Roofing Canvassing: How to Stop Chasing Low-Value Doors

Ask a roofing sales manager where the money goes and most of them point at marketing spend or material costs. The real leak is quieter than that. It is the four hours a canvasser spends each afternoon walking past roofs that will never need replacing for another decade, knocking on doors where the homeowner re-roofed eighteen months ago, and burning gas to cover a subdivision that got built in 2019. The doors themselves are free. The hours, the payroll, the truck miles, and the rep's morale are not. A "low-value door" is any door where the roof underneath cannot realistically turn into a job in your sales cycle, and the average canvasser knocks dozens of them a day without ever knowing it.

This is a solvable problem, and it does not require a bigger team or a bigger ad budget. It requires deciding which roofs are worth a knock before anyone laces up. Below is the full operating system: how to define a low-value door precisely, how to read a street before you walk it, how to score and rank houses by the two signals that actually predict a replacement (age and storm exposure), how to route a day so reps hit density instead of distance, how to script the door so a green rep sounds like a fifteen-year vet, and how to measure whether any of it is working. It is written for owners and sales managers running residential outbound, whether you knock doors, mail, or both. Everything here is operational, drawn from how the best canvassing operations actually run, and the numbers are illustrative of how the math works on a typical street, not survey statistics.

What a low-value door actually is

Before you can stop chasing low-value doors, you have to be able to name one on sight. Most crews use a fuzzy definition ("a bad door") that collapses three very different problems into one. Separate them and each becomes fixable.

1. The roof can't buy. The single biggest category. The roof is too new to fail, was recently replaced, or is in genuinely good condition for its age. No amount of sales skill turns a four-year-old architectural shingle into a re-roof this year. This is the door you most want to eliminate before the rep ever walks up, because it is invisible from the truck and the rep wastes a full knock to discover it.

2. The homeowner can't buy. The roof might be due, but the occupant is a renter with no authority, the property is a bank-owned vacancy, or the household genuinely cannot finance the work and no financing option closes the gap. Less common than category one, but it eats time in specific neighborhoods (heavy rental density, certain multi-unit pockets).

3. The door can't be worked. No-knock ordinances, gated communities you can't enter, aggressive HOAs, or a registry the homeowner is on. Knocking these is worse than low-value; it is a compliance and reputation risk.

The rest of the playbook below is built mostly around category one, because that is where the volume is and where data changes the game. But keep all three in mind, because a good route plan routes around categories two and three as well.

The hidden cost of one wasted knock

Roofers underprice their own time because the door is free. Put a real number on it. Say a canvasser is paid a blended cost of $22/hour once you load payroll taxes and a modest base. A thorough knock-and-conversation cycle, including walking to the door, the pitch, and the walk to the next house, runs three to four minutes on a productive street and longer when nobody answers and they wait. Call it four minutes fully loaded. That is roughly $1.47 of payroll per door, before gas, before the manager's time routing them, before the opportunity cost of the good door they didn't reach because they were standing at a bad one.

Now scale it. A rep knocking 60 doors a day where 40 are category-one dead roofs is throwing away about $59 of labor a day on doors that had no chance. Across a four-person crew over a 250-day work year that is north of $58,000 in payroll spent reaching roofs that could not become jobs. That is a real hire. That is a truck. And it understates the loss, because the opportunity cost (the due roofs never reached) is usually worth far more than the wage.

The entire point of targeting is to move that 40-out-of-60 ratio. You will never get it to zero. Getting it to 15-out-of-60 roughly triples the rate at which a rep stands in front of a roof that can actually buy, with no change in hours worked.

The two signals that predict a re-roof

If you only had two pieces of information about a house before knocking, which would you want? After everything, it comes down to two, and almost every effective targeting system is some combination of them.

Signal one: roof age

Asphalt shingles, which dominate U.S. residential roofing, have a usable service life that depends on the product. A basic 3-tab shingle commonly carries a 20-to-25-year manufacturer warranty and often performs below that in harsh climates; architectural (dimensional) shingles commonly carry 25-to-30-year warranties; premium and designer products longer. The NRCA and shingle manufacturers are explicit that warranty length is not the same as real-world service life, which is shortened by ventilation problems, sun exposure, installation quality, and storms. As a working rule, a roof enters the replacement conversation somewhere in the 15-to-20-year band and the conversation gets serious past 20.

The trouble is that roof age is not the same as house age, and almost everyone confuses the two. Zillow, the county assessor, and Google all show you year built. A 1994 house has had its roof replaced once or twice already. That re-roof is invisible to every public-records source, because nobody updates the assessor when they install shingles. So "year built" is a weak proxy that systematically mis-targets the exact homes that matter most: the older homes whose roofs were already replaced (now young) and the mid-age homes whose original roof is quietly aging out.

What you actually want is an estimate of when the current roof was installed, expressed as a range ("roughly 18 to 22 years") rather than a false-precise date. Aerial and satellite imagery captured over time can reveal a re-roof event (the roof visibly changes color and texture between two image dates), and the visual condition of the shingle surface correlates with age. This is why imagery-based roof-age estimation beats public records: it sees the roof, not the building permit.

Signal two: storm exposure, modeled per roof

The second signal is what the weather actually did to that roof. Most roofers treat this as a map: "it hailed in this ZIP, so we canvass this ZIP." That is better than nothing and far too coarse. A hail swath is not uniform. Within a single "hit" ZIP, stone size, fall angle, and storm motion mean one street took 1.75-inch hail driven at a steep angle and the next street over took pea-sized hail that did nothing to a roof. Wind is even more directional: a roof's leeward slopes can be untouched while the windward slopes lost tabs.

The authoritative public sources for storm data are the NOAA Storm Prediction Center and the National Weather Service (storm reports, warnings, and the Storm Events Database), plus radar-derived hail products. These tell you where storms occurred and give estimated hail sizes. They do not, by themselves, tell you what happened to a specific roof, because the same storm produces very different damage on a 22-year-old 3-tab roof than on a 3-year-old impact-rated shingle. The IBHS research on hail and impact-rated roofing makes this concrete: roof age, shingle type, and impact rating dramatically change whether a given hail event causes functional damage.

So the high-value version of the storm signal is not "did it hail here" but "given the storm that passed and the likely age and type of this roof, how probable is meaningful damage?" That is a per-roof estimate, expressed as odds, never as proof. You confirm with an inspection. But it lets you rank a street so the canvasser starts at the houses where age and storm history stack, which is exactly where the due roofs cluster.

Why you need both, not either

Age alone over-targets: a 20-year-old roof in a region that hasn't seen weather in years may keep going. Storm alone over-targets the other way: a fresh roof in a hail swath usually shrugged it off. Stack them and the signal sharpens. The roofs that are both old and storm-worn are your A-list. The roofs that are old but un-stormed are your steady B-list (age will get them eventually). The roofs that are storm-hit but young are usually a pass unless the storm was severe. The roofs that are young and un-stormed are the low-value doors you are trying to stop knocking.

How to read a street before you walk it

You will not always have software in hand, and a sharp rep can pre-qualify a block visually. Train your crew on these tells so even a manual canvass skips obvious dead doors. None of these are certainties; they are probabilities that shift where a rep spends the first knock.

Visual age and condition cues from the curb

• Granule loss and color fade. A roof that looks flat, gray, and "bald" in patches versus a roof with crisp color and visible granule texture. Fade is uneven and tells you sun exposure history.
• Curling, cupping, and lifted tabs. Edges turning up at the corners (curling) or the center (cupping) signal an aging or heat-stressed shingle. Visible from the ground on a sunlit slope.
• Streaking and biological growth. Dark streaks (often Gloeocapsa magma algae) read as "old and neglected" to a homeowner even when it is cosmetic; it correlates loosely with roof age and is a conversation opener.
• Sagging ridgelines or deck movement. A ridge that dips or a deck that visibly waves indicates deeper, longer-term problems. Higher-value door.
• Mismatched or obviously new shingles. Bright, uniform color and sharp edges mean a recent re-roof. This is your category-one tell to skip. A whole street of these is a development you should route around.
• Patches and tarps. Active problem, motivated owner. High value, knock now.

Neighborhood-level reads

• Build-era uniformity. A subdivision built in one phase tends to re-roof in waves, because the original roofs age out together. Find the era that puts the original roofs at 18-to-25 years and you have found a street where many roofs are due at once. This is the single best manual heuristic for density.
• Pride-of-ownership signals. Maintained landscaping, newer cars, fresh paint correlate with owners who will invest in the home and can finance work.
• Rental indicators. Multiple mailboxes, parking patterns, deferred exterior upkeep. Flags category two (can't-buy occupant); de-prioritize or route around.

Teach reps to make these reads from the truck or the sidewalk before committing a knock. The goal is to walk past the new roof and ring the bell at the bald one next door, instead of working the street in mechanical order.

Building a worth-it list: the data approach

Visual reads help, but they don't scale and they don't pre-plan a route. To genuinely stop chasing low-value doors, you score the houses before the day starts and hand reps a ranked list. There are three ways to get there, in increasing order of leverage.

Approach 1: Public records filtering (cheap, weak)

You can pull parcel data with year-built from county assessors or a data vendor and filter to homes built in your target era. This is better than knocking blind and costs little. Its fatal weakness is the one named above: year built ignores re-roofs, so you will still knock a lot of already-replaced roofs and skip aging original roofs on older homes. Use it as a floor, not a finish.

Approach 2: Manual imagery review (accurate, slow)

A disciplined office person can open aerial imagery for a target area and eyeball each roof for age and condition, flagging the worn ones. This produces a genuinely good list. The problem is throughput: reviewing a few thousand roofs by hand is days of work, and storm exposure still has to be cross-referenced against weather data manually. It does not scale to a real territory, and it does not refresh after each storm.

Approach 3: Imagery + storm modeling at scale (the leverage move)

The modern version automates approaches one and two and adds the storm signal: imagery-derived roof-age ranges plus per-roof storm modeling, scored across every house in an area and ranked. This is the category RoofPredict sits in, and it is worth being precise about what it does and does not do.

What it does: for the homes in an area, it estimates a roof-age range per address from aerial imagery, models the hail and wind each roof actually took (rather than the whole ZIP it sits in), and produces a per-roof priority score so you can rank streets and routes and enrich your own CRM or mailing list with age and storm signals. The honest framing matters here: roof age comes back as a range, not an install date, and storm exposure comes back as odds of meaningful damage, not a guarantee of damage. You still inspect to confirm. What it removes is the guessing about where to point the crew first, which is exactly the low-value-door problem.

What it does not do: it is not a lead-buying service and does not hand you homeowners who raised their hand; it ranks your own outbound so you knock and mail the right doors. It does not measure the roof for a bid (that is EagleView, HOVER, Roofr territory, a different job entirely) and it does not confirm damage from the curb. It sharpens targeting; your reps and inspections still do the selling and the verifying.

The practical workflow looks like this: define the area, pull the ranked list, push the A and B houses into your CRM or route app as the day's targets, and enrich any existing list (old estimates, past customers, a purchased mailing list) with the age and storm fields so you can sort it. That last move (enriching a list you already own) is often the highest-return use, because the houses are already in your book and the only thing missing was knowing which ones are due now.

A worked example: rebuilding one rep's day

Numbers make this concrete. Take a rep, call him Marcus, working a mixed older neighborhood. Below is the math, illustrative of how the ratios move, not a study.

Before targeting. Marcus works the street in order, house by house. In a 4-hour shift he knocks 60 doors at 4 minutes each. The neighborhood is genuinely mixed, so roughly:

• 38 doors: roof too new or recently replaced (category-one dead). Wasted.
• 8 doors: renter or can't-buy. Mostly wasted.
• 14 doors: roof plausibly due. Real conversations.

Of the 14 real conversations, the usual funnel applies: maybe 5 to 6 agree to a free inspection, a few of those convert downstream. Marcus spent about 62% of his knocks on doors that never had a chance.

After targeting. Same 4 hours, same 60 knocks, but the route is a ranked list of A and B houses (old roofs, storm-worn first) with the new-construction blocks and the rental-heavy pocket routed around. Now:

• 12 doors: still misses (the data is a range, not certainty; some "due" roofs are fine, some occupants can't buy).
• 48 doors: roof plausibly due. Real conversations.

Marcus didn't work harder or longer. He stood in front of a roof that could buy roughly 48 times instead of 14. Even if his per-conversation close rate stays identical, his inspection count and his job count scale with the number of qualified conversations, which more than tripled. That is the entire thesis of stopping low-value doors: you are not improving the pitch, you are improving the denominator.

There is a second-order effect that owners undervalue. Marcus is no longer eating 38 rejections from people who don't need a roof. He is having conversations that go somewhere. Rep churn in canvassing is brutal, and the number-one driver is the grind of pointless rejection. A green rep who knocks the right doors closes something in week one, makes money, and stays. Targeting is a retention tool disguised as a productivity tool.

Routing: density beats distance

A ranked list is only half the win. The other half is order. The best target list in the world is wasted if the rep drives 11 minutes between every good door. Routing is where you reclaim the windshield time.

Cluster, then sequence

Work in clusters, not sprawl. Group your A and B houses into tight geographic pockets and assign a rep a pocket for the shift, not a sprawling "territory." The goal is the highest density of qualified doors per walking minute. A canvasser who can walk from one due roof to the next without crossing three dead blocks is dramatically more productive than one zig-zagging a wide area.

A practical daily routing workflow

1. Pull the ranked list for the working area (A and B houses).
2. Drop pins in a route app (many crews use a mapping or canvassing app; even Google My Maps works to start).
3. Cluster into pockets of roughly 40 to 70 qualified doors each — about one rep-shift.
4. Sequence within the pocket to minimize backtracking: serpentine the streets, do both sides, don't cross and re-cross.
5. Set a knock window matched to answer rates (late afternoon into early evening and weekend mornings answer best; avoid dinner and post-dark for both etiquette and safety).
6. Route around the no-knock zones, gated communities, and rental-dense blocks identified in your reads.
7. Leave a tail for the next session: when a pocket is worked, the adjacent pocket is already pinned.

Density math

If reducing windshield and dead-door time lifts qualified knocks per shift from, say, 14 to 45, you have roughly tripled productive output without adding a single labor hour. The cost was an hour of planning the night before. This is the highest-ROI hour a sales manager spends all week.

The door: making a green rep sound like a vet

Targeting puts your rep in front of the right roof. The knock still has to land. The advantage of a data-driven target list is that the rep can open with something specific and true about that house, which immediately separates them from every generic "we're doing roofs in the neighborhood" pitch the homeowner has heard. Specificity earns the first 20 seconds.

A clean, honest opener

The structure that works: a specific observation, a low-pressure reason for the visit, and an easy yes.

"Hi, I'm Marcus with [Company]. We're a local roofing crew. I'll be straight with you — we look at roofs in the area, and yours looks like it's getting up there in years, probably in the range where it's worth a set of eyes before the next big storm. No charge and no obligation. Would it be alright if I did a quick visual and took a few photos so you know where it stands?"

Why it works: it is honest (a range, not a fake "your roof is 19 years old"), it leads with the homeowner's interest (know where you stand), and the ask is a free, low-commitment inspection, not a sale. A rep who knows the roof is genuinely aged isn't bluffing, so they hold the line under a homeowner's skepticism.

What not to say (and why it costs you)

Green reps reach for urgency and over-promise. Two traps:

• False precision. "Our records show your roof is exactly 17 years old." You don't know that; you know a range. If the homeowner re-roofed five years ago, you just torched your credibility on the doorstep. Say "it looks like it's in the range where..." and you're never wrong.
• Over-promising on storms. "You've got hail damage, the insurance will cover a new roof." You cannot see that from the sidewalk and you should never promise it. See the compliance section below — this language is worse than bad sales, it can be illegal.

The leave-behind

Give the homeowner something tangible that reinforces the specific, professional read: a branded one-pager or a homeowner report showing the roof's estimated age range and storm history, with a QR code to your reviews and a way to schedule. It makes a new rep look like an established operation and gives the homeowner something to act on after the door closes. The point is not a slick brochure; it is proof of specificity — you knew something about their roof.

The storm-damage door: documentation, not claims handling

A large share of canvassing happens after weather, and the conversation turns to insurance fast. This is where roofers get themselves in real legal trouble, so it deserves its own section. The targeting principle is the same (knock the roofs the storm actually wore out, ranked by age and modeled exposure), but the script has a bright line you cannot cross.

The bright line: what a roofer may and may not do

A roofing contractor may inspect a roof, thoroughly document storm damage with photos and measurements, and prepare an accurate repair estimate (commonly Xactimate-aligned) for the work they would perform. They may state facts about their own scope to the homeowner. That is legitimate, valuable, and exactly what your documentation should produce.

A roofing contractor may not, for compensation, negotiate or "handle" or adjust the homeowner's insurance claim, interpret the policy or what is covered, promise a specific payout or that the claim will be approved, promise that the deductible is waived/absorbed/covered, advertise a "free roof," or represent the homeowner against the insurer. Those activities are unlicensed public adjusting in most states, and enforcement is real — the Texas Supreme Court's 2024 Stonewater Roofing v. Texas Department of Insurance decision upheld that even advertising yourself as an insurance "specialist" who handles claims can violate the public-adjusting and dual-capacity rules. Many states treat deductible "absorption" as insurance fraud outright.

The do-not-say list, taught to every storm canvasser

Drill these out of your reps' vocabulary. Each one is a liability:

• "We'll get your claim approved" / "We handle the whole claim for you."
• "We'll negotiate with / fight / beat your adjuster."
• "We'll maximize your settlement" / "recover every dollar."
• "Your deductible is waived / we'll cover it / you pay nothing."
• "Free roof."
• "This is definitely covered" / "the insurance will pay for a full replacement."
• Calling yourself a "claims specialist" or "public adjuster."

What to say instead (safe and still effective)

The safe frame is powerful because it is honest and it positions you as the documentation expert:

"After a storm like that, it's worth getting the roof documented properly. What I do is a thorough inspection, photograph any damage, and put together an accurate, itemized repair estimate for the work. I hand that documentation to you. You decide whether to file with your insurer, and your insurer decides what's covered — that part stays between you and them. But you'll have the facts in hand either way."

This captures the exact homeowner intent (they want help with a storm-damaged roof) and delivers genuine value (rigorous documentation and an accurate estimate) while keeping you on the right side of the line. You document thoroughly, you write an accurate estimate, you hand it to the homeowner. The homeowner files. The insurer decides coverage. That is the whole compliant workflow, and it is also genuinely the most useful thing you can do for them.

Where targeting fits the storm door

The reason per-roof storm modeling matters here is that it points the documentation effort at the roofs most likely to actually have storm damage: old roofs (which are damaged more easily) in the parts of the swath that took the real hail or wind, rather than the whole "hit" ZIP. You still inspect every one to confirm — odds are not proof — but you spend your inspection hours where they pay off instead of climbing fresh roofs that the storm didn't touch. The list says which roofs likely qualify by age and storm; your inspection and documentation do the rest. RoofPredict produces that age-plus-storm ranking; it never touches the claim.

Handling the objections a targeted door still throws

Knocking a genuinely due roof does not mean the homeowner says yes. It means the objections you get are real ones worth working, instead of the flat "my roof's fine, it's three years old" that ends a low-value-door conversation in two seconds. Here are the objections that come back from a well-targeted door and how to keep them moving toward a free inspection (the only thing you're asking for at the door).

"I just had it looked at / it's fine."

Don't argue the roof. Argue the second opinion. "That's good to hear — a lot of roofs that look fine from the curb have wear you only catch from up close, especially around the flashing and the valleys. Since I'm already here, a quick visual and a few photos costs you nothing and you'll have a dated record either way. If it's solid, I'll tell you it's solid." The honesty ("I'll tell you it's solid") disarms the assumption that you're just there to manufacture a problem.

"I'm not interested / not right now."

This is reflex, not a decision. Acknowledge and shrink the ask. "Totally fair, I'm not here to sell you a roof today. All I do is take a look so you know where it stands before the next storm season — knowing is free, and it's yours to do whatever you want with." You are reframing from "buy a roof" (a big, scary commitment) to "get information" (a small, safe one). If they still pass, leave the one-pager and move on; a worked door with a leave-behind is not a wasted door.

"How much is this going to cost me?"

Never quote a roof price on the doorstep — you haven't measured it and you'll either scare them or under-bid yourself. "I can't give you a real number until I see the roof, and I won't make one up. The inspection itself is free. If it turns out you do need work, I'll put together an itemized written estimate and we'll talk about whether it makes sense and how to handle it." The word "itemized" signals professionalism and the deferral keeps you honest.

"I can't afford a new roof."

This is a genuine concern, especially on age-driven (non-storm) doors. The compliant answer is about options, not promises. "Most folks don't pay cash for a roof. There are financing options that spread it out, and if there's been storm damage, your own insurance is a separate path you'd file on. Either way, step one is just knowing the condition — there's no cost to find out." Note what this does not do: it does not promise insurance will pay, does not promise approval, and does not touch a deductible. It names that paths exist and routes back to the free inspection.

The pattern across all four: acknowledge the objection, shrink the ask to a free inspection, stay honest about what you don't know, and never over-promise. A targeted list gives reps the confidence to hold these lines, because they actually believe the roof is due.

Keeping the list clean: hygiene and re-knock rules

A target list is a living asset, not a one-time pull. The operators who get the most out of targeting treat the list like inventory and keep it clean. Sloppy list hygiene re-introduces low-value doors through the back door — reps re-knock worked houses, chase roofs that got replaced last month, or burn a neighborhood by hitting it too often.

Status-tag every door

Every door a rep touches gets a disposition the same day: not-home, not-interested, inspection-set, inspected-no-deal, sold, do-not-knock, or already-replaced. Without dispositions you have no idea which doors are worked, and reps waste shifts re-covering ground. A canvassing app or even a disciplined spreadsheet handles this; the rule is that no door closes the day untagged.

Re-knock windows

Not-home is not no — it is the most common outcome and many of those roofs are still due. Set a re-knock window (commonly a few attempts spaced across different days and times before retiring a not-home) so you catch the working homeowner you keep missing at 4 p.m. But cap it: a house knocked five times with no answer goes to a long-cooldown bucket, not the daily route. Hammering the same dead-quiet house is its own kind of low-value door.

Suppress the false positives fast

When a rep confirms a roof was actually replaced recently (the data said old, the curb says new), that disposition should immediately suppress the address from future routes and, ideally, feed back to sharpen the targeting. Roof-age estimates are ranges; a small fraction will be wrong in both directions. The fix is a fast feedback loop, not abandoning the signal.

Re-rank after every storm

Storm exposure changes the rankings. After a significant hail or wind event in your area, re-pull and re-cluster: roofs that were B-list (old, un-stormed) may jump to A-list (now old and storm-worn), and the documentation conversation becomes timely. Stale storm rankings are how crews end up canvassing last season's swath while this week's swath sits unworked.

Compliance you can't ignore while canvassing

Targeting reduces low-value doors, but a fine or a cease-and-desist is a very expensive door. Build these into your canvassing program from day one.

Do-not-knock and solicitation rules

Many municipalities require a solicitation permit for door-to-door sales and maintain no-knock registries homeowners can join. Some have time-of-day restrictions and no-soliciting-sign rules with teeth. Before you deploy a crew in a new city, check the municipal code and the local registry, and load those addresses and zones into your route app as hard exclusions. A rep who knocks a registered no-knock house is a complaint waiting to happen and, in some jurisdictions, a fine.

The federal layer

The FTC's Cooling-Off Rule gives consumers the right to cancel certain sales of $25 or more made at their home within three business days, and you must provide the required notice. Telemarketing follow-up has its own rules (the Telemarketing Sales Rule and Do-Not-Call provisions). None of this stops you from canvassing; it shapes how you paper the deal and follow up.

Worker safety on the roof

The inspections your canvassing generates put people on roofs. OSHA's fall-protection requirements for residential construction are not optional, and roofing is consistently among the most dangerous trades for falls. Your inspection and production workflow has to include fall protection. It is both a legal duty and, frankly, a recruiting and retention argument when reps see you take their safety seriously.

Measuring whether it's working

If you can't measure it, you can't prove the targeting paid off, and you can't tune it. Track a small set of metrics weekly. Resist the urge to track 30; track these.

Core canvassing KPIs

The metric that proves the low-value-door fix is the qualified-conversation rate. If targeting works, knocks per shift hold steady while qualified conversations climb. If knocks crater, your route is too sparse (clusters too small or too far apart). If qualified conversations don't climb, your targeting data or your reads are off.

A simple weekly review cadence

1. Pull the numbers for each rep and the crew average.
2. Compare to the pre-targeting baseline you captured before you changed anything. (Capture a baseline first — two weeks of "work the street in order" — so you have something to measure against.)
3. Find the drop-off stage. If inspections set are high but jobs are low, the problem is the bid, not the door. If conversations are high but inspections low, fix the script.
4. Re-pull the target list after any significant storm and re-cluster. Storm exposure changes the rankings.
5. Retire worked pockets and feed in fresh clusters so reps never re-knock a dead zone.

A/B testing your way to a better list

If you want hard proof, split it. Run two reps of similar skill: one on a ranked target list, one working streets in order in a comparable neighborhood, for two weeks. Compare qualified-conversation rate and cost per booked inspection. The gap is your ROI on targeting, in your market, with your crew. Most operators who run this stop arguing about it after the first two weeks.

Putting it together: a 30-day rollout

Here is the sequence to install all of this without blowing up your current pipeline.

Week 1 — Baseline and define. Capture two weeks of current numbers (or at least one week) on the KPIs above so you have a before. Write your own definition of a low-value door for your market. Train reps on the curb reads. Check municipal solicitation and no-knock rules for your active cities and build the exclusion list.

Week 2 — Build the first target list. Pick one neighborhood. Get a ranked list of houses by roof-age range and storm exposure — whether you build it by manual imagery review or pull it from a per-roof scoring source like RoofPredict — and enrich it with your own CRM (flag past customers and old estimates already in the area). Cluster into rep-shift pockets and pin a route.

Week 3 — Run it and compare. Work the targeted pockets. Keep one rep on the old "work the street" method in a comparable area as a control. Track the qualified-conversation rate daily. Drill the honest opener and the do-not-say list, especially if you're in a post-storm market.

Week 4 — Tune and scale. Review the gap between targeted and control. Fix whichever funnel stage is leaking. Expand the targeted approach to the rest of the crew, retire the worked pockets, and pull the next batch of clusters. Set the weekly review cadence as a standing meeting.

By day 30 you have replaced "knock everything and hope" with "knock the roofs that can buy, in a tight route, with an honest, specific opener, and measure it." That is the whole system, and it compounds: every storm re-ranks your list, every worked pocket feeds the next, and every retained rep gets sharper on the curb reads.

The bottom line

Low-value doors are not a discipline problem or a hustle problem. They are an information problem. Reps knock new roofs because they cannot see roof age from the truck, and they over-cover storm ZIPs because a hail map can't tell them which roofs the storm actually wore out. Fix the information — roof age as a range from imagery, storm exposure modeled per roof, the two stacked and ranked — and the wasted knocks fall away on their own. You don't need a better pitch or a bigger crew first. You need to put your existing crew in front of roofs that can buy.

That is the job RoofPredict was built to do: tell you which roofs are due, house by house, by roof-age range and the storms each roof actually took, so you can rank your knock list, plan tight routes, and enrich the list you already own — your old estimates, your past customers, your mailing list — with age and storm signals. It is not a lead service and it won't measure the roof for your bid or confirm damage from the curb; the range is a range and the storm read is odds, not proof, so you still inspect. What it removes is the guessing about where to point the crew. If you want to see how your own streets rank, hand us a roof you already know the age of and check whether we nailed it — then point your reps at the ones that can actually buy.

FAQ

What exactly counts as a "low-value door" in roofing canvassing?

A low-value door is any door where the roof underneath cannot realistically become a job in your sales cycle. There are three kinds: the roof can't buy (too new, recently re-roofed, or in good condition), the homeowner can't buy (renter, vacant, or genuinely can't finance), and the door can't legally be worked (no-knock zone, gated, or on a registry). The biggest category by far is the first one, because a new roof is invisible from the truck and the rep burns a full knock to discover it.

Why can't I just use Zillow or county records to find old roofs?

Because those sources show year the house was built, not when the roof was last replaced. A 1990s home has almost certainly been re-roofed once or twice, and nobody updates the assessor when shingles go on. So year-built systematically mis-targets the exact homes that matter: older homes that were already re-roofed (now young) and mid-age homes whose original roof is aging out. You want an estimate of the current roof's age, which imagery can reveal and public records cannot.

How accurate is an imagery-based roof age estimate?

It is an estimate expressed as a range (for example, "roughly 18 to 22 years"), not an exact install date. Aerial imagery captured over time can reveal a re-roof event because the roof visibly changes color and texture between image dates, and surface condition correlates with age. Treat it as a strong targeting signal that tells you which houses are worth a knock, then confirm the actual condition with an on-site inspection. Never quote a homeowner a precise age you can't back up.

If a storm hit my whole ZIP, why shouldn't I canvass the whole ZIP?

Because a hail swath is not uniform. Within one "hit" ZIP, one street took large, steeply driven hail and the next took pea-sized stones that did nothing. Wind is even more directional. And the same storm damages a 22-year-old roof far more easily than a 3-year-old impact-rated one. Modeling the storm per roof, combined with roof age, points your reps and inspections at the roofs the storm actually wore out instead of the whole ZIP, so you waste far fewer knocks and climbs.

What should a green canvasser say at the door to sound credible?

Lead with something specific and true about that house, then make a low-pressure ask. Something like: "Your roof looks like it's getting up there in years, probably in the range where it's worth a set of eyes before the next big storm — no charge, no obligation, can I take a quick look and a few photos?" Specificity earns the first 20 seconds and separates you from the generic "we're doing roofs in the neighborhood" pitch. Use a range, never a fake exact age, so you're never caught out by a recent re-roof.

Can I tell a homeowner their insurance will cover a new roof?

No. You can't see that from the sidewalk, and promising a specific outcome (approval, a full replacement, or a waived deductible) can cross into unlicensed public adjusting or even fraud depending on your state. The safe and still-effective frame: "I'll inspect the roof, document any storm damage thoroughly, and write you an accurate repair estimate. You decide whether to file, and your insurer decides what's covered." You document, the homeowner files, the insurer decides. That keeps you legal and is genuinely the most useful thing you can do.

What are the words I should ban from my storm canvassers' scripts?

Drill these out: "we'll get your claim approved," "we handle/negotiate/fight the adjuster," "we'll maximize your settlement," anything about waiving, covering, or absorbing the deductible, "free roof," "this is definitely covered," and calling yourself a "claims specialist" or "public adjuster." Each one is a liability — several are unlicensed-public-adjusting or fraud risks in many states. Replace them with documentation language: you inspect, photograph, and write an accurate estimate, then hand it to the homeowner.

How do I measure whether targeting is actually reducing wasted knocks?

Track the qualified-conversation rate: the percentage of knocks that reach a real conversation with someone whose roof is plausibly due. If targeting works, knocks per shift stay roughly flat while qualified conversations climb. Pair it with cost per booked inspection and cost per job. Capture a baseline for a week or two before you change anything, then compare. For hard proof, run one rep on a ranked list and one working streets in order in a comparable area for two weeks and compare the two.

Do I need to worry about permits or no-knock laws when canvassing?

Yes. Many cities require a solicitation permit for door-to-door sales and maintain no-knock registries, with time-of-day rules and no-soliciting-sign enforcement. Check the municipal code and registry before deploying a crew in a new city and load those addresses and zones into your route app as hard exclusions. Federally, the FTC Cooling-Off Rule gives consumers a three-day right to cancel certain at-home sales of $25 or more, with a required notice. Build these into your program from day one.

Is RoofPredict a lead service that sells me homeowners?

No. It does not sell you homeowners who raised their hand, and it won't measure the roof for your bid or confirm damage from the curb. It estimates a roof-age range per address from aerial imagery, models the storms each roof actually took, and scores houses so you can rank your own knock list, plan tight routes, and enrich the list you already own (old estimates, past customers, mailing lists) with age and storm signals. The age is a range and the storm read is odds, not proof, so you still inspect. It sharpens targeting; your reps and inspections do the selling and verifying.