How to Estimate Roof Age From the Year a Home Was Built (And Why That Number Lies)

Every roofer has done the math on the back of a truck. You pull up a street, you see a house built in 2004, and your brain does the subtraction: it's 2026, so that roof is about 22 years old, so it's due. You knock. The homeowner opens the door and tells you they re-roofed after the 2019 hail. The roof is seven years old. You just burned a knock, and worse, you sound like every other guy who pulled the same county record off the same website.

That subtraction — current year minus year built — is the most common way roofers estimate roof age, and it is wrong more often than anyone wants to admit. Not a little wrong. Directionally wrong, on a meaningful share of the houses you most want to work, in a way that quietly drains your gas, your mailing budget, and your reps' confidence.

The frustrating part is that year built is not useless. It is a real signal. It is just a weak one, and most people treat it like a strong one. The skill — the thing that separates a crew that knocks smart from a crew that knocks the whole ZIP and prays — is knowing exactly how much to trust that number, which other cheap signals to stack on top of it, and when to throw the build year out entirely.

What follows is the method a working contractor actually uses to go from a year-built number to a defensible roof-age range per address: what build year really tells you, where it breaks, the public records that override it, the curb-and-aerial signals that confirm or kill it, and a repeatable workflow you can hand to a green rep. Roof age is never a date. The best you can ever produce from the curb is a tight range with a confidence level attached. Pros who pretend otherwise lose deals to the homeowner who knows their own re-roof date.

Why "year built" feels like roof age (and why that's a trap)

When a house is brand new, the year built is the roof age. Day one, they match perfectly. That is the whole reason the shortcut feels safe — it was true exactly once, on the day the certificate of occupancy was issued.

From that day forward, the two numbers drift apart, and they only ever drift in one direction: the roof gets younger relative to the house. A 1998 house has had 28 years for something to happen to its roof. Re-roofs, storm replacements, insurance jobs, a flip where the investor slapped on a new layer to sell — every one of those events resets the roof clock while the build year sits frozen in the county database forever.

So the build-year shortcut has a built-in bias. It systematically over-estimates the age of roofs in older neighborhoods, which is exactly where you'd want to work, which is exactly why the error hurts. The houses most likely to have already been redone are the same houses your subtraction flags as most due.

There's a second trap underneath the first. Year built in public records is not always the year the house was actually finished. Depending on the jurisdiction and how the assessor coded it, "year built" can mean the year the permit was pulled, the year construction started, the year it was finished, or the "effective year built" — an adjusted figure assessors use after major renovations to reflect a property's condition. Effective year built is the sneakiest one, because it already bakes in remodels in some counties and not others, and you usually can't tell which from the public-facing record. You are subtracting from a number whose definition changes county to county.

None of this means you ignore build year. It means you treat it as the starting hypothesis — "this roof is probably original and probably old" — and then you spend the next ninety seconds trying to prove yourself wrong before you commit a knock or a stamp.

What the build year actually tells you — three real signals

Strip away the false precision and the year a home was built still hands you three genuinely useful things. Knowing what they are keeps you from over-reading the date.

1. An upper bound on roof age. This is the strongest thing build year gives you, and it's the one people forget because they're chasing the wrong number. The roof cannot be older than the house. A 2009 build has a roof that is at most about 17 years old in 2026 — full stop. That ceiling is rock solid (barring a salvaged-material rebuild, which is vanishingly rare in residential). Upper bounds are quietly valuable: they let you confidently skip new construction. A street of 2021 builds is a street you do not knock for replacement, and the build year tells you that with near-certainty. Use the date most aggressively to rule houses out, not in.

2. A first guess at the roofing system and code era. Build year tells you what was probably code and common practice when the original roof went on. Pre-2000s tract housing in much of the country leaned on 3-tab shingles with a 20-to-25-year nominal rating; a lot of mid-2000s-and-later work shifted to architectural (dimensional/laminate) shingles rated 25 to 30. Build era also hints at deck type, ventilation norms, and whether you might be looking at an older organic-mat shingle versus modern fiberglass. This is era-context, not a guarantee — but it sharpens your guess about how long this roof was designed to last, which matters as much as how old it is.

3. A neighborhood cohort. Tract developments go up in waves. If you know one house in a subdivision was built in 2003 and originally roofed once, there's a decent chance its neighbors share a birth year, an original roof spec, and a storm history. Build year, read across a block instead of one house, tells you where the cohorts are — and cohorts are how you find streets where a bunch of original roofs are aging out together. That's a route, not a single door.

Hold those three together and the honest read of a build year is: "This is the oldest the roof could be, here's roughly what it's made of, and here are its neighbors." That is a long way from "this roof is 22 years old."

The four events that secretly reset the roof clock

Your subtraction is only correct if the roof on the house today is the original roof. Every one of these events breaks that assumption, and none of them shows up in the year-built field.

Storm replacements. The big one. After a significant hail or wind event, whole neighborhoods get re-roofed inside of 18 months on insurance money. If a 2001 subdivision took a baseball-sized hail event in 2017, a large share of those "25-year-old roofs" are actually nine years old — and they all reset at the same time. This is why storm history isn't a side detail; it's often the dominant factor in a neighborhood's true roof-age distribution, and it's completely invisible in the build year.

Real-estate-driven re-roofs. Roofs get replaced to sell a house. Inspection flags a roof at end of life, the deal hinges on it, and either the seller replaces it or the buyer does right after closing. Flips are even more aggressive — investors re-roof to move the property. So a house that sold in the last several years is meaningfully more likely to have a younger roof than its build year suggests, especially if it was an older home.

Age-out replacements. The ordinary case: the original roof simply wore out and a homeowner replaced it on their own dime ten or fifteen years ago. No storm, no sale, no drama. Common on homes from the '80s and '90s, which have now had time for two roofs.

Partial replacements and overlays. The messiest. A previous owner replaced one slope after a tree-limb hit, or laid a second layer over the first (a tear-off-avoiding overlay, legal in some jurisdictions up to two layers). Now different parts of the same roof are different ages, and "roof age" becomes genuinely ambiguous. Overlays in particular fool the build-year math and fool a quick curb glance.

The practical takeaway: before you trust current-year-minus-build-year, you are really asking one question — has anything happened to this house since it was built? The rest of this is how you answer that cheaply, at scale, without climbing a ladder.

How often does the clock actually get reset?

Nobody can hand you a clean national percentage, and you should distrust anyone who pretends to. But you can reason about it well enough to know how hard to lean on build year, neighborhood by neighborhood.

Work it from first principles. A roof's odds of having been replaced go up with (a) how many years it's had to wear out, (b) how many qualifying storms have rolled over it, and (c) how many times the property has changed hands. So a 1990s home in a hail-prone metro that's sold twice is far more likely to be on its second or third roof than a 2015 home in a calm climate that's never sold. The reset rate isn't one number — it's a function of age, weather, and turnover, and those three things vary wildly by market.

That means the build-year shortcut is least reliable exactly where roofers do the most work: older housing stock, active storm corridors, normal real-estate churn. In a brand-new subdivision the shortcut is nearly perfect (the roofs really are all the build age). In a 40-year-old neighborhood that's eaten three hail seasons, the build year is almost decorative. Calibrate your trust to the neighborhood, not to a blanket rule.

There's a useful mental shortcut here: the older and stormier the area, the harder you verify. New tracts you can target off build year alone and be right most of the time. Old, beaten-up neighborhoods demand the full signal stack on every borderline house, because that's exactly where build year quietly lies to you.

A roof-age estimate is a range with a confidence level — not a date

Here is the mental model that makes everything downstream work. You are never producing "this roof is 19 years old." You are producing two things:

1. A range. "This roof is 15 to 22 years old."
2. A confidence level — how much you trust that range, based on how many independent signals agree.

A single signal (build year alone) gives you a wide range and low confidence. Each additional signal that agrees either tightens the range or raises your confidence, or both. Signals that disagree are the most valuable of all, because a disagreement is usually a re-roof you would have missed — the exact door you don't want to waste, or the exact door you do want to knock.

Think of it as stacking. Build year says "old." Imagery shows a roof with no granule loss and crisp ridges: that disagrees — knock it off your list or flag it as recently done. Imagery shows streaking, patched slopes, and a tarp: that agrees and tightens — high-confidence due. A permit record from 2016 overrides everything — you now have a near-exact date, not a range.

This is also how you talk to a homeowner without getting embarrassed. "Records show your home was built in 2003, and from the aerial your roof looks like it's in the back half of its life — does that match what you know?" invites a conversation. "Your roof is 23 years old" invites a correction, because they know their own re-roof date and you just told them you don't.

The signal stack: cheap to expensive, weak to strong

Rank your signals by how much they move your confidence per minute of effort. Start with the cheapest, stop as soon as you have enough confidence to act. You almost never need all of them.

Tier 1 — Year built (free, weak, you already have it)

Your starting hypothesis and your upper bound. Treat it as "oldest the roof could be" and as a way to rule out new construction. Never act on it alone for anything except skipping obviously new homes.

Tier 2 — Sale history (free or cheap, moderate)

When did the property last sell? A sale in the last few years on an older home raises the odds of a recent re-roof (inspection-driven or flip). Sale dates are public and usually right next to the build year in assessor and listing data. They don't confirm a re-roof, but they shift your probability and tell you which doors deserve a closer look. Old listing photos, when you can find them, sometimes show the roof condition at a known date — a free time-stamp.

Tier 3 — Aerial and street-level imagery (cheap, moderate-to-strong)

This is the highest-leverage signal most roofers under-use. You can read a surprising amount of roof age off good aerial imagery without leaving your desk:

• Granule loss / shine: Aging asphalt loses granules and can look smoother, lighter, or shinier in patches as the asphalt substrate shows through.
• Streaking and discoloration: Dark streaks (often algae) and uneven coloring read as an older, weathered roof.
• Patching: A slope that's a different color or texture than the rest screams partial replacement — and breaks your single-age assumption.
• Sagging ridges or deck deflection: Visible waviness suggests deck issues and an older, possibly multi-layer roof.
• Crisp vs. worn ridge lines and edges: New roofs have sharp, uniform ridge caps and clean edges; old roofs look soft and irregular from above.
• Tarps, missing sections, obvious storm damage: Self-explanatory, and gold.

Street-level imagery (the drive-by kind) adds shingle profile — you can often tell 3-tab from architectural, spot curling and cupping at the eaves, and see lifted or missing tabs. The catch with all imagery: it has a capture date, and that date can be one, two, even four years old. A roof that looks fine in two-year-old imagery could have been hammered last spring. Always check how fresh the picture is before you trust it, and never let stale imagery talk you out of a knock in a recent storm zone.

Tier 4 — Permit records (cheap-ish, strong)

This is the one that actually overrides the build year, and it's the step most roofers skip because it's slightly tedious. A re-roof usually requires a permit. If a permit was pulled, there is very often a dated public record of the roof replacement sitting in a county or city database. Find it and your "range" collapses to a near-exact date.

Tier 5 — On-roof / close inspection (expensive, definitive)

The ground truth. Manufacturer codes on the shingle back, the actual layer count, fastener type, flashing condition, attic decking from below. This is where a range becomes a fact — but you only spend this effort on doors that already cleared the cheaper tiers. The whole point of Tiers 1–4 is to make sure you climb the right roofs.

How to actually pull roof-replacement permit records

Permits are the cheat code, so here's the real workflow rather than the hand-wave version.

1. Find the right jurisdiction. Permits are issued by the local building department — sometimes the city, sometimes the county, sometimes a township. Figure out which authority covers the address before you go hunting. Unincorporated areas usually fall to the county; incorporated cities run their own departments.

2. Hit the online portal first. Most building departments now run a public permit-search portal where you query by street address or parcel number (APN). Search the address, then scan the permit list for anything that reads as roofing: "reroof," "roof replacement," "tear-off," "roof covering," or a roofing-trade permit. Note the issue date and the final/closed date — the final inspection date is closest to when the roof was actually finished.

3. Cross-check the assessor. The county assessor's office tracks improvements that change assessed value. A major roof job sometimes shows up as a dated improvement even when the permit is hard to find, and the assessor is also where you confirm the parcel number, sale history, and the exact build-year coding (including whether it's an "effective" year built).

4. Know the failure modes. Permits are strong but not perfect. Plenty of re-roofs happen without a permit — small jurisdictions, rural areas, cash jobs, or homeowners who just didn't pull one. So absence of a roofing permit does not prove the roof is original; it only means you didn't get a confirming signal. A permit's presence, on the other hand, is about as close to ground truth on roof age as you'll get from a desk. Treat found permits as near-fact and missing permits as "no information," never as "original roof confirmed."

5. Mind the coverage gaps. Older permits may predate digitization and live only on paper or microfilm. Some portals only go back a decade or two. If a roof was redone in 1998, the record might simply not be online. Again: missing record ≠ no re-roof.

Done right, a permit pull turns a 15-to-22-year guess into "reroof finaled 2016-08-14." That's the difference between a cold knock and a precise one — and at scale, pulling permits one address at a time by hand is exactly the kind of work that doesn't pencil out for a whole route, which is the gap the data-tooling section below speaks to.

A few words you'll see on a roofing permit, and what they tell you

Permit listings are written in clerk shorthand, and reps waste time because they don't know what they're scanning for. A quick translation so a green hire reads them fast:

• "Reroof" / "re-roof" / "roof recover": a roof replacement or recover. Your target. Note whether it says tear-off or overlay.
• "Tear-off": the old roof came off to the deck — a full replacement. The cleanest reset; the final date is your roof age.
• "Overlay" / "recover": new layer over the old one. The roof age resets for the top layer only, and you're now looking at a multi-layer roof that may have a shorter remaining life and code limits on further layers.
• "Repair" / "partial": only part of the roof was touched. Does not reset the whole roof — flag it as a partial and keep the original-age hypothesis for the rest.
• "Decking" / "sheathing" / "structural": the deck was worked, which usually rides along with a full tear-off. Strong signal of a real replacement.
• Status fields — "finaled," "closed," "CO," "passed": the work was inspected and signed off. "Issued" or "applied" without a final means a permit was pulled but you can't confirm the job finished or when — treat the date as soft.

Write down the type and the final date, not merely that a permit exists. "Overlay 2014, partial repair 2019" is a very different roof than "tear-off finaled 2014," and the homeowner will know the difference even if your notes don't.

When permits and assessor records disagree

Sometimes the assessor shows a roof-driven assessment bump in one year and the permit final lands in a different one. Trust the permit's final-inspection date over the assessment date — assessments often post a year or more after the work, on their own cycle. And if a sale closed between the two, the re-roof was very likely tied to that sale. Reading the dates against each other usually tells the whole story of why the roof got replaced, which is ammunition when you knock.

Reading roof age off the shingle itself (when you get close)

When a rep is actually at the house, the build-year math gets a reality check. Here's how to age asphalt from arm's length, roughly oldest-to-newest tells:

• Granule loss in the gutters and at downspout splash zones. Piles of granules mean the surface is shedding — a mid-to-late-life sign. Bare asphalt showing through on the field is later still.
• Curling and cupping at the edges and corners of tabs. Shingles curl up (cupping) or the corners lift (clawing) as the mat dries out and shrinks with age and heat cycling. Pronounced curling generally means the back half of the roof's life.
• Cracking and brittleness. Older shingles get brittle; you'll see surface cracking and they snap rather than flex. A shingle that crumbles is near or past end of life.
• Blistering and bald spots. Small pocks and bald patches accumulate with UV and age.
• Shingle profile and type. 3-tab (flat, uniform cutouts) skews older/cheaper-era; architectural/laminate (varied, layered, dimensional) skews newer or a higher-end original. Knowing the type tells you which nominal lifespan to apply.
• Ventilation and flashing wear. Rusted flashing, deteriorated pipe boots (cracked rubber collars), and chalky, faded color all stack onto the age picture.
• Layer count at the edge. Look at the rake or eave edge: two distinct layers means an overlay, which both tells you a re-roof happened and warns you the math is now muddy.

The most reliable on-site date, when you can get it without tearing into a sound roof, is the manufacturer's stamp on the back of a shingle — many carry a production code. You'll only see it on a loose tab, a leftover bundle in the garage, or where you're already lifting shingles for a legitimate reason, but when it's there it's the closest thing to a birth certificate.

Two worked examples

Abstractions are easy to nod along to, so here's the method on two real-shaped houses.

Example A — the trap house

The record: Built 1999. Current year 2026. Naive math: ~27-year-old roof. Looks like a slam-dunk knock.

Stack the signals:

• Sale history: Sold in 2021. Older home, recent sale — flag for possible inspection-driven re-roof. Confidence in "old roof" drops.
• Aerial imagery (captured 2024): Ridge caps look crisp and uniform; color is even; no streaking; no patching. Strongly disagrees with "27-year-old roof."
• Permit pull: "Reroof, architectural, tear-off — finaled 2021-05-02."

Estimate: Roof is ~5 years old, replaced at sale. Do not knock for replacement. The naive math was off by 22 years. Two cheap signals (sale + imagery) raised the flag; one strong signal (permit) closed it. Total desk time: a couple of minutes, and you saved a wasted knock plus the credibility hit of telling a 2021-roof homeowner their roof is 27.

Example B — the quiet winner

The record: Built 2006. Current year 2026. Naive math: ~20-year-old roof. Borderline by the numbers — a lot of crews would pass or deprioritize.

Stack the signals:

• Sale history: Original owner, never sold. Lowers the odds of a sale-driven re-roof.
• Storm history: Neighborhood took a verified significant hail event in 2011 and again in 2019. Now it matters which storms actually hit this roof hard, rather than only knowing the ZIP got weather.
• Aerial imagery (captured 2025): Visible streaking, granule-loss sheen on the south slope, one slope a slightly different shade (possible old partial repair). Agrees with "worn," and the mismatched slope hints the roof has already taken damage.
• Permit pull: Nothing roofing-related found. Treat as no information, not as "original confirmed."

Estimate: Range 15–20 years on the main roof, probably original and weathered, possibly with an old partial repair, sitting in a neighborhood that's eaten two hail events — one of them recent. High-priority knock, and you walk in able to talk specifically about the slope that looks worn and the storms the street has taken. The naive math said "meh, 20." The stack said "worn original roof in a beaten-up cohort" — a much better door than the trap house that looked better on paper.

Notice the pattern: build year alone would have mis-ranked these two houses against each other. The trap house looked older and was actually newest; the quiet winner looked borderline and was the real opportunity. Targeting off raw build year doesn't only add error — it can invert your priorities. That inversion is the expensive part.

Modeling the storm, not only the calendar

Everything above ages the roof by time and wear. But two roofs the same age, on the same street, are not equally due — because the storm that rolled through didn't hit them equally.

Hail doesn't fall uniformly. Stone size, fall angle, wind-driving, and roof orientation mean one slope gets pulverized while the slope twenty feet away barely registers. Wind loads concentrate on edges, corners, and the windward face. A roof's exposure — pitch, direction it faces, what shaded or sheltered it — changes how much a given storm actually aged it. So "the ZIP took a 2019 hail event" is a blunt instrument. The roofs that storm actually wore out are a specific subset, and they're not the same as the roofs that merely sat under the storm's footprint.

This is the ceiling on what year built can ever tell you, and it's also where the calendar method and the weather method have to fuse. The best targeting answers two questions at once: how old is this roof (age range, from the signal stack above) and what has the weather actually done to this specific roof (storm exposure, modeled per house, not per ZIP). An old roof in a calm pocket and a newer roof on the windward edge of a severe-hail core can land at the same true "due-ness" — and you'd never see that from build year, or from a hail map alone.

Where data tooling does the boring 80% for you

Everything in the signal stack works by hand. It also doesn't scale. Pulling permits, checking imagery freshness, and cross-referencing sale dates one address at a time is fine for a handful of houses and impossible for a route of four hundred. This is the honest case for tooling, and the honest limits of it.

What platforms like RoofPredict do is run the cheap tiers of that stack across a whole area at once. Instead of you eyeballing one aerial and pulling one permit, it reads aerial imagery to estimate a roof-age range per address, then layers storm exposure modeled per individual roof — more than "this ZIP got hail," a per-house read of hail and wind impact based on each roof's exposure — so you get back a ranked list of which houses are actually due versus which are new and not worth a stamp or a knock. The build year becomes one input among several instead of the whole decision.

The value isn't magic; it's the same logic this whole method describes, run at scale so a route gets prioritized in minutes instead of an afternoon of manual lookups. For a mailer, that means stamps land on worn roofs instead of the 2021 builds on the same street. For a door crew, it means a green rep knocks the houses most likely to convert and can speak to the specific roof and the storms that street has taken — so they close, make money, and stick around instead of churning out after a month of dead doors. For a shop sitting on a fat CRM, it means re-scoring your old estimate list to see whose roof finally aged into the buy zone.

Now the honest limits, because a trade compares notes straight:

• Roof age is a range, never an exact date. Imagery-based estimation narrows the window; it does not read the install stamp off the shingle. A permit you pull by hand can still beat it on a specific house, and the on-roof inspection is still ground truth.
• A storm model is odds, not proof of damage. Modeling that a roof was likely hit hard is a targeting signal — a reason to go look. It is not a damage report, not an adjuster's finding, and not something to present as established fact to a homeowner. You still inspect; the homeowner still owns whatever happens with their insurer.
• It sharpens outbound; it isn't a lead service. It tells you which of your own streets and your own past customers to work. Nobody is handing you a sold customer five other crews also bought.
• Imagery has a capture date too. Tooling inherits the same freshness problem you have — a very recent storm may not yet be reflected in the picture, which is why it pairs imagery with modeled storm data rather than relying on a single old photo.

Use it for what it's good at — collapsing a giant manual stack into a ranked route — and keep doing the close inspection on the doors it surfaces. That's the whole relationship: it makes sure you spend your expensive minutes (knocking, climbing, estimating) on the right roofs.

A repeatable workflow you can hand to a green rep

The point of all of this is a process a new hire can run without ten years of instinct. Here's the desk workflow, in order, with stop conditions so nobody over-works a house.

1. Pull the year built and set the upper bound. "Roof is at most [current year − build year] old." If the home is clearly new construction (built in the last ~8 years and nothing in the area to reset it), stop — skip it for replacement. Don't waste a knock on a young roof.
2. Check sale history. Recent sale on an older home? Raise the "possible re-roof" flag and lean harder on imagery and permits before committing.
3. Look at the aerial. Note the capture date first. Read for streaking, granule sheen, patching, ridge crispness, tarps. Decide: does the picture agree with "old" (tighten toward due) or disagree (lean toward recently done)?
4. Pull the roofing permit if the house is borderline. A found permit overrides everything — write down the final date, you now have a near-exact age. A missing permit means no information; do not upgrade it to "original roof."
5. Factor storm exposure, per house not per ZIP. Did real storms hit this specific roof's exposure, and how recently? A recent severe event can override stale imagery entirely.
6. Write the estimate as a range + confidence. "15–20 yrs, high confidence, weathered, knock" or "~5 yrs, high confidence, permit 2021, skip." Never write a single date.
7. Only then spend the expensive effort. Knock, climb, inspect — on the doors that cleared the cheap tiers. Confirm age on-site with the shingle-level tells and, when available, the manufacturer stamp.

Stop conditions matter as much as the steps. The moment you have enough confidence to act — skip the new build at step 1, knock the obviously-worn original at step 3 — you stop and move to the next house. The stack is there to get you to a decision fast, not to make you research every door to death.

What pros get wrong (the checklist)

The expensive mistakes are predictable. Run this list against your own targeting:

• Treating build year as roof age instead of an upper bound. The single most common error. Build year tells you the roof's ceiling, not its age. Use it to rule houses out, not to rule them in.
• Ignoring sale history. A recent sale on an old home is a loud "check for a re-roof" signal, and it's sitting right next to the build year for free.
• Trusting stale imagery in a storm zone. Two-year-old aerial that looks fine means nothing if the street took hail last spring. Always check the capture date, and never let an old picture talk you out of a recent-storm knock.
• Reading "no permit found" as "original roof." Absence of a permit is absence of information, not proof. Plenty of re-roofs never got permitted.
• Targeting on ZIP-level weather. "The area got hail" is not "this roof got hit." Storms land per-roof, by exposure; treating a whole ZIP as uniformly damaged wastes effort on roofs the storm barely touched and misses the ones it wrecked.
• Forgetting the effective-year-built quirk. In some counties the "year built" you're subtracting from has already been adjusted for renovations. Confirm what the number actually means before you trust the math.
• Quoting a single roof age to a homeowner. "Your roof is 23 years old" gets corrected the second they remember their 2018 re-roof. "Records show a 2002 build and your roof looks like it's in the back half of its life — does that track?" opens a conversation and keeps your credibility.
• Skipping the cheap signals and over-relying on the expensive one. Climbing every roof to age it doesn't scale; neither does eyeballing build year alone. The win is the stack, cheapest first, stopping when confident.

Turning an age range into a targeting decision

An age estimate is only worth the effort if it changes what you do next. Here's how to translate a range into an action, because "this roof is 15–20 years old" isn't a decision — it's an input to one.

Think in three buckets, set by where the roof sits in its expected life for its type and climate:

• Too new to bother (roughly the first half of expected life, no storm reset pending). Skip for replacement. The exception is a recent severe storm — a five-year-old roof that took baseball hail last month is an inspection, not a skip. Age tells you to pass; weather can override.
• The buy zone (back third of expected life, or any age with credible recent storm exposure). Your best doors. These are roofs old enough that replacement is a real conversation, or younger roofs a storm may have wrecked. Spend your knocks and stamps here.
• Past end of life (clearly beyond expected lifespan, weathered in imagery). The homeowner often already knows. High intent, but also high competition — every other crew can see an obviously-shot roof too, so your edge is getting there with something specific to say rather than just "your roof looks old."

The buy-zone definition shifts with shingle type and climate. A 3-tab roof in a hot, high-UV, hail-prone market enters the buy zone years earlier than an architectural roof in a mild one. This is why the build-era guess from earlier matters: it tells you which lifespan clock to run, so the same chronological age can be "too new" on one house and "buy zone" on the next.

And this is where the range-plus-confidence format pays off operationally. A wide, low-confidence range that straddles the buy-zone line is a house to verify further before you spend a knock — pull the permit, get fresher imagery. A tight, high-confidence range sitting squarely in the buy zone is a house to act on now. The confidence level tells your rep whether the next move is more research or a knock, which is the difference between a route that runs efficiently and one that thrashes.

One more practical note on sequencing a route: sort by cohort and storm, not by individual house. Because tract neighborhoods reset in waves, your best use of a morning is usually a block of original roofs in a storm-hit cohort, worked door to door, rather than scattered one-off due houses across town. The age work tells you which roofs; the cohort and storm logic tells you which streets, and streets are what a crew can actually walk.

The bottom line

The year a home was built is a real signal wearing the costume of a precise one. It hands you an upper bound, a guess at the original roofing system, and a neighborhood cohort — and that's genuinely useful. What it does not hand you is the roof's age, because it can't see the storm replacement, the sale-driven re-roof, the quiet age-out, or the overlay that reset the clock.

The fix isn't a better formula. It's a habit: treat the build year as a hypothesis, spend ninety cheap seconds trying to disprove it with sale history, imagery, permits, and per-roof storm exposure, and write your answer as a range with a confidence level rather than a date. Stack the signals cheapest-first, stop when you can act, and save your expensive minutes — the knocking, the climbing, the estimating — for the roofs that cleared the desk.

Do that and two things change. Your reps stop burning knocks on roofs that got redone after the last sale, and they stop walking past the borderline-on-paper houses that are actually the best doors on the street. You stop targeting the calendar and start targeting the roof. That's the whole job — knowing which roofs are actually due, before anyone gets on a ladder.

FAQ

Can I just subtract the year built from the current year to get roof age?

Only as a starting guess, and only for ruling houses out. That subtraction is the roof's maximum possible age, not its actual age — it's correct only if today's roof is the original one. Storm replacements, sale-driven re-roofs, ordinary age-out replacements, and overlays all reset the roof clock while the build year stays frozen. Use the math to confidently skip new construction, then verify with sale history, imagery, and permits before trusting it on older homes.

What's the most reliable free way to tell how old a roof is from my desk?

Stack three free-or-cheap signals: the year built (your upper bound), the last sale date (a recent sale on an old home flags a possible re-roof), and aerial/street imagery (read for streaking, granule sheen, patching, and ridge crispness — but always check the image's capture date). When a house is borderline, a roofing permit pull is the strong tiebreaker because it can give you a near-exact replacement date.

How do I find a roof replacement permit for an address?

Identify the jurisdiction that issues permits (city, county, or township), then search that building department's online permit portal by street address or parcel number. Look for entries like reroof, roof replacement, tear-off, or roof covering, and note the final inspection date — that's closest to when the roof was finished. Cross-check the county assessor for improvement records and parcel details. Note that absence of a permit doesn't prove the roof is original; many re-roofs were never permitted or predate digitized records.

How long do asphalt shingle roofs actually last?

It's a range, not a fixed number. Basic 3-tab shingles tend to run roughly 15–20 years and architectural/dimensional shingles roughly 25–30, but the National Roofing Contractors Association points to climate, roof pitch, and installation quality as the factors that push a roof toward the high or low end. Manufacturer ratings assume ideal conditions; real-world weather and ventilation usually pull actual lifespan shorter. Use the shingle type to pick which nominal lifespan to apply, then adjust for local climate and storm history.

Why does 'year built' in public records sometimes seem wrong?

Because the term isn't defined consistently. Depending on the jurisdiction, year built can mean permit-pull year, construction-start year, completion year, or 'effective year built' — an adjusted figure some assessors use after major renovations to reflect a property's condition. Effective year built can already bake in remodels, which throws off naive age math. Confirm with the assessor what the number actually represents before you subtract from it.

How can I tell a roof's age from aerial imagery?

Look for streaking and uneven discoloration, a granule-loss sheen or smoother patches where asphalt shows through, slopes that are a different color (partial replacement), sagging or wavy ridge lines, and soft versus crisp ridge caps and edges. Crisp, uniform ridges and even color suggest a newer roof; weathered, streaked, patched surfaces suggest an older one. The big caveat: imagery has a capture date, sometimes years old, so a recent storm may not show up — check the date before you trust the picture.

Two roofs on the same street are the same age — are they equally due?

Not necessarily. Hail and wind don't hit uniformly. Stone size, fall angle, wind-driving, roof pitch, and which direction a slope faces mean one roof can be pulverized while its same-age neighbor is barely touched. That's why 'the ZIP got hail' is a blunt signal — the roofs a storm actually wore out are a specific subset based on per-roof exposure, not the whole footprint. Pair the age estimate with storm exposure modeled per house to rank true due-ness.

Can software tell me the exact age of a roof by address?

No — and be wary of anything that claims an exact date from imagery. Tools like RoofPredict estimate a roof-age range per address from aerial imagery and add storm exposure modeled per roof, then rank which houses are most likely due. That's a strong targeting signal that scales across a whole route, but it's a range and a probability, not a confirmed install date and not a damage finding. A hand-pulled permit can beat it on a single house, and an on-roof inspection is still ground truth.

Does a storm model prove a roof was damaged?

No. Modeling that a roof was likely hit hard tells you it's worth inspecting — it's odds, not proof. It is not an adjuster's finding or a damage report, and it shouldn't be presented to a homeowner as established fact. You still perform the inspection and document conditions; the insurer makes coverage decisions and the homeowner owns the claim. The model's job is to point your crew at the right roofs, not to settle anything about damage.

If I can't find a permit and the imagery is old, should I skip the house?

Not automatically. A missing permit is no information, not proof of an original roof, and stale imagery can't see a recent storm. If the build year says the roof is old, the home hasn't sold recently, and the street took a recent severe storm, that house may be a strong knock even with weak desk signals — you just verify on-site. Reserve confident skips for clearly new construction or houses where a permit, recent sale, or fresh imagery shows a young roof.