How to Scan an Area for Old Roofs (Without Wasting a Season Knocking the Wrong Doors)

Most roofing companies work the whole street. A crew drives into a subdivision, picks a corner, and starts knocking every door from one end to the other. Half those houses got a new roof four years ago. A quarter of them are rentals where nobody answers. And buried somewhere in the mix are the eight or ten homes whose roofs are genuinely worn out and due for replacement this year. The crew burns a full day to find those eight doors, and on a bad day they never reach them because they ran out of daylight standing on porches of houses that didn't need a thing.

Scanning an area for old roofs is the discipline of flipping that ratio on its head. Instead of touching every door and hoping, you read the neighborhood from above first, score which roofs are likely old, and send your people to the worn ones. Done right, it turns a guessing game into a routing problem. You stop paying for gas, payroll, and stamps to reach roofs that physically can't buy a re-roof, and you put more at-bats in front of the homes that can.

The catch is that doing it right is harder than it sounds. "That roof looks old" from a truck window is worth almost nothing. Aerial imagery can lie to you in a dozen ways. County records tell you when the house was built, not when the roof was last replaced, and those two numbers diverge hard in any neighborhood older than fifteen years. Below is the full workflow pros actually use, the cheap free tools, the paid ones, the visual cues that separate a tired roof from a shadow, and the mistakes that waste a whole canvassing season.

What "old" actually means before you scan anything

Before you open a single map, get specific about what you're hunting. "Old roof" is not one thing, and the answer changes how you scan.

Most residential roofs in the United States are asphalt shingle, and most of those are three-tab or architectural (laminated) shingles. A three-tab roof installed in the 1990s or early 2000s was typically sold as a 20- or 25-year product but realistically gives 15 to 20 years of service depending on slope, ventilation, and climate. Architectural shingles, which took over the market in the 2000s, are sold as 30-year or "lifetime" products but in practice most installers will tell you 22 to 28 years of real service is the honest range, less in hail country or on a hot south-facing slope with poor attic ventilation.

So when you scan, the population you care about is roughly:

• Roofs in the 15-to-25-year-old band that are aging out on their own (the steady, all-weather opportunity).
• Roofs of any age that took a recent hail or wind event and are worn out early (the storm opportunity).
• Roofs already showing failure regardless of stated age: curling, granule loss, missing tabs, patched sections, daylight problems.

The number you can never read off a map is the exact install date. The honest output of any area scan is a range and a probability, not a birthday. A roof you score as "likely 18 to 22 years old" might be 16 or might be 24. Treat the scan as a way to rank doors by odds, not as a certificate. Reps who walk up saying "records show your roof is exactly 19 years old" get the door closed in their face, because the homeowner knows that's not knowable and now thinks you're a liar. Reps who say "a lot of the roofs on this street went up around the same time and they're getting to the age where we start seeing problems, mind if I take a quick look?" get to the ladder.

Know the material before you guess the age

You cannot age a roof you've misidentified, so the first read on any roof is always "what is it made of," then "how old is it." The common residential materials and their service lives:

• Three-tab asphalt shingle: flat, uniform, thin shadow lines, often a single color. Service life roughly 15 to 20 years. The bread-and-butter old-roof target.
• Architectural / laminated asphalt shingle: thicker, dimensional, a layered look with deeper shadow lines and color variation. Service life roughly 22 to 28 years. The dominant new-construction shingle since the 2000s.
• Wood shake / shingle: irregular, brown-to-gray, splits and curls with age. Service life varies widely; often replaced for fire code. Reads textured even when new, so don't mistake new shake for worn asphalt.
• Concrete or clay tile: rounded barrel or flat profile, strong repetitive pattern, reds/browns/grays. The tile itself lasts 50-plus years even though the underlayment fails sooner — usually not your re-roof sale, so it's a disqualifier on an old-asphalt scan.
• Slate: dark, flat, rectangular, premium homes. 75-plus years. Specialty work, not a volume target.
• Metal (standing seam or metal shingle): long straight seam lines or a stamped pattern, often reflective. 40-plus years. Watch for rust as the age tell.

Getting the material right up front kills a whole class of false positives. A textured wood-shake roof and a streaky old asphalt roof can look similar to an untrained eye on a straight-down image; the oblique view and the curb pass settle it. Train your scanners to call the material out loud before they score the age.

The two-altitude method: read it from above, confirm it from the curb

The single biggest mistake in roof scanning is trusting one source. The reliable workflow always uses two altitudes:

1. The desk pass (from above): aerial and satellite imagery plus property data, done at a computer before anyone leaves the office. This narrows a neighborhood of 400 homes down to a target list of maybe 60 to 90 that are worth a human's time.
2. The curb pass (from the ground): a rep or a windshield survey that confirms the worn roofs and discards the false positives, then knocks.

The desk pass is cheap and fast but lies a fraction of the time. The curb pass is accurate but expensive in payroll and hours. You use the cheap one to protect the expensive one. Never skip the desk pass and just knock; never trust the desk pass so completely that you knock a list without a human ever looking. The rest of this is how to run both passes well.

Step 1: Pick the area and pull the imagery

Start by drawing a boundary you can actually finish. A good first scan area is a single subdivision or a cluster of 2 to 6 connected streets, roughly 150 to 600 rooftops. Big enough to be worth the setup, small enough that you can route a crew through the confirmed list in a day or two.

Good places to focus a first scan:

• Subdivisions built 15 to 28 years ago. Pull this from the county assessor's "year built" field (more on that below). A neighborhood platted around 1999 to 2008 is statistically full of original roofs hitting the end of their service life right now, all at once, because the builder roofed them all in the same 18-month window.
• Neighborhoods inside a recent storm swath. If a hail or straight-line wind event came through in the last 6 to 24 months, the roofs there may be worn out regardless of age.
• Streets where you already did a job. You have a truck wrap, a yard sign, and a real reference. Scan the surrounding blocks.

For imagery, you have a tiered set of options from free to paid:

For a contractor just starting, Google Earth Pro plus your county's parcel viewer covers most of what you need and costs nothing. The desktop version of Google Earth Pro is the workhorse here, mainly because of its historical imagery slider (the clock icon in the toolbar). That feature alone is worth the download, and we'll come back to it.

Step 2: Get year-built data, and understand why it lies

Every county in the U.S. keeps property records, and almost all of them now publish a parcel viewer or GIS map online. Search "[your county] parcel viewer" or "[your county] property appraiser" or "[your county] GIS." These give you, per address, the year the house was built, the parcel boundary, the owner of record, and often whether the owner's mailing address matches the property address (a clue about owner-occupied vs. rental).

Year built is your starting filter, but you have to understand exactly how it deceives you, because most reps either trust it too much or dismiss it entirely. Both are wrong.

Why year built is a useful starting point: in a tract subdivision, the builder roofs every house within a year or two of construction. So a neighborhood with "year built = 2003" across the board started life with a synchronized fleet of roofs that are now around 20 years old. Even with re-roofs mixed in, the base rate of old roofs on that street is high. Year built sets your odds.

Why year built lies house-by-house: a re-roof almost never updates the county "year built" field. Reroofing usually pulls a permit (or sometimes doesn't), but that permit updates a separate record, not year-built. So:

• A 2003 house that was re-roofed in 2018 still shows "year built 2003" but has a 6-year-old roof. False positive if you target on year built alone.
• A 2003 house with its original roof shows the same "2003" and is genuinely 20+ years. True positive.
• A 1975 house re-roofed last year looks ancient on paper and has a brand-new roof. False positive.

This is the core reason you cannot scan an area for old roofs using public records alone. Year built tells you the neighborhood-level probability. The aerial imagery is what separates the original roofs from the re-roofed ones house by house. Zillow's "year built," Google's data, and the county all share this same blind spot: they know when the structure went up, not when the roof was last replaced, because nobody publishes "roof last replaced" as a clean dataset.

Some jurisdictions let you search building permits online (search "[city] permit search" or "accela [city]"). If your area has this, a re-roof permit is the closest thing to a real roof date you can get for free. It's incomplete (lots of re-roofs go unpermitted, especially older insurance jobs), but a permit reading "REROOF 2019" is a hard signal to drop that address from your old-roof list. Pulling permits is tedious one address at a time, but for a high-value street it's worth an hour.

Step 3: Read the roof from above — the visual cues that actually mean "old"

This is the skill that separates a real scan from "that one looks dark, knock it." Open the area in Google Earth Pro (or your aerial tool of choice) and zoom to where a single roof fills a good part of the screen. You're looking for specific, learnable signals. Here is what each one means.

Signs a roof is likely old or worn

• Color difference between slopes of the same house. A south- and west-facing slope ages faster from sun than a north slope. When one plane looks noticeably lighter, grayer, or more washed-out than the others on the same roof, that's UV wear and granule loss. A uniform fresh roof is one even color across all planes.
• Streaking and dark staining. Long dark streaks running down the slope are usually algae (Gloeocapsa magma), which loves older, moisture-holding shingles. Algae alone doesn't mean failure, but heavy streaking correlates with age and a tired surface.
• A "flat" or matte texture. New architectural shingles cast small, crisp shadow lines from their thickness and the laminated tabs. As shingles wear and lose granules and the mat flattens, the roof reads smoother and flatter from above, with less shadow definition.
• Visible patches or mismatched sections. A rectangle of slightly different color is a repair. Repairs mean the roof has already started failing somewhere; the rest is usually the same age.
• Sagging or wavy ridgelines and decking. A roofline that dips, waves, or shows a rippled "washboard" surface suggests old decking, moisture, or multiple layers. In oblique (angled) imagery this is easier to see than in straight-down (orthographic) views.
• Rust on metal accessories. Rusty vents, flashing, or a rusty metal roof are obvious from above and date the install.
• Moss buildup, especially in the Pacific Northwest and shaded northern slopes, reads as green-brown fuzzy texture and indicates a roof that's been damp and undisturbed for years.

Storm-specific signs (different from age wear)

• Bright, even "speckling" or pockmarking across a slope after a known hail event can be impact bruising or granule displacement, though hail damage is notoriously hard to confirm from imagery and usually needs a ground inspection.
• Missing shingles or tabs, blue tarps, or sections of exposed underlayment/decking after a wind event. Tarps are the loudest signal in any neighborhood; circle every one.
• Debris fields and downed limbs on multiple roofs in a tight cluster tell you a storm cell hit that pocket, even if the damage to a given roof isn't obvious yet.

What fakes you out (false positives to learn)

This is where most self-taught scanners go wrong. The following look "old" or "damaged" from above but usually aren't:

• Shadows. A roof half in tree shadow looks dark and worn on one side. Move the imagery date or check a different time of day if your tool allows. Don't score a shadow as wear.
• Stale imagery. The picture you're looking at might be from three years ago. A roof that looks old in 2021 imagery could have been replaced in 2022. Always check the imagery date (Google Earth Pro shows it at the bottom of the screen) and use the historical slider to find the most recent capture.
• Wet roofs. A roof photographed right after rain reads darker and "older."
• Material confusion. Some materials just look dark or textured when new: certain architectural shingles, wood shake, slate, and concrete tile. A dark slate roof is not an old asphalt roof. Learn the material first.
• Dirt and debris vs. wear. Pollen, leaves, and dirt wash off. Granule loss doesn't. From straight above these can look similar; obliques and a curb check sort them out.

Orthographic vs. oblique: use both angles

Most free aerial imagery is orthographic — a straight-down, top-of-the-roof view. That's great for reading slope color, streaking, and patches, but it flattens everything, so sagging decking and curling edges hide. Oblique imagery is shot at an angle (think 45 degrees from a plane), and it shows you the roof the way you'd see it standing across the street and looking up. Bing Maps offers oblique "Bird's Eye" views in many areas, some county GIS sites carry obliques, and paid providers sell them at high resolution from four cardinal directions.

Why you want both: a roof can read fine straight down and reveal a wavy, multi-layer edge or a sagging ridge in oblique. Curling shingles that vanish in a top-down view pop in an angled one. When a roof is borderline on your top-down read, pulling the oblique is the cheapest tiebreaker before you spend a windshield-survey slot on it. If you have access to obliques from multiple directions, check the south and west faces first — those weather fastest and tell the age soonest.

Measuring and counting from above

While you're in the imagery, two quick measurements pay off. First, roof size: Google Earth Pro's ruler and polygon tools let you rough out a roof's footprint, which helps you sort a street by job size so your biggest-ticket roofs float to the top of the knock route. Second, roof complexity: count the planes, hips, valleys, and penetrations. A cut-up roof with lots of valleys and dormers is both a bigger sale and a roof that tends to leak and wear at the transitions first, so complexity plus age is a strong combined signal. None of this replaces a real measurement for a bid, but it's enough to rank doors by potential value as well as by age.

Use the historical imagery slider as a roof-age detector

Here's a trick that pros use and beginners miss. In Google Earth Pro, click the clock icon to open the historical imagery timeline. Now scrub backward through the years for a single house. If the roof's color and texture suddenly change between two captures (say it's dark and streaked in the 2017 image and clean and uniform in the 2019 image), you've just found the approximate re-roof date: it happened between those two captures. Conversely, if the roof looks the same all the way back to the oldest available image, it's likely still the original and is at least as old as your earliest capture.

This is the single most powerful free technique for estimating roof age from imagery, because it sidesteps the year-built lie entirely. You're watching the actual roof change, not reading a record. The limits: the historical archive only goes back so far (often to the mid-2000s in suburban areas, less in rural ones), captures are irregular, and a re-roof that matched the old color closely can be hard to spot. But for any house with a clear before/after in the archive, you can date the roof to a two-or-three-year window for free.

Step 4: Build the worked scoring system

Looking at roofs one at a time is fine for a single street. To scan an area, you need a repeatable scoring rubric so two different people produce the same target list and so you can rank doors. Here's a simple, field-proven 100-point system you can run in a spreadsheet.

Score every parcel in your area. Then rank:

• 70+ : knock first. High odds the roof is genuinely due.
• 45–69 : second wave / mail. Worth a touch but lower odds; good mail targets.
• Under 45 : skip or drop. New roof, wrong material, or disqualified by permit.

The exact point values matter less than the discipline of writing them down and applying them the same way every time. The disqualifiers (negative scores) are where you save the most money, because dropping a known re-roof from the list is one fewer wasted door, stamp, and conversation.

A worked example. Say you scan a 400-home subdivision built in 2004. Year built puts the whole neighborhood at +20 base. You scrub historical imagery and find 120 homes show a clear re-roof in the last several years (drop them, −40). Another 60 are tile or have permits on file (drop). That leaves roughly 220 likely-original roofs. Of those, 75 show two or more strong wear cues and score 70+. Your crew now has a 75-door first-wave list out of 400 homes, ranked worst-roof-first. That's the entire point: you converted a vague "go work that neighborhood" into a 75-stop route where most doors have a real reason to be knocked.

Step 5: The curb pass — confirm before you commit payroll

Your desk list is a hypothesis. Before you route a full crew, do a windshield survey: one person drives the ranked list and eyeballs each roof from the street, marking confirm / drop / maybe on a phone. This costs a couple hours and routinely kills 15 to 30 percent of a desk list (re-roofs the imagery was too stale to catch, materials misread from above, shadows that fooled the scan). Every house the windshield survey drops is a door your closers never have to waste.

From the curb, the confirming cues are:

• Curling, cupping, or clawing shingles at the edges — clear end-of-life on asphalt.
• Granule loss visible as bald, shiny, or thin spots, and granules collecting in gutters and at downspout splash blocks.
• Missing tabs, exposed nail heads, or shiny exposed mat.
• Sagging, waviness, or a soft-looking deck.
• Rusty or cracked flashing, deteriorated pipe boots, daylight-aged vents.
• Multiple layers visible at the rake/edge (a thick, double-stacked edge means a roof-over, which ages worse and is a strong sale).

Google Street View can do part of this from your desk, but remember Street View imagery is frequently years old, so treat it as a preview, not a confirmation. A live drive-by beats it.

Step 6: Route, knock, and mail the confirmed list

Now you have a confirmed, ranked list. Two motions use it:

Door-knocking. Route the crew worst-roof-first so the highest-odds doors get knocked while everyone's fresh and there's daylight. Hand each rep a per-house note ("original 2004 roof, heavy south-slope wear, owner-occupied") so a green canvasser walks up sounding like they did their homework. The script stays honest: you're noticing the roofs in here are reaching the age where problems show up, and you'd like to take a quick look. You are not quoting a manufactured exact age.

Direct mail. Your 45–69 "second wave" tier is often a better mail list than a knock list, because mail reaches the doors that didn't answer and the renters' owners who live elsewhere. A postcard to 200 scored-old roofs beats a blanket mailer to 1,000 unscored homes on cost per response, because you cut the spend going to new roofs that can't buy from you.

What scanning saves you: a worked cost example

The reason to scan instead of blanket-knock isn't theory; it's a spreadsheet. Walk through the same 400-home neighborhood two ways.

Blanket approach. A two-rep crew knocks all 400 homes. At a realistic pace with no-answers, callbacks, and walking time, that's three to four full field days. Say roughly 14 percent of those roofs are genuinely due (the rest are too new, wrong material, or recently re-roofed). Your reps spend the overwhelming majority of their day on porches of homes that can't buy a roof. You also mailed the whole neighborhood: 400 postcards at, say, a dollar-plus all-in per piece, most of it landing in mailboxes of new roofs.

Scanned approach. You spend half a desk day and a couple windshield hours building a ranked list. The crew now knocks a confirmed 75-door first-wave list in about a day, where most doors have a real reason. You mail the 45-to-69 second tier — maybe 150 homes — not all 400. Same crew, same week, but the at-bats are concentrated on roofs that can convert, and your mail spend drops by more than half while reaching better targets.

The math that matters: one re-roof is worth thousands of dollars in revenue. If concentrating your reps on scored doors produces even one or two extra signed jobs per neighborhood that the blanket approach would have missed because the crew ran out of daylight before reaching them, the scan has paid for itself many times over. The cost of scanning is a few hours of desk and drive time; the cost of not scanning is a closer's entire week spent mostly on roofs that were never going to buy. That asymmetry is the whole argument.

A free-tools deep dive

Since most contractors start on the free stack, here's how to get the most out of each one.

• Google Earth Pro (desktop, not the browser version): download the desktop app for the historical imagery slider, the ruler/polygon measuring tools, and the ability to save placemarks. Workflow: drop a placemark on every scored roof, color-code by tier, and you've built a visual route map for free. The capture date shows at the bottom edge — always read it.
• County parcel viewer / property appraiser: your source for year built, parcel lines, owner of record, and the owner's mailing address (the owner-occupied vs. rental tell). Bookmark your local one; the URL and field names differ by county.
• City/county permit search: where it exists, this is your only free source close to a real roof date. Search the address and look for any "reroof," "re-roof," or "roof" permit. A recent one is a hard drop.
• Google Street View: an eye-level preview from your desk. Useful for confirming material and spotting obvious curling, but the imagery is often years old, so never let it be your final confirmation.
• Bing Maps Bird's Eye: your free oblique source. When a top-down read is ambiguous, this angle frequently settles it.
• NOAA Storm Prediction Center storm reports and the National Weather Service: map storm swaths so a storm scan targets the real footprint. After major declared events, check NOAA's emergency-response imagery site for fresh, high-resolution captures.
• USGS EarthExplorer: a deeper archive of aerial and satellite imagery, useful when you want historical coverage Google's slider doesn't have.

The free stack genuinely works for a contractor scanning a neighborhood or two at a time. Its only real ceiling is labor: every step is manual and per-address, so it caps out at the number of roofs one person can read in a day.

Edge cases that trip people up

A few situations don't fit the clean tract-subdivision model, and they're worth knowing.

• Custom-built and infill neighborhoods. When homes were built one at a time over many years, year built is scattered and the "synchronized fleet" base rate disappears. Here you lean harder on the imagery read and the historical slider, address by address, because the neighborhood-level odds don't help you.
• HOA neighborhoods with forced re-roofs. Some HOAs mandated a community-wide re-roof after a storm or for appearance, so a whole street of "old" houses may have matching newer roofs. The historical slider catches this — you'll see them all change in the same capture.
• Heavy tree cover. In mature neighborhoods, canopy hides roofs from above. Use leaf-off winter imagery if your archive has it, pull obliques from multiple directions, and accept that some of these only get resolved at the curb.
• Flat and low-slope roofs. Modified bitumen, TPO, and EPDM age differently and don't show shingle wear cues; look for ponding stains, patches, and seam failures instead. These are usually a different crew and a different sale.
• Recently painted or coated roofs. A roof coating can make an old roof read newer from above. The curb pass and the homeowner conversation sort these out.
• Solar panels. Panels both hide roof area and signal a homeowner who has already invested in the house; the roof underneath was often replaced or inspected at install. Note them, don't assume.

A note on both: the goal of the scan is to stop spreading your effort evenly across roofs that aren't all equal. One re-roof is worth thousands of dollars. Pointing your most expensive resource — a skilled closer's time — at the roofs least likely to need work is the quiet way roofing companies bleed margin.

Where a roof-scoring platform fits (and where it doesn't)

Everything above can be done by hand with free tools, and for a single street that's the sensible way to do it. The problem is that it doesn't scale. Scrubbing historical imagery house-by-house, pulling permits one address at a time, and scoring 400 parcels in a spreadsheet is a full day or more of skilled desk work per neighborhood. If you want to scan ten neighborhoods, the manual method falls apart, and the temptation is to skip the careful parts and just knock — which puts you right back to working the whole street.

This is the gap a platform like RoofPredict is built to fill. It reads aerial imagery across an area and returns a per-address roof age range plus a per-roof storm model — going past whether a storm passed through the ZIP to model how hail trajectory and wind likely loaded each individual roof — and ranks the homes by which ones are most likely due. In other words, it automates Steps 2 through 4: the year-built base rate, the imagery read, the storm overlay, and the scoring, across hundreds or thousands of roofs at once, so your team starts the day with a ranked list instead of building one.

Be clear-eyed about what that does and doesn't give you, because honest limits are the whole point of doing this well:

• It outputs a range and odds, not a roof's birthday. A score of "likely 18 to 22 years" means the model's confident the roof is in that band, not that it was installed on a date. You still confirm worn roofs at the curb. No imagery model can certify an exact install date, and anyone claiming otherwise is selling you something.
• Storm modeling is probability, not proof of damage. It tells you which roofs a storm most likely wore out so you knock the right doors. It does not replace a physical inspection, and it is never evidence that a specific roof is damaged. The roof either has damage on the ladder or it doesn't.
• It's not a lead service. You're not buying a homeowner's name that's been resold to five competitors. You're getting a scored map of your own chosen streets that you go work. The doors are yours to knock.
• You still own the relationship and the inspection. The platform points you at the right roofs faster and at scale; the sale, the ladder, and the documentation are still your crew's job.

The honest pitch is narrow: it removes the day of manual desk work and lets you scan more area than a person can by hand, while keeping the same two-altitude discipline. If you only ever work one subdivision a year, free tools are plenty. If your growth depends on scanning area after area and you can't afford to skip the careful steps at scale, that's where the automation earns its keep.

Storm scanning: a different clock

Everything so far assumes you're hunting roofs aging out on their own. Storm scanning runs on a different clock and deserves its own checklist, because a storm can wear out a 9-year-old roof overnight and none of your year-built logic will catch it.

After a hail or significant wind event:

1. Map the swath. The NOAA Storm Prediction Center and the National Weather Service publish storm reports, and SPC keeps hail and wind reports by date. Pull the footprint of the event so you scan the actual impacted area, not the whole county.
2. Grab fresh imagery if it exists. After major declared events, NOAA sometimes flies and publishes high-resolution aerial imagery within days. That's the freshest read you'll ever get, and it shows tarps, missing sections, and debris clearly.
3. Scan for acute failure first. Tarps, missing shingles, exposed deck, debris clusters — circle them, they're urgent and obvious.
4. Then scan for subtle hail wear, knowing it's hard to confirm from imagery and almost always needs a ladder. Don't promise damage you haven't seen.
5. Stay legally clean. You document conditions and provide estimates. The insurer decides coverage and the homeowner owns the claim. You do not handle, file, or negotiate the claim, you do not promise approval, and you say nothing about a homeowner's deductible. A storm model tells you where to look; it is never proof of damage and never a guarantee a claim will be paid. Keep that line bright and you keep your license. (Several states treat a roofer acting as an unlicensed public adjuster as a serious violation, so the safe role is documentation and honest estimates only.)

The storm opportunity and the aging-out opportunity stack. The aging-out scan gives you steady work in any weather; the storm scan gives you a surge when a cell comes through. The same two-altitude discipline runs both.

The mistakes that waste a season

A quick field list of the errors that quietly kill canvassing programs:

• Trusting year built as roof age. The deepest trap. It's a neighborhood base rate, not a house's roof date. Re-roofs are invisible to it.
• Reading stale imagery as current. Always check the capture date. A "worn" roof in three-year-old imagery may be a year-old roof now.
• Scoring shadows and wet roofs as wear. Both fake age. Use obliques, multiple captures, and the curb pass to sort them out.
• Skipping the curb pass. Knocking a raw desk list burns closers on false positives. The windshield survey is cheap insurance.
• Confusing material. Slate, tile, and metal have their own lifecycles. A dark tile roof isn't an old asphalt roof.
• Overclaiming age to the homeowner. Saying "records show your roof is exactly X years old" reads as a lie because it isn't knowable, and it torches trust at the door. Use ranges and "a lot of roofs around here are reaching that age."
• Treating a storm model as proof of damage. It points you to likely-worn roofs. The damage is real only if it's on the ladder.
• Knocking everything because scanning felt like work. The entire payoff is not knocking the new roofs. If you scan and then knock the whole street anyway, you wasted the scan.
• No disqualifier discipline. The money is as much in the roofs you drop (permits, recent re-roofs, wrong material) as in the ones you knock.

A repeatable weekly rhythm

To make scanning a habit instead of a one-off, put it on a cadence:

• Monday (desk, 2–4 hrs): pick the week's area, pull year-built and parcel data, run the imagery read and historical-slider check, score and rank the list.
• Tuesday (windshield, 2–3 hrs): drive the ranked list, confirm/drop, finalize the knock route and the mail list.
• Wednesday–Friday (field): crew knocks the confirmed 70+ list worst-roof-first; mail drops to the second-wave tier; reps carry per-house notes.
• Friday (review, 30 min): mark which scored roofs converted to inspections and jobs, and feed that back. Over a few weeks you learn which cues in your market predict real sales, and your scoring gets sharper.

That feedback loop is what separates a contractor who scans once and quits from one who compounds. Every week you find out which roofs your scan called right, and you tune. After a season, your desk pass gets accurate enough that the curb pass mostly just confirms, and your closers spend their days on doors that were almost all worth knocking.

Bottom line

Scanning an area for old roofs isn't a gadget, it's a two-altitude discipline: read the neighborhood from above to rank the doors by odds, then confirm the worn ones from the curb before you spend a closer's day. Free tools — Google Earth Pro's historical slider, your county parcel viewer, permit search, NOAA storm reports — get you surprisingly far on a single street, as long as you respect what each one can and can't tell you. The year built is a base rate, not a roof date. The imagery shows the roof but can be stale. The exact age is never knowable from a screen, so you work in ranges and odds.

When you outgrow doing it by hand and need to scan area after area without cutting the careful steps, that's the moment a scoring platform like RoofPredict pays off — it automates the read and the ranking and adds a per-roof storm model, while you keep the curb pass, the ladder, and the honest conversation that close the job. Either way, the win is the same: stop working the whole street, and put your best people in front of the roofs that are actually due.

FAQ

Can you really tell how old a roof is from satellite or aerial imagery?

You can estimate a roof's age as a range, not an exact date. Wear cues like uneven slope color, streaking, granule loss, and a flat matte texture indicate an older asphalt roof, and Google Earth Pro's historical imagery slider can pin a re-roof to a two-or-three-year window if the roof's color and texture visibly change between two captures. What no imagery can do is certify an exact install date, so treat the output as odds and confirm worn roofs from the curb.

Why isn't the county "year built" the same as the roof's age?

Year built records when the structure went up, not when the roof was last replaced. A re-roof almost never updates that field, so a 2003 house re-roofed in 2018 still reads "2003" while having a six-year-old roof. Year built is a useful neighborhood-level base rate because tract builders roof all the houses in the same window, but house-by-house you need aerial imagery and, where available, permit records to separate original roofs from re-roofed ones.

What free tools can I use to scan a neighborhood for old roofs?

Google Earth Pro (desktop) is the workhorse for its historical imagery slider and roof-texture reading. Add your county's parcel viewer or property appraiser site for year-built and owner data, your city or county permit search for re-roof dates where available, Google Street View for an eye-level preview, and NOAA Storm Prediction Center reports for storm swaths. That free stack covers most of a single-neighborhood scan at zero cost.

How do I tell an old roof from a shadow or a dirty roof in aerial imagery?

Shadows, wet roofs, pollen, and dirt all fake age from above. Check the imagery capture date, use the historical slider to compare multiple captures, and look at oblique (angled) views when available. Real wear shows as granule loss, uneven color between slopes of the same house, and a flat texture that persists across captures and lighting. Anything you can't confirm from above gets settled by a quick drive-by from the curb.

What's the difference between scanning for old roofs and scanning after a storm?

Aging-out scanning hunts roofs in the roughly 15-to-25-year band that are wearing out on their own, using year-built base rates and imagery wear cues. Storm scanning runs on a different clock: a storm can wear out a much newer roof overnight, so you map the event swath from NOAA/SPC reports, grab fresh post-event imagery, and look for tarps, missing shingles, and debris. The aging scan gives steady all-weather work; the storm scan gives a surge. The same read-from-above-then-confirm-from-the-curb discipline runs both.

How many doors should an area scan eliminate before I knock?

On a typical aging subdivision, a disciplined scan often disqualifies half or more of the homes before anyone knocks. Recent re-roofs caught in historical imagery, permits on file, wrong materials like tile or slate, and clearly new roofs all drop off. In a 400-home neighborhood it's common to end up with a 60-to-90-door ranked first-wave list. The savings come as much from the doors you drop as from the ones you knock.

Is a roof-scoring platform like RoofPredict a lead service?

No. A lead service sells you a homeowner's name that's often been resold to several competitors. A scoring platform reads aerial imagery across your own chosen streets and returns a per-address roof-age range and a per-roof storm model, then ranks which homes are most likely due. The doors are yours to knock and the relationship is yours to build. It automates the desk work of reading and ranking roofs at scale; it doesn't sell you customers.

Can a storm model prove a roof is damaged?

No, and treating it as proof is a mistake. A per-roof storm model estimates which roofs a hail or wind event most likely wore out so you know which doors to knock. Whether a specific roof actually has damage is determined on the ladder during a physical inspection. The model is a targeting tool and a probability, never evidence and never a guarantee that an insurance claim will be paid.

What should I say at the door so I don't sound like I'm lying about the roof's age?

Avoid claiming an exact age, because homeowners know an install date isn't knowable from a record and it reads as a lie. Use the truth: a lot of the roofs in the neighborhood went up around the same time and are reaching the age where problems start to show, and you'd like to take a quick look. Ranges and base rates are honest and credible; manufactured precision closes the door on you.

Where's the legal line for roofers scanning storm-damaged areas?

Document conditions and provide honest estimates only. The insurer decides coverage and the homeowner owns the claim. Do not file, handle, or negotiate the claim, do not promise approval, and say nothing about a homeowner's deductible. Several states treat a roofer acting as an unlicensed public adjuster as a serious violation, so keep your role to documentation and estimates and keep that line bright.