How to Find Old Roofs in a Neighborhood (A Roofer's Field Playbook)

Every neighborhood you drive through is two streets at once. There's the street the homeowner sees — mowed lawns, parked trucks, kids' bikes. And there's the street you should see: a row of roofs at wildly different ages, some installed last spring, some quietly past their service life, a handful that took a beating in a storm three years ago and nobody's looked at since. The contractors who grow without burning cash have trained themselves to read that second street. They don't knock the block. They knock the roofs that are actually due.

Most crews never learn to do this on purpose. They knock everything, mail everything, and call the 1–2% who say yes a "good neighborhood." That works, in the same way a slot machine works. You can do far better than random with about a day of learning and a repeatable process — reading aerials, pulling the right public records, and recognizing the half-dozen ground-level tells that separate a 22-year-old roof from a 6-year-old one at 25 mph from your truck.

What follows is the working method: how to find old roofs in a neighborhood from the curb, from the air, and from records — then how to stack those signals so your guys spend their hours on doors that can buy a re-roof instead of doors that just got one. It's written for owners and sales managers who already know roofing and want a sharper targeting habit, not a sales pitch. Where software earns its place in that workflow, you'll see exactly where and exactly what it can't do.

Why "old" is the signal that actually converts

Before the how, get the why straight, because it changes how you weight everything else.

A homeowner buys a new roof for roughly four reasons: it's leaking, it's visibly failing, a storm damaged it, or they're selling and an inspector flagged it. Three of those four correlate directly with age. A leak is usually an old roof that finally gave up at a flashing or valley. Visible failure — curling, granule loss, missing tabs — is age expressing itself. Storm damage lands hardest on roofs that were already brittle. Only the home sale is age-independent, and even there, the inspector's note is almost always about wear.

So when you target old roofs, you're not chasing a vanity metric. You're concentrating your outreach on the population where the buying trigger is most likely to fire soon. The math is blunt: an asphalt shingle roof in most of the country gives you somewhere around 18 to 25 years of real service, often less in high-UV or high-storm climates, sometimes more on a steep, well-ventilated north slope. The replacement window for a given roof is a band of a few years, not a single date. Your job is to find the addresses sitting inside or just ahead of that band.

The opposite mistake is the expensive one. Knock a street where half the roofs were replaced after a hailstorm five years ago and you'll burn a morning getting "we just did ours." That's not a slow day. That's a targeting failure you paid for in gas, payroll, and your rep's morale.

What "old" is worth in dollars

Run the unit economics once and the whole exercise stops feeling like busywork.

Say a re-roof nets you a few thousand dollars of margin. Say a canvasser knocks 40 effective doors in a productive afternoon and your historical close rate on a cold knock is around 1 in 50 conversations to a signed job. On a random street, that rep needs the better part of three afternoons to land one job. Now bias the same rep toward streets and houses where the roof-age band is genuinely due. You don't need a miracle — lifting the hit rate from random to even modestly informed roughly halves the doors-per-job. That's one extra job a week per rep that you were already paying for. The cost of finding old roofs is almost always smaller than the cost of not finding them.

The three layers of finding old roofs

There are exactly three places the truth lives, and good targeting uses all three:

1. The ground layer — what your eyes and your reps see from the street and the driveway. Highest fidelity on condition, lowest coverage (you can only see so many houses, and only the front slope).
2. The aerial layer — what overhead and oblique imagery shows you across a whole block at once. Great coverage and reveals slopes you can't see from the road; weaker on fine condition detail.
3. The records layer — permits, parcel data, sale dates, and year-built. Objective and searchable at scale, but full of gaps and lies of omission (a re-roof with no permit is invisible here).

Nobody layer is enough. Year-built tells you a 1998 house, but not that the roof was redone in 2016. An aerial shows a tired-looking slope, but not whether that's algae or granule loss. A curbside look nails condition on the one house you're parked at and tells you nothing about the next forty. Stack them and the false positives cancel out. Below, each layer in operating detail, then how to combine them into a route.

Layer 1: Reading a roof's age from the ground

This is the skill that pays every single day, because your reps are already standing in these neighborhoods. Teach a green canvasser to read these signs and they stop knocking blind. Here's what to look for, roughly in order of how reliable each tell is.

The high-confidence tells

Granule loss and surface sheen. A healthy asphalt shingle has a uniform, slightly rough, matte surface of mineral granules. As it ages, those granules wash into the gutters and downspout splash zones, and the asphalt mat underneath starts to show through. From the street you'll see a roof that looks shinier, darker in patches, almost "bald" in spots, with a blotchy rather than uniform tone. Glance at the bottom of the downspouts and the splash blocks — a pile of black sand-like grit there is granules, and it's one of the cleanest age tells you can read without leaving the truck.

Curling, cupping, and clawing. Old shingles lose their flexibility and the edges lift (curling) or the centers rise while edges stay down (cupping/clawing). From a distance the roof loses its crisp, flat shingle lines and takes on a rippled, restless texture — like the surface of water with a light chop. A flat, sharp-lined roof is young. A roof that looks "busy" or textured where it should be smooth is old.

Missing, cracked, or slipped shingles. Gaps, dark rectangles where a tab tore off, or a course that's visibly out of line all say the field is failing. One missing shingle is a repair; a scattering of them across the field is a roof telling you it's done.

Sagging or wavy decking. A roofline that dips between rafters or shows a gentle wave across a slope usually means moisture has been working on the deck for a while — a long-term problem that travels with old, leaking roofs.

The medium-confidence tells (read with care)

Algae streaks (the black stains). Those dark vertical streaks are a cyanobacteria (commonly Gloeocapsa magma), not dirt and not, by itself, proof of age — they're really a humidity and shade story. A 10-year-old roof on a shaded north slope can streak heavily; a 20-year-old roof in full sun can stay clean. Use streaking as a "this house has roof attention deferred" flag, not as an age measurement. It does correlate loosely, because nobody who just paid for a roof lets it go green, but don't lean your whole read on it.

Patches and color mismatch. A rectangle of slightly different shingle color is a prior repair — which tells you the roof has already had a problem and the owner chose a patch over a replacement. That's a tired roof with a frugal owner, often a good conversation.

Rusty or worn flashing and pipe boots. Look at the metal around chimneys, valleys, and the rubber boots on plumbing vents. Cracked, dry-rotted pipe boots and rusted flashing age right alongside the field and are a frequent leak origin. Visible rust streaks running down from a vent are an easy curbside catch.

The deceptive ones — what fools rookies

• Dimensional (architectural) shingles look "newer" than they are. Their built-in shadow lines read as crisp from a distance even when the roof is well-worn. Don't assume an architectural roof is young just because it has depth.
• A clean, recently power-washed roof can hide its age. Pressure-washing an old roof (which you should never do — it blasts off granules) leaves it looking refreshed for a season. Cleanliness is not youth.
• New gutters or fascia, old roof. Homeowners often replace gutters or paint trim without touching the roof. Crisp white fascia over a curling roof field is common; judge the field, not the trim.
• One bright new slope, the rest original. Partial replacements happen after localized damage. Read each slope on its own.

A 30-second curbside scoring rubric

Give your reps something they can run from the passenger seat. Score each visible roof 0–2 on five tells, add it up, and you've got a fast triage number.

A 0–2 total is a young or recently done roof — skip it. A 3–5 is a "maybe, log it." A 6+ is a roof you stop for. The point isn't precision; it's a shared language so a new hire and a 15-year vet log the same street the same way, and so you can later compare what the reps saw against what the records and aerials say.

Reading the gutters, the attic line, and the neighbors

Three more reads turn a decent canvasser into a sharp one, and none require a ladder.

The gutter and downspout story. You already know to check splash blocks for granules. Go one step further: a roof that's shedding granules heavily often shows a faint rust or dark staining streak running down the face of the gutter below a valley, where concentrated runoff carries grit and asphalt oils. Sagging or pulled-away gutters can also hint at ice-dam history or a deck that's been wet — both travel with older, leak-prone roofs in colder climates.

The roofline against the sky. Park where you can see the ridge silhouetted against the sky. A crisp, dead-straight ridge and eave line says the structure is sound. A ridge that swags in the middle, or eaves that wave, points to long-term moisture in the deck or framing — almost always an old-roof problem. This read is free and you can do it from a moving truck.

The neighborhood-cohort tell. Tract subdivisions were built in waves, and the original roofs went on within a year or two of each other. So roofs tend to age and fail in cohorts. If you see three or four roofs on a street with obvious wear and a couple already replaced, the un-replaced ones nearby are statistically likely to be the same vintage and due soon. Reading the block as a cohort — not house by house in isolation — is one of the highest-leverage habits a route planner can build, because one confirmed old roof raises your confidence on its untouched neighbors.

Mail targeting: the same signals, a different motion

If you mail rather than knock, every layer above still applies — you're just converting the candidate list into an address list for a drop instead of a route for a rep. The economics are even less forgiving of sloppy targeting, because a mail piece costs real money whether or not the roof is due, and there's no rep on the porch to course-correct a wasted send.

The single biggest lever in mail is the exclusion step. A roofer who mails an entire ZIP off year-built is paying postage to reach a pile of recently-replaced roofs. Pull recent roofing permits and drop those addresses first, then layer the aerial read, and your cost-per-piece-that-reaches-a-real-prospect drops sharply for the same budget. A specific piece beats a generic one here too: a mailer that references the home's actual roof age band or a recent storm in the area reads as "they know my house," not junk mail. Just keep the claim honest — an age range and a storm-history mention, never a fabricated "your roof is damaged" assertion you can't stand behind.

Layer 2: Reading roofs from aerial imagery

The ground layer is high fidelity but you can only see the front slope of the houses you physically pass. Aerials flip that: you can scan an entire neighborhood from a desk, see the back and side slopes the homeowner thinks nobody can see, and plan a route before anyone gets in a truck.

Where to get the imagery

• Free overhead (Google Maps/Earth satellite, Bing Maps, Apple Maps). Good enough to read field condition, see patch jobs, count slopes, and spot tarps or obviously failed sections. The catch is image age — the "satellite" view you're looking at can be one to several years old, and refresh dates vary wildly by area. A roof that looks fine in the imagery may have been replaced or may have failed since.
• Free oblique / 45° (Bing's "Bird's Eye," Google Earth's 3D and historical imagery). Obliques are gold because you see the slope face rather than only the top edge, so condition reads far better than straight-down. Google Earth's historical imagery slider is the underrated tool here — more on that below.
• Government and academic orthoimagery. Many counties and states publish high-resolution aerials through their GIS portals, and the USGS distributes national aerial imagery (the National Agriculture Imagery Program) you can browse through USGS EarthExplorer. Resolution and recency vary, but it's free and sometimes sharper than the consumer maps.
• Paid high-resolution providers. Commercial aerial-imagery and roof-measurement services sell crisp, recent, multi-angle imagery. They're built for measurement and estimating rather than age-finding, but the image quality makes condition reads easier. Know what you're paying for: that's a measurement-and-imagery category, not a "which roofs are old" category.

What aerials show that the ground can't

Back and side slopes. A homeowner keeps the street-facing slope presentable and lets the back go. The worst slope is often the one you'd never see from the road. Aerials catch it.

Patch geometry across the whole roof. From above, a prior repair or a partial replacement shows as a rectangle or a single fresher slope — instantly readable, and a strong signal the roof has a history.

Tarps, exposed decking, and active failures. A blue tarp is a leak someone hasn't fixed yet. Exposed deck or a dark patch where shingles tore off is a roof in active failure. These jump out from overhead.

Tree cover and shading context. Heavy overhanging trees mean more debris, more shade, more moisture retention, faster aging on those slopes — and they explain algae streaking so you don't misread a shaded 10-year-old as ancient.

The historical-imagery trick (the one most roofers skip)

Google Earth Pro (free on desktop) has a clock icon that opens a time slider. Drag it and you can step through every aerial capture of that address going back years, sometimes more than a decade. This is the closest thing to a free "when was this roof last replaced" tool that exists.

Here's the workflow:

1. Open the address in Google Earth Pro and click the historical-imagery clock.
2. Step backward through the captures, watching the roof.
3. Find the frame where the roof suddenly changes — a uniform fresh slope appears, color shifts, a patch vanishes. That capture date brackets a replacement: the roof was redone between that frame and the previous one.
4. If you scroll back through 12 years of captures and the roof never changes, you're likely looking at an original or very old roof — a strong "due" signal.
5. Note the bracket ("re-roofed between 2014 and 2016 capture") in your sheet.

This won't be perfect — captures are intermittent and you can miss a re-roof that happened and weathered between two frames — but it routinely turns "I have no idea" into "redone about a decade ago" for free. Do this on a few blocks and you'll start to see the storm and re-roof history of an area in the imagery itself.

Aerial scanning at neighborhood scale

Reading one house in Google Earth is easy. Reading four hundred is the bottleneck. A disciplined manual pass looks like this:

• Pull up the neighborhood in overhead view at a zoom where you can see individual roof texture.
• Work a grid, street by street, and tag any roof that reads dark/blotchy, patched, tarped, or rippled.
• Drop those into a list with the address.
• Spot-check your tagged houses in oblique and historical view to confirm before you commit a route.

This is real work — a thorough manual scan of a subdivision is a couple of focused hours — and it's exactly the part that doesn't scale by hand when you're trying to cover a whole service area. Hold that thought; it's where automated age-ranging changes the economics, covered further down.

Layer 3: Mining public records for roof age

Records give you something the eyes can't: objective dates, searchable across thousands of parcels at once. The trick is knowing which records actually speak to roof age and which ones quietly mislead you.

Year-built is a trap (and how to use it anyway)

The single most common mistake is treating year-built as roof age. County assessor data and the listing sites (Zillow, Redfin, Realtor) all show year-built, and it's tempting to say "1999 house, 26-year-old roof, knock it." That's wrong often enough to waste real money. Roofs get replaced. A 1999 house may be on its second or third roof. Year-built tells you the oldest possible roof age and nothing about re-roofs.

Use it correctly and it's still valuable: year-built sets the ceiling. A house built in 2019 cannot have a 20-year-old roof — cross it off your old-roof list immediately. Year-built is excellent for exclusion (ruling out houses too new to be due) and useless for confirmation (proving a house is due). Most importantly, the listing sites show year-built, not roof age — re-roofs are invisible to them. Don't build a campaign on a number that can't see the one event you care about.

Building permits: the closest thing to ground truth

When a roof is replaced legally, a permit is usually pulled. That permit — with a date, a description ("reroof," "tear-off and replace"), and sometimes the contractor — is the single best public record of roof age that exists. Many jurisdictions put permit data online and searchable by address, and some publish it as open data you can download in bulk.

How to work permits:

1. Find your county or city's permit portal (search "[county] building permit search" or look for an open-data portal).
2. Search by address or, better, pull or filter the dataset for permit types containing "roof," "reroof," or "re-roof."
3. A recent roofing permit (within the last several years) = exclude that house. A 20-year-old roofing permit, or a house with no roofing permit since it was built = strong "due" candidate.

The big caveat: permits are incomplete. Plenty of re-roofs happen without a permit — cash jobs, storm-chaser work, owners who didn't bother. So a missing permit doesn't prove an old roof; it just means records can't rule the house out. And portal quality is all over the map: some counties have clean searchable open data, others give you a clunky lookup one address at a time, and some have nothing online at all. Permits confirm young roofs well and old roofs poorly — the mirror image of year-built.

Property sale dates and inspection triggers

Sale history (free on the listing sites and in assessor records) is a quieter signal. A house that sold recently often had an inspection that flagged the roof, and new owners are more open to roof conversations in their first year or two than long-tenured owners who've ignored their roof for a decade. Conversely, a house owned by the same person for 20+ years with no permits is a classic deferred-maintenance profile. Tenure plus no-permit-on-record is a useful combination.

Storm and hail history

If you work storm restoration, layer in weather history. Public sources — NOAA's Storm Events Database, the National Weather Service, and the Storm Prediction Center's archives — let you see where hail and damaging wind have hit and roughly how big. Insurers and other vendors also publish or sell hail-swath maps.

The honest limitation here matters: a hail map tells you where it hailed, in a broad swath. It does not tell you which specific roofs got worn out, because impact depends on hailstone size, the angle and direction it fell, the slope orientation and pitch, the roof's age and brittleness, and shielding from trees and neighboring structures. Two roofs on the same street can take a very different beating from the same storm. So weather history narrows you to an area; it doesn't hand you the houses. Keep that distinction — it's the difference between a swath and a target list, and it's exactly where per-roof modeling earns its keep (next section).

A records-layer cheat sheet

Stacking the layers: the targeting workflow

No single signal is trustworthy alone. The power is in agreement. When the records say "old house, no recent roofing permit," the aerial shows a blotchy, patched back slope, and your rep's curbside score is a 6 — that's a roof you can knock with real confidence. When the signals disagree, you've learned something too: a tired-looking aerial on a house with a 3-year-old roofing permit is probably algae on a young roof, so skip it.

Here's a workflow a sales manager can actually run before sending a crew out.

Step 1 — Pick the area on records and weather

Start wide. Identify subdivisions where the housing stock falls into the right age band — neighborhoods built roughly 18 to 30 years ago are prime, because a lot of original roofs are aging out and the early re-roofs are themselves getting old. If you do storm work, overlay hail/wind history to prioritize areas that got hit. You're not picking houses yet; you're picking where to spend the next steps.

Step 2 — Exclude the obvious young roofs

Now subtract. Pull year-built to drop anything too new to be due, and pull recent roofing permits to drop houses that were just done. This is the cheapest, highest-value filter you'll run: every recently-roofed house you remove now is a "we just did ours" you won't pay a rep to discover in person. In a neighborhood that took a hailstorm a few years back, this step alone can cut your knock list substantially.

Step 3 — Scan the survivors from the air

For the houses that survived exclusion, scan the aerials. Use overhead to triage and oblique/historical to confirm. Tag each as young / unclear / likely-old based on surface condition, patches, and the historical-imagery bracket. You're now down to a list of addresses with a real reason to believe the roof is due.

Step 4 — Verify on the ground and convert

Route your reps to the likely-old and unclear houses, with the young ones stripped out. Reps run the 30-second curbside rubric to confirm condition, then knock. Because they're only at houses with a reason behind them, the conversation starts from "I noticed the back slope of your roof is showing some wear" instead of "hi, do you maybe need a roof." Specific beats generic every time.

Step 5 — Log what you learn and close the loop

Whatever you saw — "roof's newer than the aerial suggested," "already has bids," "interested, call in spring" — goes back into your sheet or CRM. This is the step almost everyone skips, and it's the one that compounds. Every pass makes your read of the area sharper and feeds your follow-up list.

A worked example

Take a 320-home subdivision built around 1998. Random knocking gives your rep maybe a 1–2% conversation-to-job rate across the whole tract. Run the stack instead:

• Records exclusion: year-built confirms the tract is in-band; a permit pull shows about 90 homes have a roofing permit in the last six years (a prior hail event drove a wave of replacements). Drop those 90. You're at 230.
• Aerial scan: of the remaining 230, obliques and historical imagery flag roughly 70 with clear age signals — blotchy fields, visible patches, no roof change in 12 years of captures. The other 160 look ambiguous or reasonably young.
• Route: you send reps to the 70 strong candidates first, the 160 ambiguous as a second pass.

Your rep now spends the morning at 70 doors where the roof genuinely looks due rather than 320 doors where most don't. Same labor, dramatically higher density of real conversations. You didn't manufacture demand — you stopped spending it on roofs that already got replaced.

Where software fits — and where it doesn't

Everything above works by hand. The honest problem is scale. Reading one block in Google Earth, cross-checking a permit portal, and bracketing re-roofs with the historical slider is a couple of hours of skilled desk work per neighborhood. Doing it across a whole service area, every week, is more desk time than most sales managers have. That's the gap automated tools fill — and it's worth being precise about what they can and can't do, because the category is full of overlapping claims.

Three different tools people lump together

• Measurement and imagery (aerial-measurement and roof-takeoff services): they tell you the roof's dimensions from imagery — squares, pitch, facet count — so you can estimate without climbing. They do not tell you the roof's age or which house is due. That's a different question; "measure the house" is not "which house."
• Lead marketplaces (the pay-per-lead networks): they sell you a homeowner who raised a hand, usually resold to several competitors at once, often nowhere near you. That's renting demand, not finding it on your own streets.
• Roof-age and storm targeting (the category that answers the question this page is about): tools that estimate roof age and model storm exposure per address across an area, so you get a ranked list of which roofs are likely due rather than scanning by hand.

They're not competitors so much as different jobs. You might use all three. But if your actual question is "how do I find old roofs in this neighborhood at scale," only the third one is aimed at it.

What per-roof age-and-storm modeling does

This is the niche RoofPredict is built for, so here's an honest read of what that kind of tool replaces in the manual workflow above and where its limits are.

The core idea is to do Layer 2 and the storm part of Layer 3 automatically, across a whole area, and hand you a ranked list. From aerial imagery it estimates a roof-age range per address — not an exact install date, a range, the same band-not-a-date reality you get from doing it by hand, just computed across thousands of roofs instead of one at a time. And instead of an area-wide hail swath, it models the storm exposure per roof — accounting for the fact that hail size, fall angle, slope, and shielding make neighboring roofs age differently — and pairs that with the age range. The output is the thing you were building by hand in Step 3: a sorted list of which houses on a street are most likely due, so your reps and your mail go to those first.

What that buys you, concretely:

• The neighborhood aerial scan that took two focused hours becomes a list you can act on, so coverage stops being capped by desk time.
• Mail and door-knock routes get built around "likely due" instead of "whole tract," which is the same outcome as the worked example, minus the manual scan.
• A green canvasser gets a per-house reason to knock — "the model puts this roof in an older age range and it took two notable storms" — so a new hire sounds informed without ten years of curbside reps, which is half of why new hires quit before they get good.

The honest limits — what it can't do

Keep this straight or you'll over-trust it:

• Age is a range, not a date. Imagery-based estimation gives you a band (say, an 18–22-year window), not "installed March 2004." Treat it as the targeting signal it is, and still verify condition on the ground before you quote.
• A storm model is odds, not proof. Modeling that a roof was likely exposed to damaging hail is a probability that it's worn, not evidence that it's damaged. The only proof of damage is an inspection — someone gets up there (or sends a documented assessment) and confirms it. Never present a model output as proof of damage to a homeowner or anyone else.
• It doesn't see the inside of a permit office perfectly, and it can't read a cash re-roof. Same blind spot you have by hand — an un-permitted, un-aged-out replacement can fool any remote method. Ground-truth still matters.
• It doesn't touch insurance. A roofer documents the roof's condition and writes an estimate; the insurer decides coverage; the homeowner owns the claim. A targeting tool tells you which doors to knock — it doesn't file, handle, or negotiate anything, and you shouldn't either. Keep that line clean.

Use it for what it's for: collapsing the find-old-roofs scanning into a ranked list so your people spend their hours converting instead of searching. It doesn't replace the curbside rubric or the inspection; it replaces the part where you guess which streets to drive.

Common mistakes that quietly burn money

After you've run this a few seasons, the same own-goals show up across crews. Watch for these.

Mistaking year-built for roof age. Covered above, worth repeating because it's the number one error. The listing sites show year-built; re-roofs are invisible to them. Build a mail drop off year-built alone and you'll paper a neighborhood full of recently-replaced roofs.

Reading algae as age. Black streaks are a humidity-and-shade story, not a calendar. Don't let a shaded 10-year-old roof eat your route while a clean, sun-baked 22-year-old gets skipped.

Knocking right after a hailstorm without filtering. The instinct is to flood a hit area. But a few years after a storm, a big chunk of that area already replaced. Without the permit-exclusion step you're knocking the densest concentration of "we just did ours" in town.

Trusting stale imagery. Free satellite views can be years old. A roof that looks pristine in the image may be failing now; one that looks failing may have been redone. Always sanity-check the image age and confirm on the ground before you commit real spend.

No exclusion step. Most crews only add houses ("this one looks old"). The leverage is in subtracting — removing the provably-young roofs first so every dollar after that lands on a possible buyer. Exclusion is cheaper than persuasion.

Not logging the route. If your reps don't record what they saw, you re-learn the same neighborhood every visit. The teams that compound write it down — in a sheet, in the CRM, anywhere — so next pass starts smarter.

Letting the new hire knock blind. A green canvasser with no per-house reason burns out fast — rejection with no edge is demoralizing, and rep churn is one of the most expensive line items in a roofing sales org. Hand them houses with a reason and a talking point and they stay long enough to get good.

A field checklist you can hand a sales manager

Print this, adapt it, run it before the next campaign.

Before the trucks roll (desk work):

• Pick neighborhoods in the right age band (roughly 18–30 years old housing stock).
• If you do storm work, overlay hail/wind history to prioritize hit areas.
• Pull year-built and exclude houses too new to be due.
• Pull recent roofing permits and exclude recently-replaced roofs.
• Aerial-scan the survivors (overhead to triage, oblique + historical to confirm).
• Use Google Earth's historical slider to bracket re-roof dates where you can.
• Produce a ranked address list: strong candidates first, ambiguous second.

In the field (reps):

• Run the 30-second curbside rubric on each candidate (5 tells, score 0–10).
• Stop at 6+; log 3–5 for a later pass; skip 0–2.
• Open with the specific reason ("the back slope is showing wear"), not a generic pitch.
• Note flashing/boot condition and any patches — leak-origin tells that build credibility.

After the route (close the loop):

• Log outcome and roof condition for every door into the CRM/sheet.
• Flag "newer than expected" houses so you stop re-knocking them.
• Build the follow-up list (interested-but-not-now) and schedule the next touch.
• Compare what reps saw vs. what records/aerials predicted, and tune your read.

Putting it together

Finding old roofs in a neighborhood isn't a trick or a single tool — it's a habit of stacking three honest signals. The ground gives you condition on the house in front of you. The air gives you coverage and the slopes nobody else sees. The records give you objective dates and, just as importantly, the exclusions that keep you off recently-roofed streets. Each one is wrong sometimes; together they're right often enough to roughly double the density of real conversations your reps have, on the same payroll you're already spending.

Do it by hand and you'll out-target most of your competition with nothing but Google Earth, a county permit portal, and a curbside rubric your crew shares. When the desk work outgrows the hours — when you want that ranked "which roofs are due" list across a whole service area instead of one block at a time, with storm exposure modeled per roof rather than read off a swath — that's the moment a tool built for exactly that job, like RoofPredict, stops being a nice-to-have and starts paying for itself. Either way, the principle holds: stop working the whole street. Work the roofs that are actually due, and let everything else be somebody else's wasted morning.

FAQ

How can I tell if a roof is old just from the street?

Read the field, not the trim. The strongest curbside tells are granule loss (a shiny or blotchy surface with the asphalt mat showing through, plus black grit at the bottom of downspouts), curling or cupping shingles that make the roof look rippled instead of flat, and any missing or slipped shingles. Cracked rubber pipe boots and rusty flashing age right alongside the field. Be careful with algae streaks — they're a humidity-and-shade signal, not a reliable age signal — and don't judge a roof by new gutters or fresh paint on the fascia.

Does the year a house was built tell me the roof's age?

No, and this is the most expensive mistake in roofing targeting. Year-built (shown on Zillow, Redfin, and assessor records) only tells you the oldest the roof could possibly be — it's blind to every re-roof. A 1999 house may be on its second or third roof. Use year-built to exclude houses too new to be due (a 2019 house can't have a 20-year-old roof), never to confirm that a house needs one.

What's the best free way to estimate when a roof was last replaced?

Google Earth Pro's historical-imagery slider. Open the address, click the clock icon, and step backward through years of aerial captures. The frame where the roof suddenly looks fresh, changes color, or loses a patch brackets the replacement date. If the roof never changes across a decade-plus of captures, you're likely looking at an original or very old roof. It's intermittent (you can miss a re-roof between two captures) but it's the closest thing to a free roof-age lookup that exists.

How do I use building permits to find old roofs?

Search your county or city permit portal by address, or pull the open dataset and filter for permit types containing 'roof' or 'reroof.' A recent roofing permit means exclude that house; a very old roofing permit, or no roofing permit since the house was built, flags a likely-due candidate. The big caveat: many re-roofs happen without a permit (cash and storm-chaser work), so a missing permit doesn't prove an old roof — it just means records can't rule the house out.

Is satellite imagery accurate enough to judge roof condition?

It depends on the image age. Free overhead views in Google, Bing, and Apple Maps can be one to several years old and refresh dates vary by area, so a roof may have been replaced or have failed since the capture. Oblique (45°) views like Bing Bird's Eye read condition far better than straight-down because you see the slope face. Always check the image age and confirm on the ground before committing real spend.

Why shouldn't I just flood a neighborhood after a hailstorm?

Because a few years after a storm, a large share of that area has already replaced — so without filtering, you're knocking the densest concentration of 'we just did ours' in town. Run the permit-exclusion step first to drop the recently-roofed homes. A hail map also only shows where it hailed in a broad swath; it doesn't tell you which specific roofs got worn out, since that depends on hailstone size, fall angle, slope, and shielding.

What roof age means a house is due for replacement?

Asphalt shingle roofs typically give 18–25 years of real service, often less in high-UV or high-storm climates and sometimes more on a steep, well-ventilated north slope. The replacement window is a band of a few years, not a single date. Neighborhoods built roughly 18–30 years ago are prime hunting grounds because original roofs are aging out and the early re-roofs are themselves getting old.

How is a roof-age tool different from an aerial measurement service?

They answer different questions. Aerial-measurement and roof-takeoff services tell you the roof's dimensions — squares, pitch, facet count — so you can estimate without climbing; they don't tell you the roof's age or which house is due. A roof-age and storm-targeting tool estimates an age range per address and models storm exposure per roof to rank which homes are likely due. 'Measure the house' is a different job from 'which house.'

Can software tell me exactly how old a roof is?

No — imagery-based estimation gives you a range (for example, an 18–22-year window), not an exact install date, the same band-not-a-date reality you get reading aerials by hand. Treat the range as a targeting signal that tells your reps which doors to prioritize, and still verify condition on the ground before you quote. A storm model likewise gives odds that a roof is worn, not proof of damage; only an inspection confirms damage.

Does finding storm-exposed roofs mean I can help homeowners with insurance claims?

Keep that line clean. A roofer documents the roof's condition and writes an estimate; the insurer decides coverage; the homeowner owns the claim. Targeting data tells you which doors to knock and storm exposure is modeled odds, not proof of damage — it doesn't file, handle, or negotiate anything, and neither should you. Present an inspection's documented findings, not a model output, and never frame a forecast or a swath map as proof of damage.