The Lean Roofing Sales Tech Stack for a New Company: A Layer-by-Layer Toolkit

When you start a roofing company, the software pitch hits you before the first job does. Every CRM vendor, lead seller, measurement platform, and door-knocking app wants a credit card and a multi-year commitment, and every one of them will tell you it's the thing that finally makes you scale. Most of them are selling you a stack built for a 40-truck operation while you're running two crews and a magnetic sign on your truck.

I've stood up the back office for new roofing operations and watched plenty more get strangled by software they couldn't feed. The pattern is always the same. The owner buys a beautiful, expensive platform with forty features, uses four of them badly, and quietly resents the subscription for a year before churning. Meanwhile the actual constraint — getting in front of the right homeowners and not dropping the ball between the knock and the check — never got solved, because the tool was never the bottleneck. The workflow was.

So let's build this the way an operator builds it: as a workflow, layer by layer, where each layer has a job to do and hands clean data to the next one. Data tells you which roofs are worth your time. Outreach (mail, canvass, calls) gets you to the door. A CRM catches the lead so it doesn't die in a text thread. Production and measurement turn a sold job into a built roof. Supplements and claim documentation recover the money you earned. Analytics tells you which of all that actually paid. Get the layers and the handoffs right, and the specific tool names become swappable — which is exactly what you want when you're new and cash is the scarcest thing you own.

How to think about "lean" before you buy anything

Lean does not mean cheap. Lean means every dollar of software is tied to a workflow you actually run today, and you can show the line from that tool to either a sold job or saved labor. A $300/month tool that books you two extra jobs a month is lean. A $49/month tool nobody opens is bloat, full stop.

Three rules I'd tattoo on a new owner's forearm:

1. Buy for the bottleneck, not the dream. Your bottleneck in year one is almost never "we don't have enough features." It's "we don't know which doors to knock," or "leads fall through the cracks," or "we left money on the table on the last ten jobs." Buy the layer that fixes the bottleneck that's actively costing you, and leave the rest as a spreadsheet until it hurts.

2. One source of truth, or you've already lost. The single most expensive mistake new companies make is letting customer data live in five places — a notes app, a canvasser's phone, the estimator's email, a measurement vendor's portal, and a CRM nobody updates. The day you have two contradictory records for one homeowner, your follow-up dies. Pick the system that will own the customer record early, even if it's barely more than a glorified contact list, and route everything through it.

3. Integrations are a feature you pay for in time, not money. Two tools that sync are worth more than two better tools that don't, because the gap between them is filled by a human re-typing data — and that human forgets, fat-fingers, and quits. When you evaluate any layer, the real question isn't "is this the best canvassing app," it's "does this canvassing app drop a clean lead into the system that owns my customer record without anyone re-keying it."

A practical budget frame for a new shop: expect to spend somewhere in the low hundreds of dollars a month total across software in year one, scaling with revenue — not the four-figure monthly stacks the enterprise vendors quote. You can run a real, competitive operation on a lean stack. You just have to be honest about which layers you're actually using.

The rest is organized by layer. For each one I'll cover what the layer does, the real tool categories and named examples that fill it, how it hands data to the next layer, a lean-versus-scaled version, and the integration mistakes that quietly cost new companies money.

The workflow, in one breath

Before the layers, here's the whole motion so you can see where each tool plugs in:

Targeting data tells you which roofs are due → outreach (mail, canvass, or your old customer list) gets a conversation started → that conversation becomes a lead in your CRM → a sold lead becomes a measured, estimated, scheduled job in production → after the build (or after the storm), supplement and claim documentation recover the full scope you're owed → analytics closes the loop by telling you cost-per-win so you can spend the next dollar smarter.

Every layer is a handoff. The tools matter less than whether the baton gets dropped. Let's go layer by layer.

Layer 1: Targeting data — deciding which roofs are worth your time

This is the layer almost nobody builds first, and it's the one with the highest leverage for a new company, because it determines whether every downstream dollar — mail, gas, canvasser payroll — lands on a roof that might actually buy or a roof that got replaced two years ago.

What this layer does

The targeting layer answers one question before you spend a cent on outreach: which specific houses, in which order, are most likely to need a roof soon? The honest version of "likely" is a blend of two things you can actually estimate — how old the roof probably is, and how much weather it's taken. Neither is a crystal ball, and any vendor who tells you they know a roof's exact age or that a forecast "proves" damage is selling you certainty that doesn't exist. Roof age is a range. A storm is odds, not proof. But a ranked list built on age-band and storm exposure beats knocking a street blind every single day.

The tool categories and named examples

Property/parcel data and "year built" sources. Public records, county assessor data, and consumer real-estate sites (Zillow, Redfin) give you year the house was built. Useful, but understand the trap: year built is not roof age. A 1985 house that was re-roofed in 2019 has a five-year-old roof, and the public record doesn't see the re-roof. New owners burn real money mailing 40-year-old houses with brand-new roofs because they treated "year built" as "roof age." Treat parcel data as a coarse first filter, not a targeting decision.

Measurement platforms — EagleView, Roofr, HOVER, GAF QuickMeasure. Important category, wrong layer. These tools tell you the dimensions of a roof — squares, pitch, facets — not whether it's worn out or which house to approach. They answer "measure this house," not "which house." You'll absolutely want one in the production layer (more below), but don't confuse a measurement subscription for a targeting tool. They measure the roof; they don't rank the street.

Storm and hail mapping — Interactive Hail Maps, HailTrace, and the underlying public data from NOAA's Storm Prediction Center and the National Weather Service. A hail map shows you where it hailed. That's the start of targeting, not the end. Where it hailed is a big polygon; which roofs inside that polygon actually got worn out depends on roof age, exposure, slope, and material. The Insurance Institute for Business & Home Safety (IBHS) has solid public research on how hail and wind actually degrade roofing materials, and it's worth reading so you stop treating every hail swath as a uniform opportunity.

Roof-age and opportunity ranking — this is the category that fuses the two signals (age-band + storm exposure) into a per-home ranked list, which is what you actually act on. RoofPredict lives here: it scores roofs in an area by an estimated age band and the storms each roof has taken, and ranks which homes look due, house by house, so you mail and knock the right doors and skip the new ones. Be clear-eyed about the limits — it's a heuristic ranking off aerial imagery and weather history, not a guarantee any given roof will buy, and the age it returns is a band, not an exact install date. Used honestly, it's a prioritization layer, not a lead service; it sharpens outreach you were already going to do. There are also in-house and analyst-built versions of this (some larger restoration shops build their own scoring off purchased data), but for a new company a ranking tool is the buy-versus-build call where buying almost always wins, because building it means buying the data anyway.

How it hands data to the next layer

The output of a good targeting layer is a list: addresses, ranked, ideally with the reason attached ("roof age band 18–22 years, two hail events since install"). That list is the input to your outreach layer — it's the mail file, the canvass route, the call list. The handoff that matters: can you export that ranked list as a clean address file (CSV) that your mail vendor and your canvassing app can both ingest without manual cleanup? If the targeting tool traps the list in a portal you can't export, it just became a dead end.

Lean vs. scaled

• Lean: Start with a free/cheap storm map plus public parcel data to get rough boundaries, and layer in one ranking tool for the neighborhoods you're actually working this month. Don't buy national coverage you can't knock.
• Scaled: Continuous ranking across multiple markets, automated refresh after each storm, and the ranked list piped straight into mail automation and canvass routing.

The integration mistakes

• Treating year built as roof age (covered — the single most common one).
• Buying a measurement subscription and calling it targeting.
• Mailing the whole storm polygon instead of the aged roofs inside it, then blaming the mail when response is weak.
• Trapping the ranked list where you can't export it.

Layer 2: Outreach — getting to the conversation

Once you know which doors, you have to actually get to them. For a new residential roofer there are three outreach motions worth running, and the lean move is to run one or two well, not all three badly. The three are direct mail, canvassing (door-to-door), and working your own past-customer/old-estimate list. Paid digital leads are a fourth, and I'll be honest about why I put them last.

2a. Direct mail

What it does. Mail puts a branded, physical touch on a ranked list of homes. For roofing it still works because the buying trigger (an aging or storm-hit roof) is tied to a physical address, and mail hits the address whether or not the homeowner is searching online yet. Mail is most effective when it's aimed at a ranked list rather than a blanket carrier route — which is exactly why the targeting layer feeds it.

Tool categories and examples. Print-and-mail services and EDDM (Every Door Direct Mail through USPS) for blanket coverage; targeted variable-data mail houses for ranked address lists; and integrated platforms that take your address file, print, mail, and ideally track delivery. The USPS publishes the EDDM program and Informed Delivery (which can surface your mailpiece as an email preview to the household) — both are documented on usps.com and worth understanding before you pick a vendor. Some targeting platforms, RoofPredict among them, run the mail for you off the ranked list and track per-piece delivery so you know what landed, which closes a gap most new companies have: they mail blind and never know what hit the doormat versus what the post office ate.

Handoff to the next layer. Mail's job is to generate inbound calls, QR scans, and microsite visits — every one of which has to land as a lead in your CRM with the campaign attached, or you'll never know which mail drop paid. Use a tracked phone number and a unique QR/URL per campaign so the response is attributable. The most common mail mistake isn't the creative; it's that the call comes in, gets handled on a cell phone, and never gets logged, so you can't connect the $1,800 mail drop to the $14,000 job it produced.

Lean vs. scaled. Lean: a few hundred targeted pieces a month to your ranked list, one tracked number, results logged by hand if you have to. Scaled: automated drops triggered after storms, delivery tracking, and per-piece attribution flowing into analytics.

2b. Canvassing (door-to-door)

What it does. Canvassing is still the highest-conversion, lowest-cash-cost channel for a new roofer — it trades payroll for ad spend. A rep at the door with a reason to be there outconverts almost any digital touch. The hard part isn't knocking; it's enabling the rep, tracking the territory so two people don't knock the same street, and capturing the lead before the rep's phone battery dies.

Tool categories and examples. Dedicated roofing/home-services canvassing apps — SalesRabbit, SPOTIO, Sales Rabbit-style territory tools — handle route assignment, pin-dropping each door's disposition (not home, callback, sold, not interested), and lead capture in the field. Some general field-sales CRMs cover this too. The category job is: map the territory, hand reps routes, record every door, and push hot doors into your CRM.

Enablement is the real edge for a new company. Your reps are probably green. A green canvasser who knocks the right doors with a reason — "your roof's in the age range where we're seeing storm wear on this street" — and can show a branded per-home report or QR sounds like a veteran without climbing a ladder. This is where the targeting layer pays off twice: it tells the rep which doors, and it gives the rep something credible to say at each one. RoofPredict's per-home microsites and QR reports are built for exactly this hand-to-the-homeowner moment; you can also build a simpler version yourself with a branded one-pager. Either way, the point is the rep doesn't show up empty-handed.

Handoff. Every door disposition should sync to your CRM, and every sold or callback door should create a lead record automatically. The mistake: canvassing data lives in the canvassing app, CRM data lives in the CRM, and they never talk — so the office calls a homeowner the rep already sold, or nobody follows up the "come back Tuesday" doors. Pick a canvassing tool that syncs to your CRM, or pick a CRM that has canvassing built in, but do not run them as two islands.

Lean vs. scaled. Lean: one canvassing app, one or two reps, territories managed in the app so nobody double-knocks. Scaled: multiple crews, leaderboards, automated route optimization, and full two-way sync with production so a sold door becomes a scheduled inspection without re-keying.

2c. Your own list (past customers and old estimates)

What it does. This is the cheapest "new" business you'll ever get, and new companies ignore it because in year one the list is thin. But the moment you have it — every estimate you wrote and lost, every customer you served — it's money already in your book. Old estimates age into jobs (the roof you quoted two years ago that they "weren't ready for" is now two years closer to a leak). Past customers refer and re-roof outbuildings, rentals, and family.

Tool categories. This isn't a separate tool — it's a use of your CRM. Any CRM with a tag/segment and a follow-up cadence covers it. The discipline is what's lean here, not the software: a quarterly touch to old estimates and an annual touch to past customers costs almost nothing and converts because the relationship already exists.

Handoff. Lives entirely in the CRM (Layer 3). The only "integration" is remembering to tag leads with their source and outcome so you can re-mine the lost ones later.

2d. Paid digital leads — why they're last

I'm not telling you never to buy a lead. I'm telling you to understand what you're buying. Shared lead marketplaces (the Angi-style model) typically resell the same homeowner to multiple contractors, so you're in a price-and-speed race against three other trucks before you've said hello. The Federal Trade Commission has public guidance on lead-generation practices worth reading before you sign anything, and the lesson for a lean operator is: paid leads are a channel you rent, with no compounding asset — when you stop paying, it stops. Mail, canvass, and your own list build a book you own. For a new company guarding cash, own before you rent. If you do test paid leads, treat them as a measured experiment with a hard cost-per-acquisition cap, and track them in the same CRM as everything else so you can compare honestly.

Layer 3: The CRM — the system that owns the customer

If you remember one thing from all of this: the CRM is the spine. Every other layer either feeds it or reads from it. Get this layer wrong and the best targeting and the hardest canvassing in the world leak out the bottom.

What this layer does

The CRM owns the customer record and the pipeline. Every lead — no matter which outreach channel created it — lands here, moves through stages (new → contacted → inspected → estimated → sold → in production → closed), and never falls through the cracks because the system, not a person's memory, holds the next action. For a roofer the CRM also typically carries the estimate, the photos, the documents, and the job timeline, which is why the roofing-specific CRMs blur into the production layer (Layer 4) — a lot of them do both.

Tool categories and named examples

Roofing/home-services vertical CRMs — JobNimbus, AccuLynx, Roofr, Jobber, JobProgress. These are built for the trade: pipeline stages that match a roofing job, estimate and material lists, photo capture, production scheduling, and in several cases integrations to measurement and supplement tools. For most new roofing companies, a vertical CRM is the right call over a generic one because the workflow is already shaped like your business.

Field-service / general CRMs — ServiceTitan (powerful, built for high-volume service-and-replace operations, generally heavier and pricier than a brand-new shop needs day one), HubSpot (a genuinely capable free/low-tier CRM if you're comfortable configuring it yourself and want strong marketing automation), and others. ServiceTitan is excellent but it's a scaled-company tool; don't buy the enterprise platform to run two crews.

The lean honest take: a new company is usually best served by one mid-tier vertical CRM (JobNimbus and Jobber are common lean starting points; Roofr bundles measurement and proposals which consolidates two layers) or, if you're technical and budget-zero, HubSpot's free tier as a contact-and-pipeline spine you graduate off of later. The wrong move is buying ServiceTitan or AccuLynx's full suite before you have the volume to use it — you'll pay for production and reporting features that sit idle while you fight the onboarding.

Two-way sync is the whole game

Here's where new companies quietly lose money. Your CRM needs to talk to the rest of your stack, and the word that matters is two-way. A one-way sync that dumps leads in but never sends status back means your canvassing app, your measurement tool, and your supplement system all drift out of date, and someone re-keys the difference. Mature roofing stacks sync the CRM two-way with measurement (EagleView, HOVER), with field tools (SalesRabbit, CompanyCam), with production, and with supplement/claim systems — so a sold door, a measured roof, and a submitted supplement all reflect in one record. RoofPredict, for instance, syncs two-way to the common roofing CRMs (HubSpot, ServiceTitan, JobNimbus, AccuLynx, Jobber, Roofr, SalesRabbit, CompanyCam, and others) specifically so the targeting and outreach data doesn't become a sixth island. When you evaluate any tool in any layer, the very first question is: does it two-way sync with the CRM I'm standing up?

Handoff

The CRM hands a sold job to production (Layer 4) — ideally automatically, so "marked sold" triggers "schedule inspection / order measurement." And it feeds analytics (Layer 6) the source, stage, and dollar data that makes cost-per-win calculable. If your CRM can't tag a lead's source channel, you've broken analytics before you start.

Lean vs. scaled

• Lean: One vertical CRM, pipeline stages set up to match your real sales motion, every lead tagged with source. That's it. Resist configuring 30 automations you'll never maintain.
• Scaled: Two-way integrations across measurement, canvass, production, and supplements; automated stage triggers; role-based dashboards; and marketing automation for the re-engagement cadence.

Integration mistakes

• Running a canvassing app and a CRM that don't sync (the office and the field call the same homeowner).
• Not tagging lead source, so you can never compute which channel pays.
• Buying the enterprise CRM before you have the volume — onboarding eats the season.
• Letting estimates live in the estimator's email instead of the CRM, so lost estimates can never be re-mined.

Layer 4: Production and measurement — turning a sold job into a built roof

This layer is where a sold lead becomes a real roof, and where a new company's reputation gets made or broken. The sales stack's job here is mostly to hand off clean and then get out of the way of the crew — but two sub-layers, measurement and field documentation, are tightly coupled to sales and worth getting right early.

Measurement

What it does. Produces accurate roof dimensions (squares, facets, pitch, edge/ridge/valley lengths) for ordering material and writing a tight estimate. This is the layer where the measurement platforms I told you not to use for targeting finally earn their keep.

Tools. EagleView (aerial measurement reports, the long-standing standard), HOVER (photo-based 3D measurement and visualization the homeowner can see), Roofr (instant measurement bundled with proposals, increasingly popular with newer shops for the price point), GAF QuickMeasure. For a lean new company, pick one measurement source and standardize on it; the per-report cost is small relative to the material-order accuracy and the estimating time you save. The honest tradeoff: aerial measurement saves a ladder trip and is fast, but you still verify on the inspection — measurement is a planning tool, not a substitute for getting on the roof when it's safe to.

Handoff. Measurement feeds the estimate inside your CRM, and the estimate feeds the material order and the supplement baseline (Layer 5). The clean version: order the measurement from the CRM record so the squares land back on the job automatically rather than getting copy-pasted.

Field documentation and photos

What it does. Captures dated, geotagged photos of the roof, the damage, and the completed work — the evidence that supports your estimate, your supplement, and your warranty, and protects you in a dispute.

Tools. CompanyCam is the category leader for trade photo documentation (project-organized, timestamped, share-ready), and it integrates with most roofing CRMs. This is one of the highest-ROI lean purchases there is, because photo evidence underpins the supplement layer that recovers your money — and because it's the difference between "we said there was hail bruising" and "here are 40 dated, located photos of it."

Safety is part of the stack

Not software, but non-negotiable, and a new company that skips it pays in fines and injuries. OSHA's fall-protection standards govern roofing work, and OSHA fall protection is consistently one of the most-cited standards in construction. Build the safety checklist into your production workflow the same way you build the photo checklist — it's part of the operating system, even if it lives on a clipboard.

Lean vs. scaled

• Lean: One measurement source, CompanyCam for photos, a safety checklist, and material ordered off the verified measurement.
• Scaled: Measurement auto-ordered from the CRM, photo documentation syncing to the job record, production scheduling and crew dispatch, and material orders integrated with suppliers.

Layer 5: Supplements and claim documentation — recovering the money you earned

This is the layer that separates a shop that survives from a shop that just turns crews. On insurance-restoration work especially, the difference between the carrier's first estimate and the full, correct scope of work is real money — code-required items, missed line items, proper waste, access, and detail that the initial estimate left out. A new company that doesn't document its own scope rigorously leaves that money on the table on every claim job, and at thin margins that's the difference between profit and a busy, broke season.

The boundary you must understand first

Before any tool, understand the line, because crossing it is illegal in most states and will end your company. A contractor may document its own work: its own inspection, its own estimate and scope, its own photo and measurement evidence, and it may compare its own estimate to the carrier's estimate internally, build an evidence index, request missing documents, and track its own submitted paperwork. A contractor may not act as a public adjuster — it may not represent, negotiate, or handle the claim for the homeowner, interpret the homeowner's coverage or policy rights, tell the homeowner what they're entitled to recover, advise on settlement, advertise public adjusting without a license, or charge fees based on the claim proceeds. Many states have Unfair Public Adjuster statutes and licensing rules enforced by the state Department of Insurance (Texas's TDI is a well-documented example), and the NAIC publishes model guidance on public-adjuster licensing. The clean mental model: the homeowner files, the insurer decides, and you document your own scope. Also retire any "we'll waive your deductible" or "free roof" messaging — deductible waiver is illegal in many states and the FTC and state regulators treat "free roof" claims as deceptive. None of that is legal advice; check your own state's DOI rules.

What this layer does

Within that boundary, the supplement/claim-documentation layer does a lot of honest, valuable work: OCR and classify the documents (the carrier estimate, the policy declarations, the scope), map your estimate lines against a knowledge base of code items and proper scope, flag where your documented scope includes items the carrier estimate missed (with the photo evidence and pricing to back each one), score your packet for completeness before you submit, track recoverable depreciation and your own submitted paperwork, and manage the cadence so supplements don't age out. It's a documentation-and-QA engine for your scope, not a tool for handling someone else's claim.

Tool categories and named examples

Estimating/pricing standard — Xactimate (Verisk) is the pricing and estimating language the insurance side speaks; understanding Xactimate line items and pricing is how you document a scope that's legible to a carrier's process. Verisk publishes public documentation on Xactimate. You don't have to love it, but you have to speak it.

Code reference — the International Residential Code (IRC) from the ICC governs the code-required items (ice-and-water, ventilation, drip edge, decking) that frequently get left out of a first estimate. Knowing the code that applies in your jurisdiction is how you document a legitimate scope item rather than a wish-list one.

Supplement / claim-documentation platforms and services — there are dedicated supplement tools and outsourced supplement services in the market. The category job is: take your documented scope and the carrier estimate, surface the documented gaps, and organize the evidence packet. RoofPredict's RoofClaim / RCM sits in this layer on the contractor-documentation side: claim intake, document OCR and classification, opportunity detection that maps your estimate lines to a knowledge base and flags missing scope or code items with evidence and pricing, recoverable-depreciation tracking, deductible tracking, supplement aging and cadence, and packet-completeness scoring — all on locked, UPPA-gated, contractor-documentation-only templates so it stays on the legal side of the line. Its honest limit, like everything in this layer: it documents your scope and surfaces gaps for you to support; it does not file, negotiate, or handle the claim, and it doesn't tell the homeowner what they're owed.

Handoff

This layer reads from production (the measurement and the photos) and the CRM (the estimate and the customer record), and it feeds analytics (Layer 6) the supplement-capture data — how much of your documented scope actually got recovered, which is one of the most important numbers in a restoration business and the one new companies track least.

Lean vs. scaled

• Lean: Rigorous photo documentation, a working knowledge of Xactimate and your local IRC code items, a checklist so nothing gets missed, and a disciplined cadence on submitted paperwork. You can do real supplement work with discipline and a spreadsheet before you buy a platform.
• Scaled: OCR/classification, automated gap detection against a knowledge base, packet scoring, depreciation and deductible tracking, and aging cadence — the volume play when one estimator can't keep up with the claim flow.

Integration mistakes

• Crossing the public-adjuster line because nobody read the state DOI rules (the catastrophic one).
• Using deductible-waiver or "free roof" messaging (illegal and deceptive).
• Thin or undated photos, so documented scope items can't be supported.
• No cadence, so supplements age out and recoverable depreciation never gets collected.

Layer 6: Analytics — closing the loop on what actually paid

The last layer is the one that makes the first five smarter every season, and it's almost always built last (correctly) and skipped most (wrongly). You don't need a business-intelligence platform. You need to be able to answer three questions honestly.

The three questions

1. Cost per win, by channel. Take every dollar you spent on a channel (mail printing and postage, canvasser payroll, lead spend) and divide by the jobs that channel produced. This is why source-tagging in the CRM (Layer 3) is non-negotiable — without it, this number doesn't exist. New companies routinely keep spending on the channel that feels busy rather than the one that pays, because they never built this number.

2. Actual vs. estimate. Did the jobs you sold come in at the margin you estimated, or did production, supplements, and change orders erode it? A results funnel that compares estimated job value to actual collected — including supplement recovery — tells you whether you're pricing right and recovering right.

3. Pipeline conversion by stage. Where do leads die? If everything stalls at "inspected → estimated," your problem is estimating speed, not lead volume, and no amount of new targeting fixes it. Stage conversion tells you which layer to fix next.

Tools

Lean: Your CRM's built-in reporting plus a single spreadsheet you actually update weekly. That's genuinely enough for a new company, and it's better than an expensive dashboard nobody looks at. Scaled: A results/analytics layer that pulls from the CRM, the mail/canvass data, and the supplement system to compute cost-per-win and actual-vs-estimate automatically. Some platforms (RoofPredict's results funnel among them) close this loop on the targeting-and-outreach side specifically, tying a mailed or knocked address back to the job and the cost so you can see cost-per-win by source — which is the metric that tells you where the next dollar should go. Census County Business Patterns and BLS data on the construction sector are useful free public benchmarks if you want to sanity-check your market against the industry.

The mistake

The one analytics mistake that matters: not tagging source at the lead stage, which makes every downstream number a guess. Fix that on day one and the rest of analytics is just arithmetic.

A second, subtler mistake is measuring the wrong window. A roof you knocked or mailed today might not buy for eighteen months — the roof ages into a sale, or a storm finally pushes the homeowner over the line. If you judge a channel on a 60-day window you'll kill mail and your own-list re-engagement before they ever pay, because those channels compound slowly and quietly. Hold channels to a window that matches how the buying decision actually unfolds, and keep the cohort: tag a lead with the month and the source it came in on, so when it closes a year later you can still attribute it. New companies that don't keep cohorts end up chasing whatever felt hot last week and abandoning the slow channels that quietly produce their best margin jobs.

A worked example of cost-per-win

Make it concrete with illustrative numbers — these are examples to show the math, not benchmarks to expect. Say a month of targeted mail costs you roughly $1,800 all-in (print, postage, the ranked list) and produces, over the following few months, three sold jobs. That's about $600 per win. In the same month two canvassers cost you roughly $6,000 in payroll and produce eight sold jobs — about $750 per win, but with eight jobs instead of three and far more pipeline (callbacks, not-homes to circle back on) than the mail generated. Now you have a real decision: mail is cheaper per win but lower volume; canvass costs more per win but moves more roofs and builds a routed territory you keep. Neither is 'better' in the abstract — it depends on your cash, your crew, and your market. But you can only have that conversation if you tagged source and held the cohort long enough to count the late closers. Run that arithmetic every month and your spend gets smarter on its own.

The starter stack: a one-page table

Here's the whole thing on one page — what each layer does, the lean pick versus the scaled pick, and the one handoff that matters. Tool names are examples that fit the layer, not endorsements of one over another; standardize on whatever two-way syncs with the CRM you choose.

How to sequence the buy (because you can't do it all at once)

You're not buying seven things on day one. Sequence it to your bottleneck:

1. CRM first. Even a free spine. You cannot run the other layers without something that owns the customer record. Tag source from the first lead.
2. One outreach motion. Pick the channel that fits your cash and your team — canvass if you have legs and little cash, mail if you have a little cash and a ranked list, your old list the moment it exists. Run one well.
3. Targeting to feed the outreach. Add a ranking tool so the outreach you're already paying for lands on roofs that might actually buy. This is where lean stacks get their leverage.
4. Measurement + photos. The day you're selling enough to feel the estimating drag and you need evidence on jobs, add a measurement source and CompanyCam.
5. Supplement discipline. From your first claim job, document your own scope rigorously by hand; add a documentation platform when claim volume outruns a spreadsheet. Read your state DOI rules before you touch this layer.
6. Analytics last, by computing cost-per-win from data you've been tagging all along. If you tagged source from day one, this is a free afternoon, not a project.

The whole point of building it as a workflow instead of a shopping list is that you never buy a layer you can't feed, and you never let the baton drop between two layers because they don't sync. A lean stack isn't the cheapest stack — it's the one where every dollar traces to a job and every handoff is clean. Build it in that order and you'll spend a fraction of what the enterprise vendors quote while running an operation that actually compounds, because the work you generate is work you own.

FAQ

What's the minimum software a brand-new roofing company actually needs?

One CRM that owns your customer record and pipeline, one outreach motion you run well (canvassing, mail, or your own past-customer list), and a way to document jobs with dated photos. Everything else — targeting tools, measurement, supplement platforms, analytics dashboards — you add as the corresponding bottleneck starts costing you money. The mistake is buying a full enterprise suite before you have the volume to use it.

Is a roofing-specific CRM worth it, or can I start with a free general CRM?

Both work. A vertical CRM (JobNimbus, Jobber, Roofr) comes pre-shaped like a roofing job — roofing pipeline stages, estimates, photos, production scheduling — which saves setup time. A free general CRM like HubSpot is fine if you're technical, want to spend zero day one, and are willing to configure it yourself. The non-negotiable in either case is that you tag every lead's source and that the CRM two-way syncs with the rest of your stack.

Can I use Zillow or county records to find old roofs to target?

Only as a rough first filter. Public records and Zillow give you year the house was built, not roof age — they can't see a re-roof. A 40-year-old house that was re-roofed five years ago looks like a target and isn't, and mailing those homes burns money. Use parcel data to draw coarse boundaries, then rank roofs by estimated age band and storm exposure (which is what roof-age ranking tools like RoofPredict do) before you spend on outreach.

Are measurement tools like EagleView or HOVER a targeting tool?

No — that's a common and expensive confusion. Measurement platforms tell you the dimensions of a roof (squares, pitch, facets) so you can order material and write a tight estimate. They answer 'measure this house,' not 'which house should I approach.' You want one in your production layer, but it won't tell you which roofs on a street are due. Targeting and measurement are different layers.

Should a new roofing company buy shared digital leads?

Treat them as a last resort, not a foundation. Shared lead marketplaces typically resell the same homeowner to multiple contractors, so you're in a price-and-speed race the moment you call, and when you stop paying the channel produces nothing. Mail, canvassing, and your own past-customer list build a book of business you own and that compounds. If you test paid leads, cap your cost per acquisition hard and track them in the same CRM so you can compare them honestly against your owned channels.

What does 'two-way sync' mean and why does it matter for a lean stack?

Two-way sync means two tools keep each other updated — when a lead changes status in your CRM, the canvassing app or measurement tool reflects it, and vice versa. A one-way sync only pushes data in one direction, so the other system drifts out of date and a human re-keys the difference (and forgets, fat-fingers, and quits). For a lean company two synced tools beat two better tools that don't talk, because the gap between unsynced tools is filled by error-prone manual work.

Can my roofing company handle a homeowner's insurance claim or supplement?

You can document your own work — your inspection, estimate, scope, photo evidence — and compare your estimate to the carrier's internally, request missing documents, and track your own submitted paperwork. You cannot act as a public adjuster: you may not represent or negotiate the claim for the homeowner, interpret their coverage or policy rights, tell them what they're entitled to recover, or charge fees based on claim proceeds, and you must avoid deductible-waiver or 'free roof' messaging. Many states license public adjusters and enforce these rules through the Department of Insurance — check your state's specific rules. This isn't legal advice.

Do I need supplement software when I'm just starting out?

Not at first. On your early claim jobs you can do legitimate supplement work with discipline alone: rigorous dated photos, working knowledge of Xactimate pricing and the IRC code items that get left out of first estimates, a completeness checklist, and a cadence so nothing ages out. Add a documentation platform (OCR, gap detection, packet scoring) when claim volume outruns what one person can track in a spreadsheet — and keep everything on the contractor-documentation side of the public-adjuster line.

How much should a lean roofing tech stack cost per month?

There's no single number, but a new shop can run a real, competitive operation for a fraction of the four-figure monthly stacks enterprise vendors quote — expect to scale software spend with revenue rather than front-loading it. The discipline that keeps it lean isn't picking the cheapest tools; it's only paying for a layer once you can trace its cost to a sold job or saved labor. A $300/month tool that books two extra jobs is lean; a $49/month tool nobody opens is bloat.

What's the right order to buy the layers in?

CRM first (even a free one), so you have something that owns the customer record and tags lead source. Then one outreach motion you can run well. Then a targeting/ranking tool so that outreach lands on roofs that might actually buy. Then measurement plus photo documentation when estimating drag and job evidence start to matter. Then supplement discipline from your first claim job. Analytics last — if you tagged source from day one, computing cost-per-win is an afternoon, not a project.