price is anchored to the real lead-market clearing price per vertical, and to the fact that ours is the highest-quality, exclusive, intent-rich lead in the market. Because the inputs are so efficient, every tier prints a 99%+ gross margin, which is what lets us invest in quality and still keep healthy margin. The only payment-side variable that matters is payment processing (~mid-single-digit %) on the transaction.
Red bar = our cost (barely visible, that's the point). Dashed green = sell price. Bars normalized to the $500 max.
| Tier / vertical | Market price | Our sell price | Our cost/lead | Gross margin | Why premium |
|---|---|---|---|---|---|
| Standard (home-services, SMB) | $20-80 | $25 | a few cents | 99.96% | verified + exclusive |
| Pest control | $20-50 | $22 | a few cents | 99.96% | dated intent event |
| Premium / verified mobile | +20-40% | $40 | ~a dime | 99.73% | right-person proof |
| B2B (owner + verified mobile) | $30-150 | $60 | a few cents | 99.98% | owner mobile + intent |
| Legal / Personal Injury | $250-600 | $300 | ~a dime | 99.96% | huge case value |
| Mass tort / med-legal | $400-1,200 | $500 | ~a dime | 99.98% | exclusive + qualified |
The credit model maps to buyer value, not our cost (our cost is flat at pennies). A PI lead clears at ~10× a home-services lead in the open market, so it costs 10 credits. Defensible because it tracks the external price ratio, not an arbitrary multiplier. Anchor: $1.00 = 1 credit at retail.
Why this holds up: a buyer never sees COGS, they see "1 credit = 1 home-services lead, 10 credits = 1 PI lead," which mirrors that a PI lead is worth ~10× a home-services lead in the open market. The ladder tracks the external price ratio, so it's defensible and never feels arbitrary. Retail anchor: $1.00 / credit; volume packages discount the effective $/credit.
Retail anchor is $1.00/credit. Each tier discounts the effective $/credit as volume rises ($0.80 → $0.50), that discount ladder is the upsell engine. Even the worst case (an Enterprise buyer spending every credit on 3-credit premium leads) holds ~87% gross margin; typical is 95%+.
Worst-case audit: an Enterprise buyer spends ALL 10,000 credits on 3-credit premium leads = 3,333 leads at ~a dime each ≈ ~$367 production + ~$285 payment processing ≈ ~$652 cost on $4,999 revenue = ~87% gross margin. Best case (all standard) = ~99%. Floor margin ~87%, typical 95%+. Self-serve top-up: $1.00/credit (250 min, $1.50 under 250).
Two channels can fill the funnel, performance-paid content creators ("Clippers") and traditional paid ads (Meta, Google, TikTok). They are not equal. One is the engine; the other is the throttle. The math below shows why, and two live calculators let you pressure-test every number.
| Dimension | Clippers | Paid Ads (Meta/Google/TikTok) |
|---|---|---|
| Effective CAC | ~$26 blended | ~$200-600 |
| Pricing model | Performance-only (per view) | Pay-per-impression (auction) |
| Cost per 1,000 views | ~$0.10-$1.00 | Multiples higher (auction CPM) |
| Scalability | High, recruit more creators | Capped, CAC rises with spend |
| Targeting | Algorithmic + niche fit | Exact (interest + lookalike) |
| Speed to test a message | Slower (clip must catch) | Fast (instant on/off) |
| Social proof | Native, built in | Reads as an ad |
| Cost of a miss | ~Zero (no views, no pay) | Full (you paid the impression) |
Set monthly spend and the funnel assumptions. See views, leads, paying users, MRR added, effective CAC, and first-month ROAS, recomputed on every keystroke.
Defaults: 0.08% view→lead · 8% trial · 6% trial-to-paid · $89 ARPU → ~$26 effective CAC at $0.50/1k views
Split one budget across both channels. See customers acquired via each, the CAC advantage multiple (how many times cheaper Clippers are), and Clippers payback in months of ARPU.
Funnel constants: 0.08% view→lead · 8% trial · 6% trial-to-paid · $89 ARPU (mirrors GTM-1 defaults)
Three calculators, all defaults and ranges straight from the model. Drive the cost path, sell price, and volume; pressure-test buyer ROI per vertical; and stress the SaaS LTV against churn and CAC. Nothing here is hand-waved, it's the same arithmetic the deck rests on.
Pick the cost path (the mix slider blends the standard ↔ premium identity-resolved cost path), set a sell price and daily volume. See what one lead costs, what it sells for, and the gross profit + margin at scale.
Std under 1¢ · Premium ~a dime · 60K/mo = 2,000/day baseline
From the buyer's chair. Pick a vertical preset (loads model conversion + job-value defaults), then tune lead price, conversion, and job value. See cost, revenue, ROI multiple, and the no-brainer verdict, how many leads one closed job pays back.
HeatSeak's own business. Set ARPA (or build it from leads/mo × price), gross margin, monthly churn, and CAC. Out comes customer lifetime, LTV, the LTV:CAC ratio, and CAC payback in months. Defaults = the weighted model (ARPA $804, GM 92%, churn 5%, CAC $400).
A buyer-base consuming 50,000 lead-credits/mo, priced at the conservative blended $0.65/credit. Production runs at a few cents/lead; the only meaningful COGS line is the ~mid-single-digit % payment-processing fee. Here is the whole P&L, to scale.
$1,500, 50,000 leads at a few cents blended. The actual cost of manufacturing every lead the base consumes in a month.
$1,853, ~mid-single-digit % of $32,500 revenue. The largest single COGS line, payments cost more than the leads themselves.
$190, cloud infrastructure fixed, with overage headroom. Effectively flat regardless of volume.
The deck above proves a single lead's economics. This proves the platform's, scaled by paying users/buyers, not just leads. Assumption: each buyer pays the verified blended $804/mo ARPA and consumes ~1,200 leads/mo (1,200 credits, blended std+premium). As volume rises, the discovery and identity-resolution layers step into volume tiers, so the per-lead cost falls while fixed infrastructure is amortized across millions of leads. Result: gross margin climbs from 92.00% at 100 buyers and holds in the ~93% band, peaking at 93.26%, across four orders of magnitude it never erodes.
Gross margin vs scale, bars are clipped to the 88-94% band so the improvement is visible (not a flat wall at 90%+):
| Users | Leads / mo | MRR | ARR | Blended lead $ | COGS / mo | Gross profit / mo | GM % | Heads |
|---|---|---|---|---|---|---|---|---|
| 100 | 120,000 | $80,400 | $964,800 | $0.0143 | $6,429 | $73,971 | 92.00% | 2 |
| 1,000 | 1.2M | $804K | $9.65M | $0.0103 | $58,788 | $745,212 | 92.69% | 8 |
| 5,000 | 6.0M | $4.02M | $48.2M | $0.0072 | $275,460 | $3.74M | 93.15% | 35 |
| 50,000 | 60M | $40.2M | $482M | $0.0072 | $2.75M | $37.45M | 93.16% | 160 |
| 500,000 | 600M | $402M | $4.82B | $0.0067 | $27.11M | $374.9M | 93.26% | 650 |
| 1,000,000 | 1.2B | $804M | $9.65B | $0.0067 | $54.77M | $749.2M | 93.19% | 1,100 |
| COGS = lead production (blended std/premium, volume-tiered discovery + identity-resolution layers) + tiered fixed infrastructure + ~mid-single-digit % payment-processing fee. Lead production is only 1.0-2.1% of revenue at every tier; payment processing is the single largest COGS line, and it's a fixed % that never grows faster than revenue. That is why margin holds. Blended-lead-$ shown rounded to 4dp; COGS/GP/GM computed at full precision. | ||||||||
The scaling thesis, proven: COGS is dominated by a fixed-percentage payment-processing fee (~mid-single-digit %), not by leads (1-2% of revenue) or infrastructure (<0.2%). So as users 10,000× from 100 to 1M, gross margin doesn't erode, it rises from 92.00% and holds in the ~93% band (peaking 93.26%) because per-lead cost falls into volume tiers while fixed infrastructure amortizes to nothing. The architecture must evolve single box → queue/workers → multi-region → warehouse → owned data pipeline, and headcount from 2 → 1,100, but even fully-loaded, operating margin rises from 55% → 91%. The only real ceiling at scale is data-supply rights and regulation, never the unit economics.
Scaling inputs: ARPA $804/mo · 1,200 leads/buyer/mo · 80/20 std/premium mix · discovery layer steps down across volume tiers · carrier-verification a flat fraction-of-a-cent/dip · identity-resolution layer collapses from a few cents toward a fraction of a cent by volume · infra $130→$600→$3K→$25K→$180K→$900K/mo · payment processing ~mid-single-digit % · fully-loaded head $180K/yr. Inputs validated June 2026.
The moat isn't the scraping (anyone can scrape). It's the outcome-labeled, cross-vertical, time-series intent graph that only exists because HeatSeak sits in the loop between free public signal and buyer revenue. Every lead delivered makes the next one smarter and cheaper. COGS falls, accuracy rises, the gap widens with scale.
Fuse → ship → capture outcome → train → own graph → add verticals → richer fusion. Tighter every turn.
Live discovery + open-data intent (311/DOH/DOB/NOAA) + people-search, unioned on time+geo+entity. The floor is free public data + a sub-cent verification dip, a structural cost moat over anyone buying raw lists.
Did the lead convert? Won / lost / bad-number / contacted-not-closed, captured via dashboard, GHL webhook, or Brooke disposition. The one input you cannot buy; it only exists because we're in the delivery loop.
Labeled outcomes turn "score" into a model: this signal combo converts 14% for pest, 3% for PI. We deliver ranked probability-of-conversion leads competitors can't match without our outcome data.
Every contact fused is stored once, reused forever, a ~dime identity resolution becomes a permanent asset, not a recurring cost. Time-series intent (complaint last quarter + permit this quarter) is impossible to replicate cold.
Each vertical adds public sources that cross-pollinate, a NOAA hail event feeds roofing AND restoration AND PI. Adding vertical N+1 enriches verticals 1…N for free.
Which leads buyers open, call first, and re-buy is itself signal. The model spends paid enrichment only where predicted value × close-rate clears the cost, margin protection improves as it sharpens.
At scale the outcome-labeled intent graph is a product, licensed via API, sold as enrichment, fed into partnerships. We flip from consuming data to being a data source. The Apollo-scale endgame.
More leads sold → more outcomes → better model → higher conversion → more buyers → more leads. A new entrant starts at zero outcomes and can never retroactively acquire ours.
Each vertical added enriches every existing one for ~zero marginal cost, so the Nth vertical is cheaper and more accurate than the first. Breadth compounds depth.
COGS falls with volume, the discovery and identity-resolution layers step into volume tiers, infrastructure amortizing thinner. We get cheaper precisely as we get bigger.
Contacts and resolved identities are bought once and owned forever. Marginal cost of re-serving a known entity trends to zero while its enrichment depth grows.
Buyers who feed outcomes back get scoring tuned to their close rates, leaving means losing a model trained on their own results.
An outcome-labeled, cross-vertical, time-series intent graph, the proprietary, compounding moat that only exists because we sit in the loop.
The roadmap is how the flywheel gets built. Each stage seeds the next: outcomes corpus → wider sources → trained model + API → monetized graph → the category's intent layer. Every raise is justified by the previous stage's data moat, capital follows proof, never need. The ~89-93% margins mean every round is from strength.
One vertical (pest / local home-services), real buyers paying, engine proven to ship complete leads, qualified + intent + named owner + verified mobile.
Productize. Buyers self-serve, pick a vertical, top up credits, pull leads. No concierge bottleneck.
Turn accumulated outcomes into a defensible model. Stop selling lists, sell ranked probability-of-conversion leads. Open the platform via API.
Become the lead-intelligence platform, not just a lead seller. Enterprise buyers (call centers, brokers, networks) build on top of HeatSeak.
HeatSeak becomes the owned intent dataset the rest of the market builds on, where "who needs what, where, right now" is graded across every vertical. The Apollo-of-intent: others pay to query our signal.
How the pieces connect: Stage 1 seeds the outcome corpus → Stage 2 widens the signal sources and adds usage data → Stage 3 closes the loop with the trained model + API → Stage 4 monetizes the graph as first-party data → Stage 5 turns the graph into the category's intent layer. Capital follows proof, and the ~89-93% gross margins mean every raise is from strength, not need.
No moonshot required to begin. The whole thing starts with one vertical, our own engine, and a handful of paying buyers, and the unit economics are proven live from the first dollar. The endgame is the owned, multi-vertical intent dataset the rest of the market queries.
One vertical (pest / local home-services), a tight budget, and the fusion engine we already run. We don't buy a finished product, we manufacture the lead from public sources for under a cent, gate it behind the validation call, and sell it to a real buyer who re-orders.
The same engine, generalized across every vertical, fed by years of outcome labels nobody else holds. HeatSeak flips from data consumer to data source, the graded "who needs what, where, right now" layer others pay to query.
The lead-intelligence engine is built, live, and producing leads today. What's left is not core technology risk, it's product packaging: a self-serve web app, billing, an outcome-feedback loop, and buyer-facing polish. Think of it as: the factory works and ships product. We now need the storefront, the checkout, and the loading dock.
The expensive, risky half, the engine, is the half that's done. The remaining build is the known, de-risked kind: billing flows, a polished frontend, an API, and a feedback loop. Work any competent product team can scope and ship on a timeline.
Each stage lists what we build, the milestone that proves it, and approximate engineering effort. The capital and time from here buy packaging and distribution on top of a working, instrumented, cost-proven engine, the lowest-risk dollar in a data business.
The entire core, vertical-agnostic engine, intent overlays, fusion + scoring, self-optimizing sourcing fleet, enrichment/verification stack, 2,000/day machine, internal back-office, live demo. Typed prompt → graded clean leads, live, at low single-digit-cents per lead.
Buyer signup → checkout → credit/token billing → minimal "request leads / download / manage credits" surface, wrapping the existing engine. Buyer onboarding + a quality guarantee / replacement policy.
Production-grade buyer-facing app (clean UX, lead management, masking/reveal-on-pay), plus the delivered → contacted → converted instrumentation feeding back into scoring.
Public REST API, webhooks, Zapier, CRM push. Onboard additional verticals by wiring each one's public event source (permits, court filings, DOH/DOB, NOAA, 311, LSA-adjacent signals). Compliance/DNC plumbing scaled up.
Native/PWA app for buyer management, push notifications on fresh high-score leads, on-the-go credit top-ups.
Dozens of verticals live, marketplace-grade reliability, enterprise API tiers, white-label, the outcome loop tuned across verticals. Every new vertical is incremental config on a proven engine, not new core tech.
Why the remaining risk is product, not tech: the engine already does the thing the business sells, clean, scored, intent-backed leads at low cost, at scale (2,000/day proven). Costs are measured, not estimated: we have real per-unit cost-and-yield instrumentation and a fleet that auto-routes to the cheapest productive sources. Everything left is well-trodden product work, Stripe-style billing, a clean frontend, a REST API, a mobile shell, and a feedback loop. Known patterns, predictable timelines. The capital from here buys packaging and distribution on a working machine.
People check everything, so here is the honest classification of every load-bearing number on this page. Three tiers: VERIFIED (web-checked unit prices & public market figures), MODELED (our assumptions, stated on screen, projections, not promises), and ESTIMATE (directional, attributed). The honesty is the point: if a number is an assumption, we label it an assumption.
What it means for us: the cost-per-lead floor and the market context are real, sourced facts, the inputs the whole model stands on are not invented.
What it means for us: the scaling curve and margin are math that follows IF the assumptions hold. They're illustrative, every input is on screen so you can stress them yourself in the calculators above.
What it means for us: we use these for context and sizing only. We do not anchor any hard claim (margin, price, ROI) on them, they're the backdrop, not the proof.
The honest bottom line: the floor (cost-per-lead under a cent) is verified fact, sourced unit prices you can re-check today. The ceiling (the scale curve and 93% margin) is a model that holds only if our stated ARPA + volume + cost-mix assumptions hold, which is exactly why every input is wired into the calculators above, for you to pressure-test. We're showing you the assumptions instead of hiding them, that's the credibility.
Tiering rule: a number is Verified only if it's a web-checked unit price or a public-record market figure (June 2026). Anything dependent on ARPA/volume/mix is Modeled and labeled illustrative. Sizing figures from third parties (incl. vendor self-estimates like Angi's TAM) are Estimate and never anchor a hard claim.
A model is only as honest as its risk section. Here is what can break this business, named plainly, and the one-line plan that contains each. Several of these we've already caught in the wild (the verification layer was 5× our old anchor; an identity-resolution provider went dead once), which is exactly why the architecture is built to absorb them.
We don't own the upstream. We already caught the verification layer being 5× our old anchor, and an identity-resolution provider went dead once, a single vendor change can move COGS or cut supply overnight.
The verification dip is a flat per-dip rate that doesn't fall with volume. At ~50M leads it becomes roughly 68% of the bill, the one line that doesn't enjoy economies of scale.
People-search reuse, public-records reuse, and data licensing all carry terms-of-service and provenance exposure if treated carelessly.
The buyer contacts the lead, so contact-compliance risk sits with them, but if our data makes compliance hard, that's our problem too.
A name+city identity match can resolve to the wrong person; bad numbers and "never been a customer" matches burn buyer trust and trigger refund demands.
Contact data and intent decay fast, a stale lead is a dead lead, and freshness is what justifies the price.
Early revenue can lean on a few big buyers, losing one dents MRR and skews the model.
Anyone can buy the same APIs we do, the raw inputs are not the moat, and a well-funded copycat could rent the same pipes.
Single-founder concentration early on, plus real infra scaling breakpoints (single box → queue → warehouse) that bite if hit unprepared.
The point of this register: none of these risks are fatal, but pretending they don't exist would be. The two we've already lived through (a 5× vendor price move, a dead data source) are the reason the engine is multi-provider, instrumented, and provenance-labeled by design. The structural answer to almost every row above is the same: own the data flywheel, it's what turns rented APIs into a moat, scattered buyers into a base, and a lead seller into a data platform.
Investors and buyers will raise these objections. We raise them ourselves first, with the honest answer: not spin, not a dismiss, not a pivot. Where the answer is "not yet", we say so, and say exactly what closes the gap.
Our leads carry a why-now signal, a violation, a storm-hit, a permit, that Angi, HomeAdvisor, Thumbtack, and Google LSA cannot match. That signal is in public records; competitors can get it, but none have built the pipeline.
Sub-cent per lead, with real per-unit cost instrumentation and a self-optimizing sourcing fleet. This is not a projection, it is a measured number from a running system at ~2,000/day.
Change a prompt, change a vertical. Pest → roofing → legal → auto repair. The same engine, the same economics. Every new vertical is config, not core-tech spend.
The outcome-feedback loop, once live, creates a moat that compounds. Every closed deal sharpens the model; competitors starting cold can't shortcut this. And it hasn't shipped yet, which means the moat is ahead of us, not behind.
Explore/exploit bandit auto-prioritizes best+cheapest avenues. The machine gets cheaper to run as it learns, a cost structure that shrinks over time, not one that inflates with scale.
~$26 CAC vs ~$200-600 for paid ads. Performance-pay-per-view: misses are nearly free. If it proves out at scale, it is a structural acquisition cost advantage over competitors who must buy ads.
The engine is real, running, and cheap. The business risk is distribution (does the Clippers model actually produce sub-$50 CAC at scale?) and time-to-close (the remaining product packaging opens the self-serve funnel). Neither is a technology bet, both are execution bets. The intent-backed lead thesis is differentiated versus everything in this space. The outcome-feedback loop, once live, is a genuine compounding moat. The risk-adjusted case is strong: the expensive, unknowable half is done. We are asking capital to buy the packaging.
The standard we hold ourselves to: every projection in this deck is labeled. Where we say "modeled" we are telling you the assumption is unverified at scale. The data that matters most, unit cost and throughput, is measured and reproducible. We will show you the live runs.
| Metric | Value |
|---|---|
| Cost to produce 1 standard verified lead | under a cent |
| Cost to produce 1 premium identity-resolved lead | ~a dime |
| Blended cost / lead (80/20 mix) | a few cents |
| Sell price · home-services / pest | $22-25 → 99.96% GM |
| Sell price · PI / legal | $300 → 99.96% GM |
| Credit model | 1 std · premium 3 · PI 10 · B2B 2 (1 cr ≈ $1) |
| Packages | $199 / $499 / $1,499 / $4,999 · $0.80→$0.50/cr |
| Buyer ROI | 1.3×-6× (PI: 1 case = 26+ leads) |
| HeatSeak LTV : CAC | 37:1 · payback <3 weeks |
| P&L @ 50K leads/mo | $32.5K rev → $29K GP → 89% GM |
The CFO's bottom line: input cost is so far below market price that COGS is not a pricing constraint, buyer ROI is. Price to the buyer's closed-job economics, gate by credit tier, discount on volume to drive expansion, and the business runs at ~89-99% gross margin with sub-month CAC payback. Every number on this page ships with a default and a range, wired straight into the calculators above.
Inputs & ranges: discovery layer a fraction of a cent/query (steps down at scale) · carrier verification a flat fraction-of-a-cent/dip · identity-resolution a few cents to ~a dime (collapses at volume) · infra $130/mo (65-190) · volume 60K/mo (10K-500K) · mix std/premium 80/20 · churn 5% (3-8) · CAC $400 (250-800) · ARPA $804 (700-950) · blended $/credit $0.65 (0.50-1.00). All inputs validated June 2026.