Best Roofing Direct Mail: Targeting, EDDM, Costs, Creative, and ROI

Does direct mail work for roofing companies?

Yes, when the campaign creates enough incremental signed contracts to cover the mail, the inspections it creates, and the sales capacity it consumes. That answer is deliberately different from “people called.” A high response rate can still lose money if callers are outside the service area, appointments do not hold, or the team spends hours inspecting properties that never become appropriate work. A modest response rate can be excellent if the responses become well-matched, high-margin jobs.

Before comparing a list, postcard, or vendor, make every proposal use the same funnel:

Metric	Exact definition	Denominator	Why it matters
Delivered mail	Mailers accepted as delivered or not returned	Mailers sent	Separates production from reachable inventory
Response rate	Unique responding properties	Delivered properties	Measures attention and initial interest
Booking rate	Properties with an inspection booked	Delivered properties	Tests whether the offer produces an action
Completion rate	Properties with an inspection completed	Delivered properties	Exposes no-shows and capacity failure
Contract yield	Properties signing a contract in the attribution window	Delivered properties	Connects mail to the commercial outcome
Variable acquisition cost	Outreach plus incremental inspection and sales cost	Signed contracts	Shows the cost of creating each contract
Campaign contribution	Contract gross profit minus outreach and incremental inspection/sales cost	Campaign	Shows what remains before fixed overhead and taxes

“Signed revenue” is not contribution, and contribution is not EBITDA. That distinction matters in roofing because a $15,000 contract does not contribute $15,000 toward office payroll, fixed fleet cost, debt service, or owner profit. Materials, field labor, commissions, and other job costs come first.

The direct-mail break-even equation

Use your own inputs:

break-even contracts = (mail cost + incremental inspection/sales cost) ÷ gross profit per signed contract

Then divide break-even contracts by delivered mailers to get break-even contract yield. If 20,000 delivered mailers cost $18,000 all-in before inspections, each completed incremental inspection costs $110, the campaign creates 120 incremental inspections, and gross profit per contract is $4,500, total acquisition cost is $31,200. The campaign needs 6.94 contracts to break even, so round up to seven. Its break-even contract yield is 7 ÷ 20,000, or 0.035%.

That worked example is not a PorchRocket benchmark. It shows why the correct threshold is specific to your print/postage, gross margin, inspection burden, and sales model. A vendor that calculates return using only postcards and signed revenue is leaving out the two inputs most likely to make the answer look worse.

Decision rule: approve a pilot only after you can state the outcome, denominator, attribution window, gross-profit assumption, capacity limit, and stop condition in one paragraph.

Which roofing companies benefit most from targeted direct mail?

Targeted direct mail fits a residential roofer with a defined service area, positive job-level economics, unused response and inspection capacity, and a truthful property-relevant reason to make contact. It can work for an owner-led shop or a multi-branch operator; company size is not the deciding variable. The useful question is whether better address selection can put scarce marketing and sales capacity in front of more credible work.

The channel is strongest when the contractor wants to earn homeowner attention before that homeowner starts comparing roofers in an online auction. It is weak when the company cannot answer the response, does not know what work it wants, or needs a model to manufacture urgency. Mail makes a promise in the company’s name. The operation behind the phone number has to keep it.

Roofing-company situation	Sensible first use	Primary decision metric	Main reason to wait
Owner-led startup with a small service radius	One staged local cohort built around a credible inspection or planning offer	Qualified completed inspections and conservative contribution	Owner already misses calls or cannot absorb the inspections
Growing retail-replacement company	Selected properties released against weekly appointment capacity	Incremental contracts and contribution per delivered mailer	Service area, gross profit, or source tracking is still undefined
Established local brand with an installed base	Selected net-new mail plus separately governed customer/estimate reactivation	Incremental contribution and installed-base/neighbor response by cohort	Old customers and unknown households are being mixed into one generic list
Proven operator entering a new market	Small market-specific cohort with local proof and dedicated response tracking	Qualified booking, contract yield, and contribution versus a local holdout	Home-market response assumptions are being treated as a forecast
Restoration-capable local roofer	Preparedness or verified-event sequence paced to safe inspection and production capacity	Incremental completed inspections, contracts, complaints, and backlog	Copy implies damage or coverage, or the company cannot sustain post-event service
Repair/service-led roofer	Narrow lifecycle or service offer where the ticket and repeat value support acquisition	Collected contribution by service line, including dispatch cost	Replacement economics are being used to justify small-ticket leads
Sparse commercial or large-loss seller	Usually an account-based plan, not household mail	Qualified account progression	Residential address and roof-age logic does not identify the actual buyer

Owner-led and newly launched roofers

A new roofing company can use mail, but it has to underwrite the trust deficit and the owner’s calendar. A recognizable local incumbent may receive a response because the household has seen its trucks, reviews, installations, or signs. A new company has to earn that confidence on the mailer and at the response path: accurate business identity, licenses or registrations where required, proof it can substantiate, a professional phone experience, a useful visit, and no borrowed social proof.

The best first cohort is rarely the biggest one the printer will discount. Start with one serviceable geography and one job type. Reserve actual inspection slots before the drop. Route calls that the owner cannot take. If ten qualified responses arriving on the same day would create a service failure, the company is not ready to mail a volume designed to create twenty.

Mail can still be attractive for a new roofer because it lets the owner choose a bounded audience and build recognition under the company’s own name. The tradeoff is that early conversion has more uncertainty. Use a lower response assumption, a conservative job margin, and a staged release. A startup does not become established by sending 50,000 mailers; it becomes credible by keeping the promises made to the first 1,000 households.

Growing retail-replacement companies

This is the clearest fit. The contractor knows its normal replacement ticket, job gross margin, inspection cost, close process, and production limits. The future inspection calendar is thinner than the business wants, but the operation can serve more booked work. In that setting, property selection can reduce money spent on poor-match homes and reduce estimator time consumed by the wrong work.

The constraint matters. If the company has six estimators but only forty open inspection slots in the next month, a list that is expected to create 200 completed inspections is not efficient. Raise the selection threshold, stage the mail, or reserve the next cohort. The value of prequalification is not merely fewer stamps. It is more gross-profit opportunity per scarce call, appointment, drive, and follow-up hour.

Established brands and installed-base operators

An established company has two distinct assets: local recognition and a history of known properties. Use them differently. Net-new targeted mail can carry the familiar brand into selected households. Prior customers, repairs, estimates, no-sales, and warranty relationships belong in a history-aware CRM reactivation program, with appropriate channel eligibility and exclusion.

Nearby completed work can also support a truthful installed-job halo: the contractor is active in the area and can offer a neighborhood availability window. That is not permission to name a customer, publish an address, imply an endorsement, or infer that every nearby roof needs work. Recognition is the mechanism; manufactured social pressure is not.

New-market and multi-branch expansion

A proven campaign is not automatically portable. Brand awareness, housing stock, roof type, weather history, permitting records, mail delivery, competitive density, homeowner expectations, call staffing, inspection travel, close behavior, and job mix can all change in a new market. Preserve the operating method: serviceability, exclusions, truthful evidence, holdout, source tracking, and contribution. Treat the market’s response and economics as new parameters.

Multi-branch companies need an additional control: one property, one current communication owner. Branch boundaries, recent contacts, active work, and source rules must reconcile before selection. Otherwise a centralized campaign can mail a homeowner who is already waiting on a branch estimate, or two branches can claim the same response. The mailing file should expose branch assignment and conflict resolution before production, not after commissions are disputed.

Restoration-heavy and event-driven roofers

Mail can prepare a market or support verified post-event outreach, but event work increases the cost of a bad assumption. A modeled footprint is not a roof inspection. An inspection is not a claim decision. A contract is not carrier approval. The campaign should preserve those boundaries in its list reason, creative, call script, and reporting.

Preparedness mail can offer documentation of current condition, emergency contact information, or a planning inspection that remains useful if no event occurs. Post-event mail can name a verified area event at the source’s actual level of confidence and offer a qualified inspection. The roofing storm-response guide owns the evidence ladder, activation timing, safety, and surge-capacity gates.

When targeted mail should not be the next move

Do not mail because the sales meeting feels quiet. First fix the constraint if:

• callers routinely reach voicemail and legitimate callbacks take hours;
• current leads lack a next action or confirmed appointment;
• the CRM contains a large, eligible historical audience nobody is working;
• inspection, estimating, production, or collections are already outside service levels;
• the business cannot state gross profit by the job type being promoted;
• the target area is outside the company’s license, service, crew, supplier, or warranty reach;
• the only “reason” for contact is an unsupported damage, coverage, or replacement claim;
• list sources, field dates, permissible use, exclusions, or outcome exports cannot be explained.

In those conditions, direct mail can amplify a leak rather than create a growth program. Use call coverage, CRM reactivation, or the lead-to-cash operating guide to repair the first broken transition. Then return to the mail decision with capacity and economics that can survive a successful response.

What did PorchRocket’s 128,000-mailer field experiment find?

In this field experiment, property selection produced the largest difference, and property-relevant personalization produced additional lift inside the selected group. The strongest arm recorded a 0.750% signed-contract yield; the broad generic arm recorded 0.109%. Those are study results, not a forecast for every roofer or market.

The study began with a 144,000-address universe. A total of 128,000 addresses were mailed in four arms of 32,000 each, and 16,000 addresses were held out. The design crossed two list strategies, higher-opportunity property selection and broad selection, with two creative strategies, generic and property-relevant. Assignment of personalization within the risk stratum was randomized. Selection into the higher-opportunity stratum was operational, not randomized, so differences between that stratum and the broad list include the selection system and may also include underlying population differences.

Results through signed contract

Mailing arm	Response rate	Booking rate	Completed-inspection rate	Signed contracts	Contract yield
Higher-opportunity, property-relevant	4.65%	3.30%	2.97%	240 / 32,000	0.750%
Higher-opportunity, generic	2.87%	Not reported	Not reported	125 / 32,000	0.391%
Broad, generic	1.10%	Not reported	Not reported	35 / 32,000	0.109%
Broad, property-relevant	1.55%	Not reported	Not reported	55 / 32,000	0.172%

The em dashes mean the abbreviated case-study table did not preserve a value for that cell; they do not mean zero. PorchRocket reports only observed fields available from the study record rather than reconstructing intermediate counts after the fact.

Within the higher-opportunity stratum, where personalization was randomized, property-relevant mail produced a response relative risk of 1.62 (95% confidence interval 1.50–1.76) compared with generic creative. Its signed-contract relative risk was 1.92 (95% CI 1.55–2.38). In ordinary language, both response and contract outcomes were materially higher in the personalized arm during this study, and the intervals quantify sampling uncertainty around those comparisons.

The higher-opportunity personalized arm versus the broad generic arm recorded a response relative risk of approximately 4.23 and a signed-contract relative risk of 6.86 (95% CI 4.81–9.77). That contrast is operationally important but not a clean randomized estimate of targeting alone because list membership was selected. The responsible conclusion is “the complete selected-and-personalized strategy performed much better here,” not “this scoring model causes a 6.86× result everywhere.”

Holdouts provided a second reality check. Higher-opportunity holdout properties recorded a 0.15% contract yield, while general holdouts recorded 0.05%. Organic activity was therefore not zero. Reporting mailed contracts without holdouts would overstate incrementality, especially when a storm, local advertising, referrals, or an existing brand can create conversions without the mailer.

Campaign economics, with the assumptions visible

The economic model assumed:

• average signed contract value: $15,000;
• job gross margin: 35%;
• gross profit per signed contract: $5,250;
• campaign contribution: gross profit minus outreach and incremental inspection/sales cost;
• fixed overhead, taxes, debt service, and owner compensation excluded.

Under those assumptions, the higher-opportunity personalized campaign generated $1,260,000 of modeled contract gross profit across 240 contracts. After $188,530 in campaign and incremental selling expense, its modeled campaign contribution was $1,071,470. Its variable acquisition cost was $786 per signed contract after rounding.

The broad generic arm generated 35 signed contracts. Its variable acquisition cost was $1,456 per contract, and its modeled contribution was approximately $132,790. The difference between those two campaign contributions was $938,680 in this experiment.

The study also ran a conservative sensitivity rather than pretending the base economics were fixed. At a $10,000 average signed contract and 30% gross margin, $3,000 gross profit per contract, the higher-opportunity personalized arm still modeled $531,470 in campaign contribution after the same campaign cost. Under the base assumptions, its modeled break-even contract yield was approximately 0.112%.

Replace every economic input before using those numbers. If your mix includes repairs, if commissions sit outside job gross margin, if inspection costs are higher, or if cancellations differ by source, the result changes. Signed contracts that cancel or never collect should not survive into a mature contribution report.

What this experiment establishes and what it does not

The experiment supports four useful conclusions:

1. List strategy and creative strategy should be measured separately.
2. A property-relevant message can beat generic creative within a selected audience.
3. Response rate alone understates the importance of downstream conversion and cost.
4. Holdouts matter because some properties convert without mail.

It does not establish that a remote score diagnoses roof condition, that every selected property needs replacement, that a homeowner has covered damage, or that another market will reproduce these yields. Local brand strength, weather, list quality, creative, season, offer, appointment capacity, sales execution, price, and job mix all change the result. The right use of the study is to design a better pilot and require better reporting, not to paste its yield into a forecast as a guarantee.

Targeted roofing mail or USPS EDDM: which should you use?

Use targeted mail when address-level evidence changes who should receive the mailer or what the mailer should say. Use Every Door Direct Mail when the route itself is the audience and broad local awareness is worth paying for. Neither format is inherently “best”; they buy different kinds of reach.

USPS Every Door Direct Mail allows a business to select carrier routes and reach every active residential address on those routes without buying a household name list. Under the current USPS description, EDDM Retail mailings must contain at least 200 mailers and no more than 5,000 mailers per day per ZIP Code. Mailers use simplified addressing such as “Postal Customer,” prepare bundles and facing slips, and deliver the mailing to the correct Post Office under the applicable rules. Always verify the live Domestic Mail Manual EDDM standards because dimensions, preparation, entry, and eligibility matter.

Targeted mail uses a specific address file. That introduces list acquisition, resolution, hygiene, exclusion, and permissible-use work, but it also allows the campaign to omit a recently replaced roof, a duplicate record, an active customer job, a nonserviceable property, or a record under a company-specific exclusion. It can also assign different truthful offers to different known histories.

Decision factor	Targeted address mail	EDDM
Selection unit	Property or household record	Postal carrier route
Names required	No; address targeting can still use occupant addressing	No
Best use	Constrained capacity, high-value property signals, CRM/history-aware outreach	Dense awareness, installed-job halo, broad route launch
Exclusion precision	High if the data process is sound	Limited because the route is intentionally comprehensive
Personalization	Property/context-specific	Route/area-specific
Data burden	Higher	Lower
Attribution	Strong with property IDs and matchback	Possible, but route-level spillover is common
Main failure mode	Bad data or overclaimed signal	Paying for poor-match doors because reach is the product

Do not compare the two by postage alone. Compare all-in acquisition cost per incremental signed contract. Include list/data, creative, printing, postage, landing/phone tracking, response handling, inspection labor, commissions or incremental selling expense, returns, and the internal time required to operate the campaign.

A useful hybrid is narrower than “mail everybody twice.” A roofer might use targeted mail for the highest-priority properties, then use EDDM around a completed installation where route-wide brand recognition is the purpose. The installed job is evidence that the contractor works nearby; it is not evidence that every neighbor has roof damage.

USPS prices changed in July 2026. This guide intentionally does not freeze a postage rate in an evergreen formula. Use the live USPS price-change and Notice 123 resources or a USPS calculator, then record the price date inside the campaign budget.

Is roofing direct mail better than Google Ads, shared leads, or canvassing?

No acquisition channel is universally best. Targeted direct mail is a strong choice when address-level selection can reduce wasted exposure, the roofer wants to earn new homeowner attention under its own brand, and the full response-to-contribution path can be measured. Paid search captures active local intent, shared leads buy access to a reseller’s inquiry, and canvassing adds a live field conversation. Compare the job each channel performs before comparing its price.

The common “cost per lead” table is usually misleading. A Google click, a resold form, a delivered postcard, and an attempted door are not the same unit. Nor is a call always a qualified roofing opportunity. Normalize every channel from its purchased inventory through collected-job gross profit and the capacity it consumes.

Channel	What the roofer buys first	Best operating role	Control the roofer retains	Scarce capacity consumed	Failure hidden by raw lead cost
Targeted direct mail	Exposure at selected addresses	Earn attention in a serviceable property cohort	Audience rules, creative, timing, response path, first-party outcome record	Calls, bookings, inspection travel, follow-up	Bad data, returns, organic conversions, poor-match inspections
EDDM	Every active residential address on selected routes	Route-wide local awareness or installed-job halo	Route, creative, timing, response path	The same downstream capacity, spread across a broad route	Poor-match doors are included by design
Paid search/Local Services Ads	Click, call, or platform-defined lead around existing intent	Capture people already looking	Offer, budget, geography, landing/call experience within platform rules	Fast response, qualification, appointment and sales capacity	Platform attribution, invalid/duplicate inquiries, bid inflation, weak call handling
Shared or exclusive lead provider	A provider-defined inquiry	Add immediate lead volume	Acceptance/dispute rules and the company’s response process	Near-immediate contact, repeated follow-up, inspection slots	Resale count, consent/origin, duplicates, geography, credits, low contact rate
Canvassing	Paid rep time and attempted doors	Create a human conversation in a dense eligible territory	Territory, script, route, rep training, door disposition	Rep hours, drive, management, inspections, safety oversight	Dead doors, duplicate knocks, territory quality, no-shows, turnover
CRM reactivation	Outreach to an eligible known relationship	Recover value from prior customers, estimates, repairs, or inspections	Segment, history-aware message, timing, channel, outcome	Data review, replies, bookings, inspections	Duplicates, old ownership, permissions, organic repeat work

Compare channels on one economic ledger

For each channel and cohort, calculate:

realized contribution = realized job gross profit − channel/outreach cost − incremental response, inspection, and selling cost

Then add the capacity ratios:

• realized contribution per qualified response;
• realized contribution per completed inspection;
• realized contribution per sales hour or canvassing hour;
• realized contribution per production slot consumed;
• cash collected per dollar spent and days from spend to collection.

Suppose two channels each create twelve completed, collected jobs at $4,500 realized gross profit per job. Both create $54,000 of job gross profit. Channel A costs $18,000 in media and $9,000 in incremental answering, inspection, and sales work, leaving $27,000 of contribution. Channel B costs $25,000 in media but only $4,000 in downstream work, leaving $25,000. Channel A wins the contribution comparison by $2,000, but it may still be the wrong next channel if its inspection burden consumes the calendar needed for higher-margin work.

That is an illustrative equation, not a channel benchmark. Replace the jobs, margin, spend, cancellations, downstream labor, and timing with cohort data. The point is that lower media cost can purchase more operational work, and a higher cost per initial response can still purchase a better job mix.

When targeted mail can be more efficient

Targeted mail can outperform broader acquisition when the selection system removes properties the company cannot or should not serve and concentrates spend where an inspection offer has a credible basis. In PorchRocket’s field experiment, the complete higher-opportunity, property-relevant arm produced materially more contracts and modeled contribution than broad generic mail on the same 32,000-mailer denominator. The cleanest randomized result is the personalization comparison within the higher-opportunity stratum; the selected-versus-broad difference is an operational strategy comparison, not a causal estimate of scoring alone.

The mechanism is not mysterious. Fewer poor-match exposures can mean fewer irrelevant responses, shorter travel, more appropriate inspections, and more signed work per unit of selling capacity. That advantage disappears if the property data are wrong, the copy overstates what they mean, or the front desk mishandles the response.

Direct mail also avoids a live click auction and a reseller’s lead definition. That does not make the mailbox free or “owned.” The contractor still pays for data, creative, production, postage, response handling, and learning. USPS controls delivery rules, and homeowners control whether they engage.

When search, shared leads, or canvassing should come first

Paid search can be the first growth dollar when the local market already has meaningful high-intent search volume and the company has a strong page, Business Profile, reviews, fast call handling, geographic control, and reliable downstream attribution. A household already searching “roof leak repair near me” is in a different state from a selected homeowner receiving a replacement-planning offer.

Shared or exclusive leads can be useful when the company needs immediate volume, the provider can document origin and outreach authority, the contract defines acceptance and credits, and sales can respond within the required window. They become dangerous when the owner buys a monthly number without knowing how many competitors received the same inquiry or whether the company can reject a duplicate, out-of-area, renter, spam, wrong-service, or unreachable record.

Canvassing can be superior when a dense, safely workable territory and a credible field conversation justify paid rep time. It also produces observations and objections a postcard cannot. Its economics should include drive, supervision, no-knock restrictions, safety, duplicate attempts, rep compensation, inspections, and turnover, not merely doors per hour. The canvassing and virtual territory guide owns that operating decision.

The practical portfolio order

For many roofers, the sequence is:

1. capture inquiries already arriving by answering and following up;
2. work eligible known opportunity in the CRM;
3. earn selected net-new attention through targeted mail or well-governed field work;
4. capture the branded searches and category interest those efforts may create;
5. scale only the next cohort whose marginal contribution and capacity burden remain acceptable.

The order changes when the facts change. A new company with no historical file cannot reactivate it. A company with a six-week production backlog should not accelerate acquisition merely because media is available. The canonical roofing lead-generation guide provides the full portfolio and budget decision; this guide owns whether direct mail is designed and measured well enough to deserve a place in it.

How does direct mail help a roofer build a controlled local brand asset?

Direct mail can build local recognition because the roofer controls the selected cohort, sender identity, creative, response experience, and first-party outcome history. That is more precise than calling direct mail an “owned channel.” The contractor does not own the mailbox or the homeowner, and an address on a purchased file is not consent for unrelated calls, texts, or email.

The durable asset is the operating memory created around the campaign:

• which property was eligible and why;
• which evidence and creative version were used;
• what the household received and when;
• whether the mail returned, the household declined, or a response occurred;
• which calls, forms, inspections, contracts, cancellations, and collected jobs followed;
• what the holdout did without the mail;
• which message, score band, market, and job mix produced conservative contribution.

That history is reusable. It can prevent duplicate contact, inform a later property review, coordinate mail with safe canvassing, and help the contractor distinguish a new homeowner from a prior estimate or active customer. It should remain exportable if the vendor relationship ends.

Brand recognition is a mechanism, not an attribution shortcut

A physical mailer may make a later truck, yard sign, referral, branded search, or main-number call more recognizable. Keep the sender name, visual identity, phone experience, inspection promise, and field presentation consistent enough that the household recognizes one company. Local proof must be real: actual service area, substantiated reviews, appropriately used project context, and credentials the company currently holds.

Do not credit every later branded search or direct call to mail. A useful journey record might show:

selected mail exposure → branded search → direct website visit → main-number call → completed inspection → contract

Store mail as the original campaign exposure, branded search and the website as assisted touches, the receptionist as an operational touch, and the property as the joining key. A holdout or staged geography helps estimate whether branded searches and direct responses changed beyond the background trend.

Separate direct response from broad awareness

A direct-response cell asks selected households to take a measurable action and is evaluated through incremental contracts and contribution. A brand-awareness cell may intentionally reach a broader carrier route around an installed job or market launch. Its near-term measures can include reach, site/direct traffic by staged geography, branded search, qualified calls, and later contracts, but the company should set a longer decision window and accept that household-level attribution will be weaker.

Do not let “brand” become a shelter for a campaign that failed its stated direct-response objective. Decide the purpose before selection. Targeted mail is usually the stronger fit when budget or inspections are scarce. EDDM can be rational when route-wide awareness is the product. Both require recognizable service and honest measurement.

Coordinate the physical and service experience

The mailer makes a claim about how the company operates even when the headline contains no performance promise. A clean, specific planning offer followed by a confused call, unavailable calendar, late estimator, generic inspection, or missing report weakens the brand. Use the same reason/boundary/value language in the response script and the field handoff. If the offer promises documented observations, define the document and who delivers it.

This is where direct mail can become more than rented attention. The company retains a governed relationship and outcome record, learns which local promise it can keep profitably, and compounds that knowledge across cohorts. The asset is not a stack of addresses. It is a trustworthy, measured path from selected property to served customer.

What should a roofer decide before requesting direct-mail quotes?

Decide the job, geography, offer, response ceiling, economic floor, and experiment before asking for a price. Otherwise, vendors quote different products under the same label: one may include list work and call handling while another quotes only printing and postage. The cheapest column can then win for no defensible reason.

Start with a one-page buying brief. It should be specific enough that a printer, an agency, and an integrated platform are bidding on the same operating requirement.

Buying-brief field	A usable answer	An answer that is still too vague
Commercial objective	“Create incremental retail replacement inspections in these three counties”	“Get more leads”
Eligible job	Asphalt replacement, owner-occupied residential, minimum estimated job value defined internally	“Roofing”
Geographic boundary	Polygon or exact ZIP/municipality list checked against licenses and drive time	“The metro area”
Capacity ceiling	No more than 35 new completed inspections per week during the pilot	“We can handle plenty”
Reason for contact	Roof-age planning signal with source date and uncertainty language	“AI says they need a roof”
Homeowner value	Documented inspection and replacement-planning review, with no coverage promise	“Free estimate”
Primary outcome	Incremental signed-and-not-cancelled contracts within 90 days	QR scans
Economic floor	Positive contribution at $4,200 gross profit per completed job and stated selling cost	10× revenue ROAS
Comparison	Selected property-relevant mail versus selected generic mail, with a locked holdout	“Try three postcards”
Stop condition	Pause if scheduling lag exceeds five business days or source writeback falls below 95%	“Stop if it feels slow”

That brief also exposes whether direct mail is the next bottleneck. If calls already roll to voicemail, fix roofing call coverage before creating more calls. If thousands of permissioned estimates and prior customers have no next action, the CRM reactivation workflow may be cheaper to test first. If the offer depends on a recent verified event, sequence it under the storm-response operating plan. Those pages own their respective systems; this guide owns the mailing decision.

Calculate the maximum affordable mailer cost backward from the job

Vendors naturally quote cost per mailer. An owner should calculate the ceiling from the other direction.

Let:

• G = expected gross profit per completed, collected job from this source;
• S = incremental selling and inspection cost per signed contract;
• Y = conservative signed-contract yield per delivered mailer;
• C = required contribution per signed contract after acquisition; and
• M = maximum all-in mail cost per delivered mailer.

Then:

M = Y × (G − S − C)

Suppose the conservative plan uses $4,500 of gross profit, $700 of incremental inspection and selling cost, a 0.20% contract yield, and a requirement that $1,000 per contract remain after variable acquisition. The maximum mail spend is:

0.002 × ($4,500 − $700 − $1,000) = $5.60 per delivered mailer

That is a ceiling, not a recommendation to spend $5.60. It tells the buyer that a $1.05 all-in proposal has room under those assumptions. Recalculate at a weaker yield, lower margin, and higher cancellation rate. If the proposal works only at the optimistic edge, the campaign is not underwritten; it is hoped for.

Separate fixed setup cost from marginal scale cost

A pilot often carries creative, data integration, landing-page, call-routing, and analysis work that will not repeat at the same level. Keep those costs visible in two views:

1. Pilot cash view: every dollar required to run and evaluate this test.
2. Repeat-unit view: the expected cost if the approved system runs again with controlled creative and data updates.

Do not remove setup cost to make the first campaign appear profitable. Do not assume the first campaign’s cost per mailer remains constant at scale either. A serious quote shows setup, data, print, finishing, postage, tracking, response handling, integration, analysis, taxes, minimums, reprints, and change-order rates as separate lines.

How does property prequalification work for roofing direct mail?

Property prequalification is a sequence of serviceability, exclusion, evidence-confidence, expected-value, and capacity gates used to decide which addresses deserve a mailing. It does not prequalify a homeowner for financing or insurance, diagnose roof condition, or certify that a replacement is needed. The output is a better-ordered inspection offer.

The distinction matters because “targeting” can hide several different decisions. Eligibility asks whether the contractor may and can serve the property. Ranking asks which eligible properties have the strongest current reason for an appropriate offer. Pacing asks how many of those properties the operation can contact now. A high opportunity score should never override an active-job exclusion or an unserviceable roof type.

Gate 1: serviceability

Start with facts the business controls:

• licensed or otherwise authorized geography and solicitation boundaries;
• drive time and branch assignment;
• residential property and roof types the company accepts;
• retail, repair, restoration, financing, or other job mix being offered;
• minimum or maximum job characteristics relevant to the actual operating model;
• supplier, crew, inspection, and production reach;
• warranty and customer-service coverage after installation.

This gate prevents a common waste: scoring every address in a ZIP and discovering after response that the property is across a licensing boundary, too far from crews, a flat/commercial system the team does not install, or a repair the campaign economics cannot support.

Gate 2: exclusions and conflicts

Remove records that should not enter the ranking competition. That includes invalid/vacant addresses, duplicate property identities, active jobs, current estimates with a different owner, unresolved complaints, company exclusions, recent replacements supported by reliable evidence, recently contacted households under the frequency policy, and records whose data purpose or permissible use is not established.

An exclusion is not a low score. Keep its reason code. Otherwise the next refresh may put the same property back into a campaign simply because it continues to look attractive to the model.

Gate 3: evidence and confidence

For remaining properties, assemble only the evidence relevant to the offer. Potential inputs include:

• roof installation or permit history, including whether it represents full or partial replacement;
• contractor CRM inspections, repairs, estimates, jobs, and stated prior outcomes;
• roof type, approximate size/complexity, and property characteristics tied to service fit;
• remotely observed indicators with imagery/source dates and quality flags;
• cumulative or recent verified weather proximity, with event date and spatial uncertainty;
• ownership/address continuity and mail deliverability;
• route or branch cost and available inspection capacity.

No source deserves silent authority. Permit coverage varies. Public and commercial records can lag. Imagery may be old, obscured, or incorrectly matched. A structure can have more than one roof section. A recent replacement may not be recorded. Each field needs a source, as-of date, unit, missing-value behavior, and confidence or review status.

Gate 4: conflict resolution and permitted language

When sources disagree, use a declared hierarchy and preserve the conflict. A contractor invoice or reliable permit showing a full replacement last year may outweigh an older inferred roof-age band. A partial permit should not silently reset the whole roof. An uncertain address match may require review instead of inheriting another property’s history.

The selection record should also control what the creative may say. A practical address-level reason card looks like this:

Field	Example operational value
Property identity	PR-0048217; normalized service address; match confidence 0.98
Eligibility	Residential asphalt replacement; inside Branch North service polygon
Primary signal	Estimated 18–24-year installation band
Source and date	Commercial property/permit synthesis; refreshed July 10, 2026
Confidence	Medium; no full-replacement permit found, records may be incomplete
Corroborating context	Roof type/size fits crew; no active job or recent company contact
Permitted use	Prioritize a documented condition and replacement-planning offer
Permitted wording	“Records suggest your roof may be in an age range where a current inspection can clarify planning.”
Prohibited inference	Exact roof age, current damage, required replacement, remaining useful life, insurance action or coverage
Field question	“Do you know when all or part of the roof was last replaced?”
Fallback	Neutral local-planning creative or exclude if the field conflict cannot be resolved

Printing an internal score would not help the homeowner understand the offer. Convert the reason into the narrowest truthful category, keep the score and provenance internal, and give the responder or inspector a verification question.

Gate 5: expected contribution and capacity

A probability of contract is not enough. Rank the expected economic outcome after mail and downstream cost:

expected contribution per property = P(realized job) × expected realized job gross profit − mailer cost − P(completed inspection) × incremental inspection/sales cost

That equation can include cancellation risk, job mix, travel, or branch capacity when those values materially differ. It should not encode protected characteristics or direct proxies. Nor should it create false precision: if margin by service line is unreliable, show a range and use a conservative threshold.

Release the highest eligible expected-contribution cohort that fits call, inspection, estimate, production, and cash capacity. The remaining eligible properties stay below the current cutoff; they are not “bad roofs.” A future evidence refresh, new capacity, or different truthful offer may change their order.

Gate 6: audit, correction, and feedback

Before mailing, sample across score bands and reason categories. After response, capture field corrections: new roof date, wrong roof type, renter/nonowner, sold property, recent repair, service mismatch, or explicit decline. Feed verified corrections into exclusion and future selection without overwriting the historical campaign snapshot.

This is the real prequalification loop:

define → resolve → exclude → evidence → rank → capacity-limit → inspect → correct → measure

If a vendor cannot export that path for a property, it is selling a list score, not an accountable prequalification system.

Can roof age be used to target roofing mail in Florida?

Yes, an estimated roof-age range can help prioritize a truthful inspection or replacement-planning offer in Florida, but it cannot establish roof condition, remaining useful life, insurability, a claim, or a required replacement date. Florida’s insurance rules make careful wording more important, not less.

The current Florida Statutes section 627.7011 says, in covered homeowners-policy circumstances, that an insurer may not refuse to issue or renew solely because of roof age when the roof is less than 15 years old. For a roof at least 15 years old, the provision gives the homeowner an inspection path before replacement is required as a condition of issuance or renewal; the statute also addresses the circumstance in which an authorized inspection indicates at least five years of useful life remaining. Read the live text and applicable policy with qualified advisers; this summary is not a substitute for either.

That rule does not mean:

• every roof becomes an insurance problem on its fifteenth birthday;
• every roof older than 15 years needs replacement;
• age alone determines condition or remaining useful life;
• an inspection guarantees issuance, renewal, price, discount, or coverage;
• a roofing contractor or marketing platform controls an insurer’s underwriting decision;
• a remotely estimated installation band is the authorized inspection contemplated by the statute.

The Florida Department of Financial Services consumer explanation describes the roof-age and inspection provisions, while its broader homeowners guidance explains that insurers may review roof age, type, construction, condition, location, prior claims, and other underwriting information. “The law has an age-related protection” and “the property meets all underwriting criteria” are different statements.

Citizens Property Insurance Corporation’s current roof-age answer, updated March 17, 2026, asks for documentation supporting at least five years of remaining useful life for underwriter review in the circumstances it describes. That is a Citizens-specific underwriting example. It should not be presented as the rule or decision of every admitted, surplus-lines, or other insurer.

Do not turn a legal threshold into a fear headline

The most dangerous Florida mail is a mailer that converts an estimated age into a threat: “Your roof is over 15 years old, so your insurer will drop you.” That claim can be wrong at every link. The estimated age may be wrong. The relevant roof work may be partial. The actual insurer, policy, inspection, remaining useful life, condition, and other underwriting criteria may differ. The mailer also does not know whether the homeowner has received any insurer communication.

A defensible planning message stays within the contractor’s role:

Records available to us suggest the roof may be in an age range when current documentation can help with maintenance and replacement planning. Those records may be incomplete and do not establish condition, remaining useful life, insurance eligibility, or coverage. A qualified inspection is needed to evaluate the roof itself.

Local legal/compliance review should adapt the message to the sender, source, license, service, and actual offer. Insurance questions should be referred to the homeowner’s insurer, licensed insurance professional, or other appropriate authorized party. The contractor can explain its own inspection and construction work without pretending to make the underwriting decision.

Florida roof-age data needs its own verification controls

At minimum, preserve:

Control	Why it matters in Florida targeting
Last full-roof replacement evidence	The statute contains a roof-age calculation rule tied to when 100% of the surface area was built/replaced; a partial permit can be misleading
Partial replacement history	Different sections may have different dates and conditions
Roof type and geometry	Expected service, inspection, and useful-life context differ; age alone is not condition
Source/as-of date and match confidence	Permit and commercial data can be absent, stale, or linked to the wrong structure
Current contractor CRM history	A recent replacement, active estimate, repair, complaint, or prior inspection can change eligibility or copy
Service/license/solicitation boundary	Florida market eligibility and local operating reach must be verified before contact
Hurricane-season response capacity	A good audience can still overwhelm calls, safe inspections, dry-in/emergency service, materials, crews, and cash
Approved response script	Staff must not upgrade a planning signal into an insurance or damage representation

Use a higher-confidence, lower-claim approach when a threshold is commercially or emotionally salient. If reliable evidence conflicts with the estimated age, correct or exclude the record. If the only way the creative converts is by making the homeowner afraid of losing insurance, reject the creative.

Check enacted law, not a bill’s proposed effective date

Florida SB 808 in the 2026 session proposed changes to the roof provisions and displayed a proposed July 1, 2026 effective date. The official history says the bill died in Banking and Insurance on March 13, 2026; its identical House bill also died. A proposal’s text and planned date are not current law.

This is why an evergreen Florida campaign needs a dated legal/source review before each material reuse. Link the live statute and current agency or carrier source inside the approval packet. Archive the version reviewed. Do not let last year’s postcard, or a vendor’s sales deck, become the company’s legal research.

How should a roofing company build a mailing list?

Begin with serviceability and capacity, resolve records to properties, apply explicit exclusions, and only then rank opportunity. Scoring a dirty, unserviceable universe produces a sophisticated order of the wrong records.

1. Define the serviceable universe

Draw the actual service polygon, not a list of ZIP Codes copied from the footer of a website. Account for municipal boundaries, drive time, roof type, licensing, product/service mix, crew reach, supplier constraints, and the number of inspections the sales team can complete without breaking follow-up. If a company can absorb 150 extra completed inspections during the pilot window, a list designed to create 600 is not more successful; it is uncontrolled work-in-progress.

2. Resolve people and history to a property

A single property may appear under a customer, spouse, old lead, canvassing attempt, imported marketing list, and permit record. Preserve those source records, but create one current property-level campaign identity. A safe merge does not silently overwrite contact permissions, exclusion status, prior outcomes, or the provenance of a field. Uncertain matches stay separate or enter review.

3. Exclude before ranking

At minimum, consider:

• invalid, vacant, returned, or otherwise nonserviceable addresses;
• duplicate property or household records;
• properties outside the licensed and serviceable footprint;
• active jobs, unresolved complaints, and current estimates where another communication owner exists;
• company do-not-mail or broader do-not-contact records;
• recent prior mailings under the frequency cap;
• property types or job types the contractor does not serve;
• records that fail the data provider’s permissible-use or quality requirements.

A campaign should produce a exclusion report with counts and reason codes. “The AI chose 12,400 homes” is not an audit trail.

4. Rank credible opportunity, not alleged damage

Useful ranking inputs can include roof age ranges, known roof type, property characteristics relevant to the company’s work, prior customer/estimate history, verified service events, and dated weather intersections. Each has uncertainty. Data may be stale, a permit may be missing, imagery may be obscured, the structure may have changed, and a storm footprint does not prove a particular roof was affected.

The output should say, in effect, “this address has a stronger reason to offer an inspection,” never “this home needs a roof.” A licensed or otherwise qualified inspection, performed safely and under applicable rules, is where condition is evaluated.

5. Audit the list supplier

Ask any list or scoring provider:

1. What is the source and permissible use of each field?
2. When was the field last refreshed, and is its date exposed?
3. What match rate and uncertainty categories should we expect?
4. How are property changes, moves, duplicates, and returns corrected?
5. Can we supply our own exclusions, and are they enforced at the property level?
6. Which sensitive attributes and protected-class variables are excluded? How are direct proxies reviewed?
7. What data is returned to our CRM, what is retained by the provider, and how is it deleted?
8. Can we export the mailed universe, creative version, reason codes, and holdout assignments?

If a vendor will not show why a record was included, it cannot help you diagnose why a campaign worked or failed.

Build an eligible-universe waterfall that reconciles to one number

The final mail count should be reproducible from the original universe. Do not accept a spreadsheet whose row count merely happens to match the print invoice. Require a waterfall with mutually exclusive reason codes, or a clearly documented priority order when one property violates several rules.

Here is an illustrative waterfall for a replacement contractor. The counts are examples, not PorchRocket benchmarks:

Waterfall stage	Removed at stage	Remaining	Control question
Raw property universe	0	86,420	Is every record inside the requested geographic source?
Outside service polygon	9,870	76,550	Does the polygon reflect actual licensing and drive-time limits?
Nonserved property/job type	11,410	65,140	Are multifamily, flat, metal, repair-only, or other exclusions explicit?
Invalid/vacant/unmailable	3,280	61,860	Which validation date and status created the exclusion?
Duplicate property/household	4,960	56,900	Which record survived and why?
Active job/open complaint/current estimate	1,140	55,760	Did the CRM snapshot arrive before list lock?
Company exclusion/frequency cap	2,730	53,030	Are direct mail and other channel exclusions reconciled?
Missing required evidence or stale signal	8,530	44,500	Is staleness defined per field rather than globally?
Reserved holdout	4,500	40,000	Was assignment locked before production?
Capacity-limited pilot selection	30,000	10,000 mailed	Does expected response fit the inspection calendar?

The 30,000 properties not selected at the final stage are not automatically a control group. They may rank differently from the mailed records. Keep “eligible but below pilot cutoff,” “random holdout from the treated population,” and “ineligible” as different states.

Every removal needs at least: property ID, rule ID, rule version, decision timestamp, source-as-of date, and enough provenance to reproduce the decision. Save the pre-exclusion and post-exclusion counts. If the CRM extract changes after assignment, log the delta instead of silently rebuilding the cohort.

Treat address quality as an operational system, not a vendor checkbox

Address hygiene has at least four layers: syntactic standardization, postal deliverability, move/address updating, and company-specific eligibility. Passing one does not imply passing the others. A deliverable address can still be outside the service area. A current address can still belong to an active customer who should be excluded.

For mail claiming qualifying commercial prices, the USPS Move Update standard describes approved methods and generally requires the addresses used for the mailing to have been updated within 95 days before the mailing date. Applicability depends on the mail class and price claimed. Make the mail owner or qualified provider document the method, processing date, and records affected rather than copying “NCOA complete” into a status email.

Returned mail is feedback, not trash. Load the return reason, address, campaign, and date back into the property record. Distinguish a transient production problem from evidence that the address should be excluded or corrected. When a corrected address is obtained, preserve the old address and provenance; do not erase the history that explains prior nonresponse.

Review a sample before the list is locked

Draw a human-review sample across the top, middle, and bottom of the proposed score, not just the most impressive records. For each sampled property, ask:

• Is it inside the actual service area?
• Is the property and job type one the contractor accepts?
• Are the source fields dated and intelligible?
• Would the proposed copy still be truthful if one uncertain field were wrong?
• Does the CRM show a reason to exclude or change the message?
• Can an operator explain inclusion in one sentence without claiming damage?

Record defect types and rates. A list does not pass because ten cherry-picked addresses look sensible. Set acceptance thresholds before review and define whether a defect requires correction, exclusion, or a lower-confidence creative treatment.

What should a roofing postcard or letter say?

A strong mailer states a truthful reason for contact, respects the boundary between signal and inspection, gives the homeowner a concrete benefit, asks for one low-friction next step, and makes the sender unmistakable. Personalization is useful when it changes relevance, not when it merely inserts a first name into generic copy.

Use this five-part message architecture:

1. Reason: why this company is writing now, stated at the level the evidence supports.
2. Boundary: what is known and what still requires inspection.
3. Value: what the homeowner receives, such as a documented inspection, maintenance review, or replacement planning conversation.
4. Next step: one phone number and one short mobile path to request an appointment.
5. Identity: legal/brand name, recognizable local contact information, licensing disclosures where required, and a clear way to decline future mail.

A property-relevant example

We work with homeowners in North Ridge whose roofs are reaching the age when an inspection can clarify maintenance and replacement timing. Public and commercial property records can be incomplete, so this is not a statement that your roof is damaged or needs replacement. If you would like a qualified roofer to document its current condition, request a time at the address below.

That copy makes the reason relevant and leaves the condition decision to inspection. The exact language should be reviewed for the contractor’s jurisdiction and the actual source behind the selection.

Copy to reject

Your home has storm damage. Insurance owes you a new roof. Scan now before your claim expires.

That version makes multiple assertions a mailing system is not positioned to make: property damage, coverage/benefit entitlement, and potentially a deadline. It also turns uncertainty into fear. A high scan rate would not make the claims truthful.

The FTC’s advertising guidance applies truth-in-advertising principles across media: advertising should be truthful, not deceptive or unfair, and objective claims need support. Direct mail is not a compliance-free channel because it arrives on paper.

Design for action, not decoration

• Put the homeowner value and next step in the first visual hierarchy.
• Use a readable type size and sufficient contrast; do not hide disclosures in pale six-point text.
• Make the phone number, short URL, and QR code lead to the same offer.
• Keep the landing page fast and usable on a phone; prefill only what is appropriate to expose.
• Give each creative a campaign/version ID without making the mailer look machine-generated.
• Test QR codes from multiple devices and physical print proofs before production.
• Ensure the response page never reveals sensitive household or prior-customer information to anyone who merely has the URL.

USPS Informed Delivery for business mailers can add a digital preview or interactive campaign for eligible mail. Treat that as an additional measured touch, not as a replacement for a coherent physical mailer or a license to double-count one conversion.

Write the creative brief before writing headlines

A creative brief keeps evidence, offer, and production constraints from drifting as a mailer moves among a strategist, copywriter, designer, data vendor, printer, and contractor. The approved brief should travel with the campaign record.

Brief field	What to record
Audience	Exact eligible segment, exclusions, size, geography, and data-as-of date
Selection basis	The highest-level truthful reason the segment was prioritized, including uncertainty
Desired decision	One homeowner action, such as request an inspection window or replacement-planning call
Homeowner value	Deliverable the company can consistently provide, not a vague “free consultation”
Proof	Only substantiated credentials, local work, reviews, warranties, or operating facts approved for use
Prohibited claims	Damage diagnosis, insurance entitlement, false scarcity, unsupported savings, unearned awards
Objections	Why now, how long it takes, what the visit includes, whether the owner is committing to work
Response paths	Phone, URL, QR, operating hours, after-hours behavior, source capture, failure fallback
Required identity/disclosures	Legal sender, address, license or registration disclosures where applicable, privacy link
Measurement	Campaign ID, creative ID, treatment, holdout, attribution window, primary outcome

This prevents a common failure: the data team selects “older roof-age estimate,” the copywriter changes it to “your roof is old,” and the final headline becomes “your roof is failing.” Each step sounds only slightly stronger than the prior one. The printed claim is still unsupported.

Match the offer to the homeowner’s stage

“Free inspection” often asks the homeowner to accept an appointment before the mailer has answered what the inspection is for or what they receive. Make the deliverable concrete.

For an evergreen replacement-planning audience, the offer might include documented observations, age/maintenance questions, photographs taken during a qualified inspection when appropriate, and a written next-step category such as monitor, maintain, repair discussion, or replacement discussion. Do not promise that a remote model can produce that conclusion.

For a prior estimate, the useful next step may be a scope and timing review, not a generic inspection as if the relationship never happened. For a completed-job halo route, the offer may be a neighborhood availability window and evidence of nearby work. For a verified recent event, urgency may be legitimate, but the event intersection still does not prove condition or coverage. Keep the event workflow on the storm-response guide and use the mail to state only what the source supports.

Offer tests should change one meaningful decision variable at a time. “Free inspection” versus “documented roof-planning review” tests positioning. A postcard versus a letter tests format. A selected audience versus a broad route tests list strategy. Changing all three in one comparison may find a winner, but it will not tell the operator why it won.

Use an evidence-to-copy matrix for personalization

Define the strongest permitted wording for each data class before generating variants.

Available evidence	Defensible use	Wording to avoid
Dated roof-age range from a named source	Offer planning around an estimated age range; disclose that records may be incomplete	“Your 19-year-old roof must be replaced”
Prior estimate in contractor CRM	Acknowledge the prior conversation accurately and offer a current review	“We inspected your roof” when the old record says no-show
Nearby completed installation	State that the contractor recently completed or is completing a named/appropriately described nearby project if true and permitted	Imply the neighbor endorsed the company without consent
Property intersects a verified weather footprint	Say the area experienced a dated event and offer a qualified inspection	“Your home has hail damage”
Model-generated opportunity rank	Decide whether and how often to mail; use reason categories that can be audited	Print the score or present it as a diagnosis
No reliable property-level evidence	Use truthful route/local-service creative or do not personalize	Fabricate specificity from a first name and stock aerial image

The matrix should include a fallback. If a merge field is null, stale, or outside its allowed range, the record should receive an approved generic variant or be excluded. Never let the print system improvise around missing evidence.

Build the response path as part of the mailer

Call the tracking number during open hours, after hours, from a blocked number, and after simulating a carrier failure. Scan the QR code from the printed stock under ordinary light. Type the short URL manually. Submit the form with a duplicate, an out-of-area address, and an incomplete entry. Confirm what reaches the CRM and what the homeowner sees.

The booking flow should ask for the minimum information needed for the next step. Explain who will contact the person and why. If the form offers specific appointment times, those times must be real or clearly described as requests pending confirmation. A high-converting form that creates false bookings transfers friction to the office and damages trust.

For call scripts, preserve the boundary used in the mail. The person or AI answering should not upgrade “we identified your area for a planning offer” into “our system found damage.” Required intent handling includes: appointment request, price question, existing customer, opt-out, complaint, insurance question, emergency leak, wrong address, renter/nonowner, and out-of-area inquiry. High-risk or uncertain calls should reach an authorized person with the context intact.

How do you prevent expensive print and postal mistakes?

Use three signed acceptance gates: data proof, physical creative proof, and production/induction reconciliation. A PDF approval alone does not confirm that the correct households received the correct variable message on an eligible mailpiece.

Gate 1: data and merge proof

Before rendering the full run, create records that exercise every branch: longest and shortest names, unit numbers, nonpersonalized fallback, every offer, every reason category, non-English or special-character cases if supported, and deliberately missing fields. Check that:

• the treatment and creative version match the assignment file;
• no holdout record renders a mailer;
• every exclusion is absent;
• merge text remains truthful and grammatically correct;
• barcodes, campaign IDs, phone numbers, URLs, and QR destinations map to the same campaign;
• household members are not receiving duplicate or contradictory mailers;
• exposed URLs do not reveal a property’s private CRM history.

Approve counts by treatment, segment, ZIP, creative, and exclusion status. Hash or otherwise fingerprint the final production file so a later replacement can be identified.

Gate 2: physical mailpiece proof

Review a production-equivalent physical proof, not only a screen. Stock, coating, fold, trim, ink coverage, window placement, tabbing, and finishing can change legibility and machinability. Run QR and OCR checks from the physical mailer. Read the disclaimers without magnification. Confirm that the return address and sender identity are clear.

Mailpiece dimensions and characteristics affect eligibility and price. USPS Business Mail 101 advises checking physical characteristics, and the USPS Mailpiece Design Analyst program provides design and automation-compatibility assistance. Use a qualified postal specialist or request a review before committing to a large print run; do not assume a designer’s template is postal approval.

Gate 3: production and induction reconciliation

After production, reconcile:

eligible assigned − holdout − final exclusions − documented spoilage/rejects = mailers entered

The printer should report planned quantity, rendered quantity, spoiled/rejected quantity, reprints, final entered quantity, induction date and location, and supporting mailing documentation. Differences need reason codes. If a run contains 20,000 assigned mailers but 19,842 entered, the missing 158 cannot be quietly counted as delivered exposure.

USPS Informed Visibility can provide near-real-time tracking information for eligible barcoded mail and includes scan-derived, assumed, or logical events depending on the data. That makes it useful for operational timing and anomaly detection. It is not address-level proof that a homeowner held or read a mailer. Keep “entered,” “tracked through network,” “estimated in home,” “not returned,” and “responded” as distinct states.

Define reprint responsibility before a defect occurs

The purchase order or statement of work should allocate responsibility for wrong data, wrong merge logic, print defects, postal rejection, delayed induction, material substitution, and tracking-link failure. It should state who decides whether the run is stopped, destroyed, corrected, or reprinted; who pays; and how affected records are flagged so they are not mailed twice without approval.

A cheap vendor with no defect protocol can become expensive in one afternoon. Ask for a recent redacted reconciliation and defect report. The quality of the artifact matters more than a claim that mistakes are rare.

How many times should roofers mail the same home?

There is no universal number. Set frequency from the useful life of the signal, the homeowner’s decision window, prior contact, complaint/opt-out behavior, campaign incrementality, and your ability to serve the response. “Seven touches” is not a law of direct response.

A conservative pilot might use one primary mailer, one meaningfully different follow-up after enough time to measure delayed response, and a stop or nurture decision. The follow-up should have a reason: a different planning resource, a seasonal maintenance prompt, an installed-job context, or a newly verified event. Sending the same urgency postcard four times is frequency, not a strategy.

Create one household-level contact ledger across mail, phone, email, text, and canvassing. It should answer: what was sent, why it was permitted, when, under which campaign, what happened, whether the household declined, and when it becomes eligible again. A exclusion in one tool that never reaches the others is not a real exclusion.

Event-triggered outreach has a shorter relevance window but a higher risk of overstatement and capacity overload. Evergreen roof-age or maintenance planning has a longer window and can be paced. For the operating timeline around a verified hail or wind event, use the separate roofing storm-response playbook; this page owns the mail itself.

How large should a roofing direct-mail pilot be?

Size the pilot against two constraints: enough observations to change a decision and few enough expected responses to protect service quality. The smaller of statistical capacity and operational capacity controls the launch.

Start with the decision the pilot must support

There are three different pilot questions:

1. Feasibility: Can the data, mail, response, booking, and CRM chain work without material defects?
2. Direction: Is there enough signal to justify a larger comparison?
3. Confirmation: Is the likely economic difference precise enough to approve a major scale decision?

A 2,000-mailer run can expose broken phone routing and merge errors. It will rarely estimate a low contract yield precisely. Calling it a “statistically proven campaign” does not make the denominator larger.

For a single observed contract yield, an approximate planning standard error is:

SE(p) ≈ √[p(1 − p) ÷ n]

If the planning yield is 0.30% and 10,000 mailers are mailed, the expected contracts are 30 and the approximate standard error of the yield is 0.055 percentage points. A rough 95% interval would extend about 0.11 percentage points on either side, although exact binomial intervals should be used for reporting sparse counts. That range may be adequate to reject a disastrous result but inadequate to distinguish two close vendors.

For comparing two proportions, sample-size planning must include the baseline yield, the smallest difference worth acting on, the allocation between arms, desired power, and error rate. Have an analyst calculate it before the print quantity is fixed. Do not reverse-engineer a power claim after seeing outcomes.

Put inspection capacity into the sample-size sheet

Expected inspections are:

mailed mailers × expected booking rate × expected show rate

Apply an upper scenario, not just the average. If 12,000 mailers at a 2.5% planning booking rate and an 85% show rate imply 255 completed inspections, but the team can absorb only 120 within the response window, reduce or stagger the run. A campaign that overwhelms scheduling can make the strongest audience appear weak because speed-to-contact and appointment delay collapse downstream conversion.

Capacity should include:

• daily call-answer and callback slots;
• inspection appointments by territory and rep;
• travel time and safe roof-access constraints;
• estimate-writing and follow-up hours;
• production backlog if contracts arrive;
• reschedule and no-show buffer;
• after-hours and weekend coverage promised by the response path.

Staggered induction can preserve the experiment if batches and in-home windows are recorded and treatment proportions remain balanced. Do not mail the “best” segment first and the comparison later into a different weather or staffing environment unless time is part of the design.

Pre-register the pilot in plain language

Before the first mailer enters the mail, freeze a short experiment record:

• eligible population and exclusions;
• assignment method and randomization unit;
• treatment and holdout counts;
• primary outcome and denominator;
• attribution start/end and maturation rule;
• economic inputs and treatment of cancellations;
• analysis populations, including intention-to-treat;
• known concurrent campaigns;
• minimum operational and economic success thresholds;
• pause rules for complaints, capacity, or technical defects;
• planned subgroup analyses and which are exploratory.

This is not academic ceremony. It stops a team from switching the primary outcome from contracts to scans after contracts disappoint, excluding hard-to-reach mailed records, or searching dozens of ZIP/creative combinations until one looks exceptional.

How do you measure roofing direct-mail ROI?

Persist a property-level campaign identity from the mailed file into the CRM, measure the entire funnel, and compare against a holdout or credible counterfactual. Unique phone numbers and QR codes help diagnose response, but neither is enough: homeowners may type the company name, call the main line, respond weeks later, or mention the mailer at the door.

Minimum attribution setup

• immutable campaign ID and creative version;
• mailed property ID plus normalized address hash/reference;
• selection reason category and data-as-of date;
• mail induction/drop and expected in-home windows;
• unique phone, short URL, QR, and optional promo/response code;
• original source and latest touch stored separately in the CRM;
• fixed response, inspection, and contract attribution windows;
• address-level matchback for contracts that used another response path;
• holdout assignment locked before launch;
• return, complaint, and opt-out status.

Do not overwrite “original source = direct mail” when a receptionist later books the caller. The receptionist is an operational touch; the campaign remains the discovery source. Conversely, do not credit the mail if the property was already an active opportunity unless the analysis explicitly measures assisted reactivation.

The scorecard an owner should receive

Layer	Report these fields
Reach	selected, excluded by reason, mailed, returned, delivered estimate
Engagement	unique calls, web responses, scans, qualified responses, opt-outs/complaints
Sales	bookings, completed inspections, estimates, contracts, cancellations
Economics	signed and collected revenue, gross profit basis, outreach cost, incremental inspection/sales cost, CAC, campaign contribution
Incrementality	holdout conversions, mailed-minus-holdout lift, confidence interval where appropriate
Capacity	days to appointment, backlog, response-handling SLA, no-show, inspection utilization

A holdout should be selected before the file is mailed, drawn from the same eligible population, and protected from intentional campaign touches during the measurement window. Contamination must be documented: a holdout property may still see a truck, receive another campaign, search for the company, or be referred. Holdouts improve the estimate; they do not create a laboratory.

Scale only when four tests pass together:

1. contribution remains positive under a conservative margin and conversion sensitivity;
2. the completed-inspection and follow-up queues remain inside their service levels;
3. complaints, opt-outs, and misleading-copy defects remain below explicit thresholds;
4. the result is not carried by a tiny number of outlier jobs or an attribution mistake.

Pause when source writeback breaks, holdouts are contaminated beyond interpretation, the team cannot schedule responses, the underlying signal expires, or creative no longer matches the evidence. Stopping a campaign can be good capital allocation.

Estimate incremental contracts before incremental contribution

For a simple randomized mailed-versus-holdout design with comparable observation windows:

incremental yield = mailed contract yield − holdout contract yield

estimated incremental contracts = incremental yield × mailed eligible count

If 20,000 mailed properties produce 70 contracts (0.350%) and 5,000 holdout properties produce 7 contracts (0.140%), the estimated lift is 0.210 percentage points. Applied to 20,000 mailed properties, that is 42 estimated incremental contracts. The other 28 mailed conversions are not “fake”; the counterfactual estimate says some similar properties likely would have converted without this campaign.

Next calculate:

incremental contribution = incremental contracts × gross profit per realized job − campaign variable cost − incremental inspection/sales cost

Use realized or cancellation-adjusted jobs when the window permits. Report the raw mailed outcome and the incremental estimate side by side. Owners need to know both how much work the campaign touched and how much it likely created.

If the treatment and holdout sizes differ, compare rates rather than raw counts. If assignment is clustered by carrier route to reduce spillover, analyze it as a cluster design; treating every household as independent will overstate precision. If the holdout receives phone, email, canvassing, or another overlapping campaign, tag the contamination and decide whether the comparison estimates “mail added to business as usual” or has become uninterpretable.

Reconcile attribution with a conversion ledger

Every claimed contract should have one row that an operator can audit:

Field group	Required values
Exposure	property ID, assignment, rendered/entered status, batch, expected in-home window
Response	first response time, response path, call/form ID, campaign recognized or matchback only
Sales	booking, completed inspection, estimate, contract, cancellation, collected-job status
Source	immutable original source, latest operational touch, assisted sources, override reason
Economics	contract value, expected and realized gross-profit basis, incremental selling cost
Audit	data timestamps, match confidence, analyst version, exceptions and reviewer

Matchback should use normalized property identity and a documented confidence rule. A last name plus ZIP may create false matches; a changed owner or mailing address may create false misses. Separate exact, strong, weak, and manually adjudicated matches. Run sensitivity with and without uncertain matches instead of hiding them in the total.

Diagnose the funnel before changing the list

Low contract yield is a symptom. Locate the first stage that materially diverged from plan.

Pattern	Likely places to inspect first	Do not conclude yet
High returns, low response	address age, vacancy, production file, induction	“Homeowners hate the offer”
Low response, normal booking among responders	selection relevance, envelope/postcard visibility, offer, deliverability	“Sales cannot close”
High response, low qualified-booking rate	misleading curiosity, out-of-area calls, weak call script, wrong CTA	“We need more mail”
Strong bookings, low completion	appointment delay, confirmation, routing, season/weather, false booking UX	“The list is low intent”
Normal inspections, low estimates	qualification mismatch, inspection standard, job-type exclusions	“Creative failed”
Normal estimates, low contracts	pricing, scope presentation, follow-up, financing, rep variation	“Direct mail does not work”
Good contracts, poor contribution	gross-margin assumption, inspection burden, cancellations, job mix	“The campaign is profitable because revenue is high”
Early batch strong, later batch weak	capacity saturation, segment-order bias, timing, creative fatigue	“The model decayed”

Review outcomes by treatment and predeclared operational segments, but protect against small-number storytelling. A ZIP with three contracts and zero in another is not automatically a territory insight. Show denominators and intervals; label exploratory findings; require replication before encoding them into permanent selection rules.

Audit the PorchRocket economics line by line

The case-study contribution can be reconstructed rather than accepted as a headline. In the higher-opportunity property-relevant arm:

1. 240 signed contracts × $15,000 average contract = $3,600,000 modeled signed revenue.
2. $3,600,000 × 35% gross margin = $1,260,000 modeled contract gross profit.
3. $1,260,000 − $188,530 outreach and incremental inspection/sales cost = $1,071,470 modeled campaign contribution.
4. $188,530 ÷ 240 = $785.54, reported as $786 variable acquisition cost per signed contract.

For the conservative sensitivity:

1. 240 × $10,000 × 30% = $720,000 modeled contract gross profit.
2. $720,000 − $188,530 = $531,470 modeled campaign contribution.

Those are arithmetic identities under the stated study inputs, not independent market estimates. Fixed overhead and taxes remain outside the contribution definition. If a contractor’s $15,000 “average job” includes supplements that are uncertain at signing, or its 35% margin excludes commissions that vary with the campaign, those inputs must be normalized before comparison.

How should you compare roofing direct-mail companies?

Make every provider prove the operating system around the mail, not merely show a postcard portfolio. A beautiful mailer sent to an unserviceable, duplicated, or unmeasurable list is still bad direct mail.

Score providers on a 100-point rubric:

Capability	Weight	What good evidence looks like
Audience and property selection	20	Field provenance, dates, uncertainty, reason codes, service-area controls
Hygiene and exclusion	15	Property resolution, returns, opt-outs, active-job and frequency controls
Creative relevance and truthfulness	15	Evidence-to-copy rules, human approval, accessible proof, version control
Print/USPS execution	10	Specs, proofs, induction records, returns, current postal compliance
Measurement and incrementality	20	Funnel denominators, matchback, locked holdouts, economics, confidence
CRM and call integration	10	Durable source, property ID, bookings, dispositions, retries/export
Data governance and exitability	10	Minimal access, retention/deletion, vendor controls, full export

Ask the finalist to show, not promise, five artifacts:

1. a sample exclusion report with counts and reason codes;
2. a property record showing source, date, uncertainty, and why it is eligible;
3. a real campaign report with denominators through contracts and costs;
4. a holdout assignment and the method used to protect it;
5. the export and deletion procedure if the relationship ends.

Red flags include guaranteed jobs, damage, or response; “AI targeting” with no field provenance; reporting only scans/calls; no ability to accept your exclusions; a list that cannot be exported; creative that implies insurance coverage; and a contract that owns or resells your responder/customer data without a clear, acceptable purpose.

Run a paid proof with the same artifacts required in production

A sales demo shows the vendor’s best path. A proof should show your path. Provide a small, appropriately protected sample and ask each finalist to return:

• a field dictionary with source, definition, refresh date, uncertainty, and permissible-use notes;
• the resolved property table and all merge decisions;
• the exclusion waterfall with reason codes;
• a ranked file with human-readable inclusion reasons;
• three rendered records per creative branch, including missing-field fallbacks;
• the CRM writeback payload and error behavior;
• a campaign/economics report populated with synthetic outcomes;
• a complete export and deletion attestation.

Pay for serious work and state that the proof data cannot be used for another purpose. Use the same scoring rubric across providers. Do not reward a finalist for showing extra sensitive data you did not ask for; reward data minimization and explainability.

Put operating requirements in the statement of work

Commercial language should match the workflow. At minimum, define:

Contract area	Questions the document should answer
Scope	Who owns service-area definition, list resolution, exclusions, creative, proofing, print, postage, response capture, CRM writeback, analysis?
Acceptance	Which counts, samples, physical proofs, tracking tests, and reconciliation files must pass before induction?
Timing	What are file-lock, proof, production, induction, expected in-home, reporting, and outcome-maturation dates?
Change control	Who can change selection rules, copy, stock, quantity, postage class, URLs, or call routing; how is approval recorded?
Data rights	Who owns raw input, derived selection fields, responders, outcomes, creative, and analysis? Can any data train a shared model or be resold?
Security	What access, MFA, encryption, subprocessor, incident, retention, deletion, and audit terms apply?
Measurement	Which denominators, outcomes, holdouts, match rules, cost fields, and exports are mandatory?
Defects	What happens after wrong merge, duplicate mail, postal rejection, late induction, broken response path, or reporting loss?
Exit	In what usable formats are data, creative, logic, exclusions, and history returned; when are copies deleted?

Avoid a contract in which the vendor guarantees “leads” but defines a lead as any call, including wrong numbers, vendor solicitations, and existing-customer service requests. The definition should specify unique property/person logic, geography, service fit, intent, duplicates, spam, existing records, and the dispute window. Even better, do not anchor the relationship on a lead guarantee; anchor it on transparent execution and a jointly measurable business outcome.

If a provider touches CRM or homeowner information, grant the minimum fields required and time-bound access. Separate production data from broader customer history where possible. Require named subprocessors, breach/incident notification terms, deletion timing, and evidence that access is removed at exit. Direct mail is physical; its operating system is still a data system.

Is PorchRocket the right fit?

PorchRocket’s direct-mail system is a strong potential fit for a residential replacement contractor with a defined service area, known gross profit, capacity for the pilot response, a working CRM or destination system, and an honest inspection or planning offer. It is especially relevant when the owner wants property-level selection, personalized creative, call capture, and downstream measurement in one loop.

It is not a fit when the company wants a guaranteed number of jobs, treats a model as a roof diagnosis, has no one accountable for inspections and follow-up, cannot honor exclusions, lacks the licensing/authority to serve the target area, or wants to imply coverage or waive deductibles. It can also be the wrong first move when the real bottleneck is unanswered calls or an unworked CRM; the roofing lead-generation channel guide helps sequence that decision.

How should a roofer scale direct mail after a winning pilot?

Scale one constraint at a time: audience depth, geography, weekly volume, creative variants, or channel coordination. Multiplying all five converts a known pilot into a new, unmeasured system.

Build a marginal scale curve

The first 10,000 selected properties may have stronger evidence than the next 50,000. Report performance by locked score bands or eligibility tiers, not only as a blended average. For each tier, show:

• eligible, mailed, returned, and holdout counts;
• response, completed-inspection, contract, cancellation, and collected-job rates;
• gross profit and incremental campaign contribution;
• appointment delay and sales-hour consumption;
• complaints, opt-outs, and data defects;
• confidence/uncertainty and concurrent-market conditions.

Expansion stops where the next band’s conservative incremental contribution no longer clears the company’s required return or where capacity erodes the conversion of the existing bands. Average campaign CAC can remain attractive after the marginal tier has become unprofitable. Budget the next mailer, not the historical average mailer.

Use cells that preserve learning

A mature program maintains a controlled baseline while testing one material improvement. One practical allocation is a large proven-control cell, a smaller challenger cell, and an untouched holdout. The percentages depend on volume and risk; there is no universal 80/10/10 rule.

Keep creative and selection versioned. A “control” that quietly receives new imagery, a new landing page, a different call center, and a broader list is not a control. When a challenger wins on the primary business outcome and passes quality checks, promote it deliberately and archive the previous version with its dates and results.

Protect the operating calendar

Mail volume should be released against forward capacity, not annual budget. Before each batch, project calls, bookings, completed inspections, estimate workload, likely contracts, and production starts under low/base/high response. Include open work from prior cohorts. A weekly release meeting can be short if the ledger is sound:

1. Are all sources, exclusions, and response paths healthy?
2. Is the next audience still supported by current evidence?
3. Can front desk, field, sales, and production meet their service levels?
4. Did recent marginal contribution remain above the floor?
5. Is any weather, pricing, staffing, or concurrent campaign change large enough to alter the design?

The decision is release, reduce, delay, reroute, or stop. “The postcards are already in the warehouse” is not an economic reason to create a scheduling failure.

Refresh evidence without rewriting history

Property and CRM fields change. Refresh them before a new cohort is assigned, but preserve the snapshot used for each prior decision. Never rescore a historical mailing with today’s fields and imply those values were known at launch. Monitoring should distinguish:

• data drift: source coverage, age, or match quality changed;
• population drift: the available properties differ from the pilot population;
• execution drift: call handling, inspections, creative, or sales behavior changed;
• outcome drift: response or economics changed after the other factors are accounted for.

That distinction determines the fix. Data drift may require a source correction. Capacity drift may require pacing. Outcome drift may justify a new test. None is solved by putting a brighter badge on the same postcard.

A 30-day roofing direct-mail launch plan

Use the first month to earn the right to mail a controlled pilot, not to race from a purchased list to 50,000 mailers. The work before the drop determines whether the result can be understood.

Week 1: economics and serviceability

• Set the service polygon and job types.
• Calculate gross profit per signed and collected contract.
• Measure inspection capacity and current backlog.
• Define maximum all-in CAC, minimum acceptable contribution, and stop conditions.
• Choose the owner of calls, web responses, bookings, and CRM corrections.

Week 2: data and experiment

• Import and resolve the property universe.
• Apply invalid, duplicate, active-job, prior-contact, and company exclusion rules.
• Audit selection fields for provenance, dates, uncertainty, and prohibited features.
• Choose one primary comparison and lock a holdout before creative launches.
• Size the pilot to learning and capacity, not to a round print quantity.

Week 3: creative and handoff

• Write the reason/boundary/value/next-step message.
• Have qualified local counsel or compliance ownership review claims and disclosures.
• Build the mobile response page and test privacy boundaries.
• Configure phone/URL/QR tracking and CRM writeback.
• Train the person or agent who answers on the exact offer and disallowed promises.

Week 4: proof, drop, and observe

• Approve physical print and data proofs.
• Verify every QR/URL/phone path and failure fallback.
• Record induction and expected in-home dates.
• Review response, booking, capacity, returns, and complaints daily, but do not declare ROI before the conversion window matures.
• Document any operational change that could affect the comparison.

At the end of the attribution window, rerun the economics with collected/cancel-adjusted outcomes, compare the holdout, inspect segments for sample-size fragility, and decide to scale, modify, retest, or stop.

Roofing direct-mail questions owners ask before buying

Is a postcard or letter better for roofing?

Neither wins universally. A postcard exposes the offer without an open and is efficient for a simple local message. A letter creates more room for context, documentation, and a higher-consideration explanation. Test formats against the same eligible population and economic outcome; do not compare one list/offer on a postcard with a different list/offer in an envelope and attribute the difference to format.

How much should a roofing direct-mail pilot cost?

Enough to produce a decision without exceeding response capacity. Build from the all-in cost and minimum detectable business difference rather than a generic “$5,000 minimum.” If the list is small, emphasize exact confidence intervals and qualitative failure review instead of pretending the pilot has population-level precision.

Can a new roofing company use direct mail?

Yes, if it can establish trust, licensing/insurance disclosures, a serviceable territory, credible call handling, and inspection capacity. A new brand may need a different response assumption than an established local contractor. Use a smaller, controlled pilot and avoid borrowing social proof the business has not earned.

Can I mail my customer or old-estimate list?

Often, but treat that as CRM reactivation, with channel eligibility, prior relationship, history-aware copy, and one contact-pressure ledger. The detailed process and 72,000-record experiment belong in the roofing CRM reactivation guide.

Should I buy a “hail-damage list”?

Buy only data whose source, date, spatial uncertainty, permissible use, and correction process you understand, and never turn a footprint or model into a claim that a specific property is damaged. Use it to prioritize a qualified inspection offer. For timely versus official weather sources and activation timing, see roofing storm response.

How long should I wait before judging the campaign?

Set the window before launch from expected in-home timing, homeowner consideration, appointment availability, sales cycle, and contract lag. Watch leading indicators immediately for operational failure, but do not close the outcome cohort while legitimate contracts are still maturing.

Does direct mail replace canvassing or digital advertising?

No. It is one controllable outbound channel. It can cover addresses that should not be knocked, warm a territory before safe field work, or create a measured alternative to auction-based media. The contractor controls the creative and outcome record; it does not own the mailbox or the homeowner. Compare channels on contribution and constrained sales hours. The roofing canvassing software guide owns field execution.

Are 500 or 1,000 postcards enough to test roofing direct mail?

They can test whether the list renders correctly, the mail arrives, the phone and form work, and the office follows the script. They usually cannot estimate a low signed-contract rate with useful precision. If 1,000 mailers produce two contracts, the observed yield is 0.20%, but the uncertainty around two events is wide. Treat a tiny run as an operational feasibility test, not proof of a scalable ROI.

Does a roofing postcard need a discount?

No. A discount can reduce margin, attract the wrong job, or create conditions the company cannot honor. Test a concrete homeowner benefit first: documented condition observations, replacement-planning clarity, a defined neighborhood inspection window, or a relevant current review. If price is the offer, state eligibility and limitations clearly and model the reduced gross profit in the campaign economics.

Should a roofing mailer use a QR code, phone number, or short URL?

Offer at least a clear phone path and one simple mobile web path when the audience can use both. A QR code is convenient but should not be the only response method. Point QR and short URL to the same fast, privacy-safe offer; use campaign parameters without exposing property history; test both from the physical proof; and keep the main-number matchback because some homeowners will search the company or call the number they already find online.

What if the roof-age estimate on the mailer is wrong?

Apologize plainly, correct the property record, honor any decline or exclusion, and do not defend the model as if it overrules the homeowner’s documents. The creative should have disclosed that records can be incomplete, and the response script should ask for the actual replacement history. Track the error source and rate; repeated errors may require a source change, a confidence cutoff, neutral creative, or stopping the treatment.

What does PorchRocket add beyond a printer or USPS EDDM?

A printer produces the mailer, and USPS provides mailing products and delivery infrastructure. PorchRocket’s direct-mail workflow is designed around the decisions surrounding that production: resolve the serviceable property universe, apply contractor-specific exclusions, rank credible opportunity, control evidence-to-copy language, pace volume to capacity, capture calls/forms, return outcomes to the CRM, preserve a holdout, and reconcile contribution. The contractor should still compare the proposed scope, evidence, price, data rights, and exit terms; the platform does not make postal, roof-condition, insurance, or legal determinations.

Methodology and limitations

PorchRocket’s direct-mail evidence is a field study with a pre-specified funnel, intention-to-treat reporting, holdouts, and confidence intervals; it is not a universal customer promise. The detailed result table above uses the study record in PorchRocket_CaseStudy.docx, prepared before this guide. No outcome was invented to complete a missing cell.

The four 32,000-address mailed arms crossed operational higher-opportunity selection versus broad selection with property-relevant versus generic creative. Personalization within the higher-opportunity stratum was randomized. Higher-opportunity versus broad membership was not randomized, so selection comparisons can reflect both the system and underlying population differences. Holdout yields help estimate background conversion but can be contaminated by other touches.

Reported confidence intervals are 95% intervals from the case-study analysis. They describe sampling uncertainty under the study method, not every source of uncertainty. They do not cover data drift, a new market, creative change, seasonality, brand strength, economic changes, implementation error, or capacity saturation.

Economic results assume a $15,000 average signed contract, 35% job gross margin, and $5,250 gross profit per contract unless the conservative sensitivity is explicitly named. Campaign contribution subtracts outreach and incremental inspection/sales cost. It does not subtract fixed overhead, taxes, financing costs, debt service, or owner compensation and is not EBITDA. Readers should substitute actual collected-job economics and define treatment of cancellations.

Remote condition, age, property, and weather fields can be stale or wrong and are opportunity signals only. They do not establish physical damage, cause, repairability, code applicability, insurance coverage, or homeowner eligibility. A qualified inspection and the appropriate licensed/authorized parties make those determinations. Protected-class variables and direct proxies should be excluded; exclusion, opt-out, address hygiene, state marketing rules, and human review remain mandatory.

Vendor and USPS facts were checked against the first-party sources listed below on July 22, 2026. Postal prices and requirements change; the current Domestic Mail Manual and Notice 123 control. This guide is operational information, not legal, insurance, tax, or postal advice.