How to Stop Knocking Dead Doors: A Roofing Sales System for Knocking Roofs That Are Actually Due

A dead door is any house where the conversation was never going to go anywhere, no matter how good your pitch was. The roof was nearly new. The owner re-roofed two years ago with the company down the street. The house took no real storm wear and the shingles have a decade of life left. You can have the cleanest open, the best damage talk in the market, and a closer's instinct, and that door is still dead because the house was wrong before you ever walked up the driveway.

Most reps who burn out think they have a pitch problem. Usually they have a targeting problem. They are spreading the same effort across every house on a street when maybe one in eight of those houses has a roof that is genuinely due — aged out, storm-worn, or both. If you knock all eight at random, you grind through seven frustrating conversations to find the one that matters, and by the time you reach it your energy is gone and your numbers look like proof you can't sell.

What follows is a system for fixing the input side of the funnel: how to figure out, before you set foot on the street, which roofs are most likely to need work, how to sequence them into routes that don't waste daylight, and how to script the knock so the houses that are actually due convert. It is written for the owner or sales manager building the canvassing program, and for the rep who wants to stop walking dead streets. None of it requires you to abandon door knocking. It requires you to knock the right doors in the right order, and to stop treating every address as equally worth your time.

What a "dead door" actually is, and why pitching harder doesn't fix it

Before you can stop knocking dead doors, you have to be honest about why a door is dead. There are really only a few root causes, and almost none of them are solved by talking better.

The roof has no need. This is the big one. A roof installed four years ago, with no meaningful storm exposure, has nothing wrong with it that a homeowner is going to pay to fix. You cannot manufacture need that isn't there, and you shouldn't try. Pushing a sound roof toward replacement is how reps and companies get reputations that follow them around a market.

The need exists but the timing is wrong. Some homes have aging or storm-worn roofs but the owner isn't in a buying window — they listed the house, they're mid-divorce, they have a kid heading to college this fall. You can't see this from the curb, and it's a smaller slice than people think. Don't let it become your excuse for poor targeting.

The need exists but you reached the wrong person. Renters, adult kids watching a parent's house, someone who genuinely doesn't make the decision. Real, but again a minority.

The need exists and you found it — but your approach killed it. This is the only category where pitching better actually helps, and it's the smallest one.

Here's the uncomfortable math. If you canvass a random suburban street, the share of homes with a roof that is both aged-out or storm-worn and owner-occupied and in a plausible buying window is often somewhere in the single digits to low teens as a percentage. Everything else is, for your purposes that day, a dead door. When reps say "door knocking doesn't work anymore," what's usually true is that random door knocking has terrible odds and always did — they just used to have more hours and more reps to brute-force it.

So the entire game is shifting your knock distribution away from random. If you can move from knocking a random street (say one in ten houses worth a real conversation) to knocking a scored list where one in three or one in four houses is genuinely due, you have roughly tripled the productivity of every hour your crew spends on the street without changing a single word of the pitch. That is the leverage. The pitch matters, but it's the second lever, not the first.

The two questions that decide if a roof is due

Every targeting decision comes down to two questions about a specific roof:

1. How old is it, roughly? Asphalt shingle roofs in the U.S. typically live somewhere in the 15-to-30-year range depending on shingle grade, ventilation, slope, and climate. A roof in its first decade is rarely a real prospect absent storm damage. A roof past 18-20 years is a candidate on age alone.
2. What has the weather done to it? Two roofs of identical age in the same neighborhood can be in completely different shape because hail and wind are local and uneven. One street caught the core of a hail swath; the next one over got grazed. One elevation faced the wind; the other was sheltered.

If you can answer those two questions for a list of addresses before you knock, you can rank the list. Ranking the list is the whole ballgame. The rest of what follows is how to answer those two questions at scale, how to turn the answers into routes, and how to knock them.

Put a real dollar figure on a dead door

Reps and managers talk about dead doors like they're free — a shrug, a wasted minute, on to the next house. They are not free, and once you price them you start treating targeting like the financial decision it is.

Walk the math for a single rep. Say a canvasser costs you, fully loaded, somewhere around $25-35 an hour once you fold in base, commission draw, fuel, phone, and the overhead of managing them. In an hour of cold random knocking they might physically reach 25-35 doors, of which maybe 12-18 get answered, of which — on a random street — perhaps one or two have any real chance of needing roof work. The rest are dead doors. So in that hour, the rep spent roughly 90 percent of their answered conversations on houses that could never have become customers. You paid full freight for a 10-percent hit rate on the only thing that matters.

Now change one input. Hand the same rep a list where one in three or four answered doors is a genuine candidate instead of one in ten. Same hourly cost, same number of knocks, same pitch — but the count of real conversations per hour roughly triples. You didn't make the rep faster or better. You stopped paying them to talk to houses that were never going to buy. Across a five-person crew working full weeks through a season, that gap between a 10-percent and a 30-percent candidate rate is the difference between a canvassing program that loses money and one that's the cheapest acquisition channel you have.

The other hidden cost of dead doors is morale, and morale shows up in your turnover line. Canvassing is hard enough when the doors are good. A rep who spends three days getting rejected by houses that had nothing wrong with them concludes — reasonably, from their seat — that the job doesn't work, and quits. You then eat the cost of recruiting and ramping a replacement who will quit for the same reason. Good targeting is a retention tool as much as a productivity tool: reps stay when the doors convert, and doors convert when they're chosen well.

Why "more knocks" is the wrong goal

The instinctive fix for a slow canvassing program is a volume mandate: knock more doors, log more attempts, push the activity number. It feels like accountability and it's easy to measure. It also quietly makes the dead-door problem worse, because raw volume rewards the rep for blowing through houses rather than working the right ones, and it spreads the crew thinner across more random territory.

The goal you actually want is candidate conversations per hour — how many times per hour a rep stands in front of a house that has a legitimate reason to need work. You can raise that number two ways: knock better doors, or get more doors answered. Both are targeting and timing problems. Almost none of it is a "knock more" problem. A rep doing 25 well-chosen, well-timed knocks will out-produce one doing 60 random ones, and will still have energy left for the conversation when a real one shows up. Hold your crew accountable to candidate conversations and quality dispositions, not to a raw knock count, and the whole system points in the right direction.

Building a target list before you ever knock a door

The goal of pre-work is a ranked list of addresses, sorted so your best-odds doors are at the top and your dead doors are either at the bottom or off the list entirely. You build this from a few layers of data, most of which you can assemble yourself with patience, and some of which is faster to buy.

Layer 1: Roof age signals

You will almost never get an exact install date, and you don't need one. You need a range good enough to sort houses into "too new to bother," "watch," and "due." Several signals get you there:

• County permit records. Many jurisdictions publish reroof permits. Where they're available and reasonably complete, a pulled permit from 14 years ago tells you a roof is aging toward its window; no permit in the last 20 years on an older home is itself a signal. The catch: permit data is wildly inconsistent between counties, plenty of reroofs happen without permits, and matching permits to current parcels is tedious.
• Home age and sale history. A house built in 1998 that has never had a recorded reroof and last sold in 2003 has, statistically, a roof that's either original-and-ancient or replaced-without-record. Either way it's worth a look. Public assessor and parcel data (and home-age fields in property datasets) feed this.
• Visual age cues from imagery. Granule loss, color fade, streaking, patched sections, and shingle style all read off recent aerial or street-level imagery once you train your eye. A 3-tab roof on a house that's changed hands twice is telling you something. Architectural shingles with heavy fade and dark streaking are telling you something else.

None of these is precise on its own. Stacked, they sort a list surprisingly well. The honest framing — and the one you should carry into every conversation — is that you are estimating an age range, not a date. "This roof looks like it's in the back half of its life" is a defensible, useful statement. "This roof was installed in March 2009" is usually a guess dressed up as a fact, and it'll burn you when the homeowner corrects you.

Layer 2: Storm exposure

Age alone gives you the "aging out" pile. Storm data gives you the "got worn down early" pile, which is often the more urgent and more winnable work because the damage is recent and the homeowner may not even know it's there.

The public sources worth knowing:

• NOAA's Storm Prediction Center and the Storm Events Database publish hail and wind reports, including estimated hail size and report locations. This is the backbone of "did a storm hit here."
• National Weather Service local offices issue post-event summaries and damage surveys.
• IBHS (Insurance Institute for Business & Home Safety) publishes research on how roofing materials actually fail under hail and wind, which helps you reason about severity rather than just presence.

The limitation of raw public storm data is resolution. A hail report pinned to a town center tells you a storm passed through; it does not tell you that this roof, on this elevation, with this shingle, took a beating while the identical house two doors down was fine. Hail swaths are narrow and patchy. Wind loading depends on exposure and direction. The gap between "a storm was reported near this ZIP" and "this specific roof was damaged" is exactly where a lot of canvassing wastes itself — reps blanket a whole storm-affected ZIP when only a few streets within it actually got hit hard.

A few practical things about reading storm exposure that separate a useful read from a sloppy one:

• Hail size is a rough threshold, not a guarantee. Small hail (under an inch) rarely does functional damage to asphalt shingles; it cosmetically marks them at most. The damage curve climbs sharply once you get into 1-inch and larger stones, and the type and age of the shingle change where that line sits — an older, brittle roof bruises at sizes a newer roof shrugs off. So a 1.25-inch hail report over a 20-year-old roof is a much stronger signal than the same report over a 5-year-old roof. Read size and roof age together, never separately.
• Direction and density of impacts matter more than a single report. Hail driven by wind hits vertical and angled surfaces differently than it hits flat-pitched roofs; the side of the roof facing the storm's approach usually takes the worst of it. You can't fully resolve this from a report point, which is the whole reason per-roof modeling exists, but you can at least note that a single isolated report is weaker evidence than a cluster of reports forming a clear swath.
• Wind events damage roofs through uplift and debris, on a different pattern than hail. A straight-line wind or downburst event tends to lift and crease shingles at edges, ridges, and corners, and to drop tree limbs. The vulnerable houses are the exposed ones — corner lots, homes above their neighbors, roofs with older or under-nailed shingles. Age interacts here too: seal strips on older shingles let go at lower wind speeds.
• Recency decays. A storm from three weeks ago is a live, urgent reason to knock; one from three years ago is mostly relevant only as a clue that a roof may have been worn down and never addressed. Weight recent events heavily and treat old ones as soft age signals rather than damage signals.

The honest summary: public storm data is excellent for telling you which general areas deserve attention and roughly how hard they may have been hit. It is poor at telling you which individual roofs within that area actually took meaningful wear. Closing that last gap — area to specific roof — is the single highest-value step in storm targeting, because it's the difference between blanketing a ZIP and walking straight to the houses that were worn down.

Layer 3: Ownership and basic fit

Finally, filter for the obvious disqualifiers so you're not knocking rentals and vacant lots:

• Owner-occupied vs. absentee owner (mailing address differs from situs address is a common flag).
• Single-family detached vs. units you can't easily sell to.
• HOA constraints in communities where roof work is governed centrally — sometimes a reason to skip, sometimes a reason to target the HOA itself, but either way good to know going in.

Stack these three layers and you have a list where every address carries a rough age range, a storm-exposure read, and a fit flag. Sort by a simple priority and you've replaced "drive to a neighborhood and start walking" with "work this ranked list from the top."

A simple scoring model you can build yourself

You don't need anything fancy to start. Here's a transparent point model you can run in a spreadsheet:

Knock the highest scores first. A roof scoring 8+ (old and storm-hit) is your A-list. A 4-5 (aging or storm-hit, not both) is B-list worth a sweep. Anything at or below 1 is where dead doors live — deprioritize or drop. The exact weights matter less than the discipline of having a score and working top-down. You'll tune the numbers as you see what converts in your market.

A worked example: scoring four houses on the same block

Numbers make this concrete. Picture four owner-occupied single-family homes on one street, six weeks after a hail event that dropped 1.25-inch stones across the western half of the neighborhood.

• House A — built 1999, no reroof permit on record, sits in the part of the block the swath clipped, imagery shows streaking and some granule loss. Age 20+ (+4), significant recent hail (+4), visual cues (+2), owner-occupied (+1). Score: 11. A-list. There's a real, recent, age-plus-storm reason to be here.
• House B — built 1999, reroof permit pulled 3 years ago, same storm exposure. Age under 10 (-3), significant recent hail (+4), no strong visual cues, owner-occupied (+1). Score: 2. The roof is new; even with the storm, this is a marginal door — knock only if you're already on the block, and expect "we just had it done."
• House C — built 1996, no reroof on record, but it's on the eastern half the swath missed. Age 20+ (+4), no qualifying storm (0), visual cues (+2), owner-occupied (+1). Score: 7. B-list on age alone. Worth a sweep with the age opener, not the storm opener.
• House D — built 2017, no storm in its half, pristine roof in imagery. Age under 10 (-3), no storm (0), owner-occupied (+1). Score: -2. A dead door. Skip it. There is nothing here, and you'd only be training the rep to expect rejection.

Four houses on one street, scores from -2 to 11. A rep working them top-down spends real energy on A and C, makes a quick courteous pass at B, and never wastes a knock on D. A rep working the street blind treats all four identically and spends a quarter of the block on a house that was never going to buy. Multiply across hundreds of doors and you can see where a season's productivity goes.

Where roof-age and storm-modeling data tools fit

Everything above can be assembled by hand. The reason most contractors don't keep it current is that it's a grind: pulling permits, matching parcels, cross-referencing storm reports, eyeballing imagery, then re-doing it every time a new storm rolls through. The work is real, and it competes with actually running the business.

This is the gap a tool like RoofPredict is built to fill. The idea is straightforward: instead of you manually stacking age signals and storm reports address by address, it estimates a roof-age range per house from aerial imagery and models storm physics per individual roof — going past "a storm was reported in this ZIP" to how hail and wind loading likely played out on a specific roof given its location within the swath, slope, and exposure. The output is the ranked, house-by-house list described above, kept current as new storms come through, so your crew knocks the roofs the weather actually wore down and the roofs aging out — in route order.

A few honest points about what this kind of data is and isn't, because overselling it is its own way to burn a market:

• It gives you odds and ranges, not certainties. A roof-age estimate is a range, not an install date. A storm model is a probability that a given roof took meaningful wear, not proof that it did. The only thing that confirms damage is an inspection on the roof. Treat the data as a way to point your inspections at the right houses, never as a finding you assert to a homeowner.
• It does not handle, approve, or decide claims, and neither do you. Your role in storm work is to document the roof's condition and provide an estimate. The insurer decides coverage. The homeowner owns the claim and the relationship with their carrier. Any tool or rep who talks like they can guarantee a claim outcome, a covered replacement, or a deductible result is heading somewhere you don't want your company to be.
• It replaces guesswork, not judgment. The data tells you where to look. Your crew's eyes on the roof, your documentation, and your honesty about what you find are still the job.

Used that way, the value is simple and large: you stop spending crew hours discovering, one dead door at a time, which houses were never worth knocking. You walk up to a far higher share of doors where there's a real, legitimate reason to be having the conversation. Whether you build that list by hand or buy it, having a ranked list at all is the difference between random canvassing and a sales system.

Turning the list into routes that don't waste daylight

A perfect target list still fails if your crew works it inefficiently. Two reps can knock the same scored list and one does 80 quality knocks in a shift while the other does 35, purely because of how they move through the area. Routing is the unglamorous half of the system and it's where a lot of theoretical productivity leaks out.

Cluster, don't crisscross

The single biggest routing mistake is letting reps chase scores across a wide area — knock a 9 here, drive six minutes to a 9 over there, double back for an 8. High scores scattered across town means you spend your shift in the truck. Instead, cluster first, then sort within the cluster. Pick a dense pocket of decent scores you can work on foot, exhaust it, then move the truck once to the next pocket. A tight grid of B-list doors you can walk often out-produces a sprinkle of A-list doors you have to drive between.

A practical rule: a rep should be parking the truck and working an area on foot for at least 45-90 minutes before moving it. If they're back in the truck every few doors, the territory is too spread out and you're paying for windshield time instead of conversations.

Sequence for the conversation, not only the geography

Within a walkable cluster, two small sequencing habits help:

• Work the same side of the street in a continuous sweep, then come back the other side. Crossing back and forth wastes steps and makes it easy to skip houses.
• When you get a real conversation or an inspection, knock the immediate neighbors right after. Social proof is strongest at its freshest. "I'm doing a roof check for your neighbor two doors down" is a true, low-pressure opener that works far better on the adjacent houses than on a cold street. This is also why storm clusters convert: the houses that got hit are physically next to each other.

Time-of-day and day-of-week discipline

Dead doors aren't only about the wrong house — they're also about the wrong moment. A great target list knocked at 1 p.m. on a Tuesday hits a lot of empty houses. General canvassing wisdom that holds up: late afternoon into early evening on weekdays (roughly 4-7 p.m., adjusted for season and daylight) and mid-morning to afternoon on Saturdays get the most answered doors in owner-occupied suburbs. Knock too early and you annoy people; knock after dark and you spook them. Respect this and your answered-door rate climbs, which multiplies the value of good targeting.

Also respect the legal layer: many municipalities require a solicitation permit, observe posted No Soliciting signs (skip them — it's both the law in many places and just good practice), and mind any local registry or do-not-knock list. Getting a crew cited or generating complaints is an expensive way to learn this. Check the local rules before you deploy a team into a new city.

A sample shift structure

Here's how a productive canvassing shift actually flows when the targeting and routing are done right:

1. Pre-shift (15 min): Each rep gets a ranked list for one or two adjacent clusters, loaded on their phone, sorted by score and laid out as a walking route. No "figure out where to go" time on the clock.
2. Cluster 1 (90 min): Park once. Sweep the high-score doors on foot, one side then the other. Log every door: not home, not interested, conversation, inspection booked.
3. Reset (10 min): Quick check of what's converting. Move truck to cluster 2.
4. Cluster 2 (90 min): Repeat, ideally timed so the back half lands in the prime answered-door window.
5. Post-shift (15 min): Disposition every knock in the CRM. This is the data that tunes next week's list.

The difference between this and "go knock that neighborhood" is enormous, and almost none of it is about being a better talker.

Assigning territory without overlap or gaps

When you run more than one rep, the routing problem grows a second dimension: keeping reps out of each other's territory and making sure no scored door falls through the cracks. Two reps double-knocking the same house in the same week is worse than a dead door — it makes the company look disorganized and irritates a homeowner you wanted to convert.

A workable approach: carve the scored map into contiguous territories sized to a day or two of work each, assign one rep per territory at a time, and don't let a territory get re-worked until its dispositions are in and a cooling-off window has passed. Use natural boundaries — major roads, rivers, school-district lines — so reps don't have to remember invisible edges. When a storm reshuffles priorities, redraw territories around the new high-score clusters rather than forcing the new work into the old grid. The point is simple: every scored door should belong to exactly one rep at any given time, and every door should belong to someone.

Don't let the easy doors crowd out the right ones

A subtle routing failure: reps gravitate to the doors that are pleasant to knock — nice streets, easy parking, friendly-looking houses — rather than the doors the scoring flagged. Left unmanaged, a crew will slowly re-randomize itself toward comfortable territory and away from the worn-down, less photogenic blocks where the real candidates often are. Storm-worn and aging roofs don't care how nice the street looks. Hold reps to the scored route, spot-check their dispositions against the list they were given, and treat "I didn't get to the lower part of the route" as a coaching moment, not a footnote. The whole system only works if the crew actually knocks the doors the data chose.

The knock itself: what to say when the door you knocked is actually due

Good targeting means a higher share of your doors have a real reason behind them. That changes the pitch — you're no longer manufacturing concern, you're surfacing something real. The tone that works is calm, specific, and consultative, because you can afford to be when you're knocking houses that are actually candidates.

The opener

Keep it short, specific, and honest. Some patterns that work on a scored door:

• Storm-cluster opener: "Hi, I'm [name] with [company]. We've been checking roofs on this street after the hail that came through last month — a few of your neighbors had wear they didn't know about. I'm not selling anything today; I'm offering to take a look so you know where your roof stands. Have you had anyone up there since the storm?"
• Age opener: "Hi, I'm [name] with [company]. We work this area, and from the street your roof looks like it might be getting toward the end of its life. I can't tell for sure from down here — I'd need to actually look — but I wanted to offer a no-cost check so you're not caught off guard by a leak this winter."

Notice what these don't do. They don't claim damage you haven't verified. They don't promise an outcome. They don't say the storm "proves" anything about this specific roof. They invite an inspection, which is the only thing that actually establishes condition.

Handle the reflex objections

Most first responses are reflexes, not real positions:

• "My roof's fine." — "That's honestly the most common answer, and a lot of the time it's true. The thing about hail and age is it's invisible from the ground until it starts leaking. The check is quick and there's no obligation — worst case, you find out everything's solid and you can stop thinking about it."
• "I just had it done." — Great, and a great disqualifier. "Perfect, then you're set — who'd you go with?" Note it, move on, don't waste anyone's time. (This is also a data point: if your list flagged a recently-done roof as due, your age signal needs tuning.)
• "I'm not interested in a sales pitch." — "Fair, and I'm not here to give you one. I'm offering a look at your roof. If there's nothing wrong, I'll tell you that."

The inspection is the product

The knock's only job is to earn the inspection. On the roof is where condition is actually established — hail bruising, granule loss, mat exposure, flashing, seals, ventilation. Document it straight: photos, measurements, an honest assessment of remaining life. If the roof is fine, say so. The reps and companies that build durable books of business are the ones whose inspections are trusted because they sometimes end with "you've got several good years left."

And to stay on the right side of the line: you document conditions and provide an estimate. The homeowner owns any insurance claim and the conversation with their carrier; the insurer decides what's covered. You don't approve claims, you don't promise covered replacements, and you never frame a storm forecast or model as proof that a given roof is damaged. Those aren't only ethics — in many states, overstepping that line is how contractors get into real regulatory trouble with the department of insurance.

What a clean inspection actually documents

Because the inspection is the product, it's worth being precise about what a defensible one captures. A rep who climbs a roof, glances around, and comes down with "looks like hail damage" has produced nothing a homeowner or adjuster can trust. A clean inspection produces a record:

• Wide and close photos of each slope, so the condition of the whole roof is visible, not cherry-picked spots.
• Test squares where appropriate — a marked area (commonly a 10-by-10-foot square) with the hail impacts within it identified and counted, which gives a density read rather than a vague impression.
• Close documentation of the specific failure modes you find: bruising or mat fracture from hail, creased or lifted shingles from wind, granule loss exposing the mat, damaged or worn flashing, deteriorated pipe boots and seals.
• Collateral evidence that corroborates a storm event: dents on soft metals like gutters, downspouts, vents, AC fins, and fascia. Soft metals dent at lower energies than shingles bruise, so they're a useful tell that hail of a given size actually fell here.
• The ventilation and overall system condition, because remaining roof life depends on more than the shingle surface.
• An honest statement of remaining service life as a range, tied to what you saw.

That record is what lets you have a straight conversation with the homeowner about their options, and it's what stands behind any estimate. It also protects you: a documented inspection that honestly concludes "this roof is fine for now" is a credibility deposit you'll draw on for years in that neighborhood.

Safety is part of the system

None of this works if your crew gets hurt, and roofing has one of the highest fatal-fall rates of any trade. Inspections mean people on ladders and on steep, sometimes storm-damaged surfaces. Build the safety basics into how you canvass: proper ladder setup and footing, fall protection on steep or high roofs per OSHA guidance, no roof access in wet, icy, or windy conditions, and a hard rule that a marginal roof gets documented from a drone or from the eave rather than walked. A rep who breaks a leg on a dead-door inspection is the most expensive dead door of all. Treat field safety as part of the targeting system, not a separate topic — the right house, inspected the wrong way, still costs you.

The tools that hold the system together

A targeting system is only as good as the rails it runs on. You don't need an expensive stack, but you do need a few things to connect, or the discipline leaks out between the cracks.

A canvassing app or CRM with mapping. Reps need their ranked list on their phone as a map they can work door to door, with one-tap dispositions (not home, not interested, conversation, inspection booked, sold). If logging a door is slow or clunky, reps won't do it, and without dispositions the whole feedback loop dies. The disposition data is the fuel; make capturing it nearly frictionless.

A clean handoff from list to route to rep. Whatever produces your scored list — a spreadsheet, a data source, or both — has to flow into the field app without a rep retyping addresses. Every manual re-entry step is a place where the good targeting you built gets diluted by human shortcuts.

Inspection and estimate tooling. Aerial measurement and estimating tools turn a documented inspection into an accurate, fast proposal, which is what converts an inspection into a contract while the homeowner's attention is fresh. The faster the path from "we found it" to "here's exactly what it costs," the higher your inspection-to-contract rate.

A reporting view the manager actually watches. The metrics later in this piece only change behavior if someone looks at them weekly and coaches off them. A dashboard nobody opens is the same as no dashboard. Keep it to the handful of numbers that matter and review them on a fixed cadence.

The tools don't replace the thinking. They keep the thinking from evaporating between the office and the doorstep, which is where most well-intentioned canvassing programs quietly fall apart.

Measuring whether your targeting is actually working

If you don't measure, you can't tell good targeting from a lucky week. The metrics that tell you whether you've stopped knocking dead doors:

The one to watch hardest is knocks-per-inspection alongside inspection-to-contract. If both improve together, your targeting is genuinely working — you're knocking fewer doors and the ones you knock are better. If knocks-per-inspection falls but contract rate also falls, you may be over-filtering or your scoring is selecting the wrong thing. Tune from there.

Close the loop on every knock

The disposition data from each shift is what makes next week's list smarter. Every "just had it done" tells you your age signal misfired on that house. Every storm-cluster door that converts confirms your exposure read. Feed it back. A targeting system that doesn't learn from outcomes slowly drifts back toward random. The discipline of logging every door — even the no-answers — is what keeps the list sharp.

What pros get wrong

A few recurring mistakes, even among experienced canvassing operations:

Treating a storm ZIP as a target. A storm hit the area, so the whole ZIP gets blanketed. But hail swaths are narrow and patchy; within an "affected" ZIP, maybe a quarter of the streets took real damage. Blanketing the ZIP recreates the dead-door problem at larger scale. Target within the storm footprint, roof by roof, not ZIP by ZIP.

Confusing roof age with a date. Reps who tell a homeowner "your roof is 22 years old" and get corrected to "we did it in 2015" lose all credibility for the rest of the conversation. Speak in ranges. "It looks like it's in the older half of its life" is honest and can't be flatly contradicted.

Pitching damage before inspecting. Saying a roof is damaged because a storm came through, before anyone's been on it, is both bad selling and a real liability. The storm data points you to the inspection; the inspection establishes the facts. Keep that order and you stay honest and effective.

Over-relying on the pitch. The rep who thinks the answer is always a better open is ignoring the input side. The best closer in the world, knocking random streets, will lose to an average rep working a tight, well-scored list. Fix targeting first.

Letting the list go stale. Storm data changes weekly in active seasons; ownership and roof condition drift over months. A list built once and never updated decays. Whatever your method, rebuild on a cadence — weekly during storm season, monthly otherwise.

No disposition discipline. Crews that don't log every door can't improve their targeting because they have no feedback. The five minutes of post-shift logging is the cheapest high-leverage thing in the whole system.

Putting it together: a 30-day rollout

If you're starting from random canvassing, here's a realistic way to install this without blowing up your current pipeline:

Week 1 — Build the first list. Pick one or two neighborhoods. Pull what age and storm signals you can, even crudely, and produce a ranked list with the simple point model above. Don't aim for perfect; aim for better than random.

Week 2 — Route and run a pilot. Put one or two of your better reps on the scored list with proper clustered routes. Run it for a week. Track knocks-per-inspection and dead-door rate against your historical numbers.

Week 3 — Compare and tune. Look at where the list was right and wrong. Did "just had it done" responses cluster around a bad age signal? Did storm-flagged doors convert? Adjust the scoring weights. Decide whether the manual build is sustainable for you or whether a roof-age-and-storm data source earns its keep by giving you a current ranked list without the weekly grind.

Week 4 — Scale and systematize. Roll the scored-list approach to the full crew. Lock in the shift structure, the disposition discipline, and a rebuild cadence. Set the handful of metrics above as your weekly scoreboard.

Thirty days in, the change you're looking for isn't "we knock more doors." It's "we knock fewer doors and book more inspections," because the doors you knock are the ones that were due all along. That's what stopping dead doors actually looks like: not more effort, but effort pointed at the roofs that were going to need a roofer anyway — the ones aging out, and the ones the last storm quietly wore down.

FAQ

What exactly counts as a dead door in roofing canvassing?

A dead door is any house where the conversation had no chance regardless of your pitch, usually because the roof has no real need. The most common cause is a roof that's too new or took no storm wear, so there's nothing a homeowner would pay to fix. Other causes are wrong timing, reaching a non-decision-maker, or a renter. The biggest lever is the no-need case, which is solved by better targeting, not by talking better.

Can I really tell which roofs are due before I knock?

You can't get certainty, but you can get a strong probability ranking, which is what matters. By stacking roof-age signals (permit records, home age, sale history, visual cues from imagery) with storm exposure (hail and wind reports) and basic ownership fit, you can sort addresses into too-new, watch, and due. Knocking the due pile first sharply raises the share of doors with a legitimate reason behind them. The estimate is always a range and a probability, never a confirmed date or confirmed damage.

How accurate is roof-age estimation from aerial imagery?

It's an estimated range, not an install date. Imagery and property data can place a roof in a life-stage band (for example, older half of its life vs. newer) by reading granule loss, fade, streaking, patching, and shingle style, combined with home age and sale history. That's plenty to prioritize a knock list. It is not precise enough to assert a specific year to a homeowner, and trying to will cost you credibility when they correct you.

Isn't door knocking dead? Why not simply buy leads instead?

Random door knocking has always had poor odds; what's dead is knocking streets blindly, not knocking itself. Targeted knocking on a scored list is one of the most cost-effective acquisition channels in roofing because you control the cost and own the relationship. Bought leads can supplement it, but they're often shared, expensive, and late. A ranked, storm-aware knock list lets you reach homeowners before anyone else and without per-lead fees.

How does storm data tell me which specific roofs to knock?

Public storm data from sources like NOAA's Storm Prediction Center and Storm Events Database shows where hail and wind were reported, which gets you to the affected area. The harder, more valuable step is going from area to individual roof, because hail swaths are narrow and patchy and wind loading varies by exposure. Per-roof storm modeling estimates how a storm likely loaded a specific roof, so you target within the footprint house by house rather than blanketing an entire ZIP.

What does RoofPredict do, and what doesn't it do?

RoofPredict estimates a roof-age range per address from aerial imagery and models storm physics for each individual roof, then ranks houses so crews knock the roofs aging out and the roofs a storm actually wore down, in route order. What it doesn't do: it doesn't confirm damage (only an inspection does), it gives ranges and odds rather than dates and certainties, and it has nothing to do with handling, approving, or guaranteeing insurance claims. It points your inspections at the right houses; your crew's eyes and documentation do the rest.

Can I show a homeowner that a storm damaged their roof using forecast or model data?

No, and you shouldn't try. Storm models and reports indicate the odds that a roof took wear; they are not proof of damage to a specific roof. The only thing that establishes condition is an inspection on the roof. Use the data to decide where to inspect, then document what you actually find. Presenting a forecast as proof of damage is misleading and, in many states, can put you crosswise with insurance regulations.

What's the right way to talk about insurance without crossing a line?

Stay in your lane: you document the roof's condition and provide an estimate. The homeowner owns the claim and the relationship with their insurer, and the insurer decides what's covered. Don't promise covered replacements, deductible outcomes, or claim approvals, and don't act as the homeowner's representative to the carrier unless you're properly licensed to. Keeping this clean protects both your customer and your standing in the market.

How many doors should a rep knock per inspection with good targeting?

There's no universal number because it depends on market, season, and how aggressive your scoring is, so track your own baseline and watch the trend. The point isn't a magic ratio; it's that knocks-per-inspection should fall as targeting improves while inspection-to-contract rate holds or rises. If both move the right way together, your list is working. If knocks-per-inspection falls but contracts also fall, you're likely over-filtering or selecting the wrong signal.

How often do I need to rebuild my target list?

Rebuild on a cadence tied to how fast your inputs change. During active storm season, weekly is reasonable because new hail and wind events reshuffle which roofs just got worn down. In quieter periods, monthly is usually enough to catch ownership changes, recent reroofs, and roofs crossing age thresholds. A list built once and never refreshed slowly decays back toward random, which is the problem you were trying to escape.