Start here, because it saves an argument with a vendor later: no county in the United States records when a roof was last replaced.
Assessors record a build year. Many record a roof material. Almost none record a roof date, and none of them update one when a crew tears off shingles on a Tuesday. So every product that shows you a roof age has inferred it, and the useful skill is knowing which inference you are looking at and how often that particular one is wrong.
There are three, and they are not close in quality.
The evidence ladder
Grade A — a recorded roofing permit. A permit pulled on the property, dated, classified as roofing. This is not an inference about the roof’s age, it is a record of the last time someone worked on it. Everything else on this page is a proxy for this.
The scale is national: roughly 13.7 million roofing permits across 1,312 counties, 12.9 million of them residential, with 732,943 residential roofing permits issued in the trailing twelve months. The full breakdown by year, state and interval is in roof replacement statistics.
Grade B — visible roof condition. Aerial imagery can show a roof that is streaked, patched, missing tabs or discoloured in planes. This is direct evidence about the physical object, which permits are not, and it catches the roof that was replaced without a permit. What it cannot do is give you a date, and condition varies with material and climate far more than most people assume.
Grade C — build year, material and climate. The fallback when there is no permit and no usable image: assume the roof is original, apply a typical service life for the material, and see whether the house is past it. It is the weakest link in the chain and it is what most cheap “roof age” fields are actually made of.
| Evidence | Gives you | Fails when |
|---|---|---|
| Roofing permit | A date, and often a declared valuation | The work was unpermitted, or the jurisdiction does not publish |
| Imagery condition | Present-tense state of the physical roof | A worn-looking roof is new, or a clean-looking one is twenty years old under a mild sky |
| Build year + material | A prior, when nothing better exists | The roof has been replaced once already — which, on a house built in 1979, it almost certainly has |
Why the fallback is so weak
The median American single-family home was built in 1979 and is 47 years old, across 79,636,129 homes carrying an assessor build year. Half the country predates 1979.
Set that against a median observed replacement interval of 14 years and the arithmetic is unkind: the typical house has had two or three roofs since it was built, and its build year tells you almost nothing about the one currently on it. Build-year inference is only defensible on genuinely young stock, and even there it is a prior rather than a fact. American housing stock age has the full distribution, including the states where the median home predates 1960.
Fourteen years is a national number, and you do not sell nationally
Observed intervals between consecutive roofing permits on the same home vary by state in a way that changes what “old” means:
| State | Median observed interval | Intervals observed |
|---|---|---|
| Ohio | 8 years | 5,123 |
| Texas | 9 years | 8,725 |
| Colorado | 10 years | 18,595 |
| Michigan | 11 years | 5,099 |
| Minnesota | 12 years | 15,477 |
| California | 12 years | 35,161 |
| Florida | 16 years | 185,740 |
A ten-year-old roof in Ohio is past the local median. The same roof in Florida is nowhere near it. Any scoring rule with a single national threshold will over-target one of those markets and under-target the other, and the fix is to hold the threshold per state rather than per product.
The reason for the spread is not shingle chemistry. Ohio, Texas and Colorado sit in or near the hail belt, where insurance-funded replacement after a storm resets the clock early and often. Which produces a second, less obvious consequence: in those markets the last permit date is frequently a storm date, not a wear-out date, and the households that just had a claim-funded replacement are exactly the ones you should be suppressing. Storm damage leads is the other half of that motion.
Material changes the number, and it is on the record more often than roof age
Roof material is recorded by many assessors — for 50,687,261 single-family homes nationally, of which 77.4% carry asphalt or composition shingle. Metal is 6.3%, tile or concrete 5.4%, wood shake 3.2%, slate 0.4%.
For most of the country that means the material field adds little, because almost everything is asphalt and the interval assumptions were built on asphalt. Where it matters is the exceptions, and they are concentrated: Nevada is 69.7% tile or concrete, Vermont is 29.3% metal, and a tile or standing-seam roof on a twenty-year-old house is not a replacement prospect on the same schedule as a shingle one. Running an asphalt-shaped threshold across Nevada produces a territory of bad doors. Roofing material statistics has the state and decade breakdown, including the 405,280 homes still recorded with an asbestos roof material.
A scoring recipe that survives contact with a territory
Build it as a ladder rather than a single filter, so every home lands on the best evidence available for it and you know which rung it came from.
- If a roofing permit exists, use its date. Score the home on years since that permit against your state’s median observed interval, not against fourteen.
- Suppress recent claim-funded work. A permit inside the last three years in a hail market is a stop, not a lead.
- If no permit exists but imagery is current, score on visible condition and mark the record as condition-derived so nobody quotes a date at the door.
- If neither exists, fall back to build year plus material — and hold this tier separately. It is a prospecting pool, not a list you should be making claims about.
- Add the second signal. Roof age alone is a large, undifferentiated population. Pair it with recent purchase, tenure, or storm exposure and it becomes a reason to call rather than a demographic.
The discipline that makes this work is carrying the rung through to the rep. A file with a roof age column and no provenance column will produce someone saying “our records show your roof was replaced in 2011” about a house where the records show nothing of the kind.
Volume, so you can size it
732,943 residential roofing permits were issued in the trailing twelve months, and the median home carrying one was 42 years old at the time — a quarter of permits go to homes under 25. That is the population that already replaced. The prospecting question is the inverse of it: which homes look like the ones about to.
At a median declared valuation of $15,000, a scoring rule that moves your contact rate by a point is worth more than most of what a CRM migration buys. How to get roofing leads covers the channels this list feeds, and roofing sales scripts covers what to say once the door opens.
Where Scout Data fits
Disclosure: Scout Data is our product, and this is the list it was built to produce. List Builder holds roofing permit history across 1,312 counties joined to roof material, build year, ownership and current roof imagery, so the ladder above is a set of filters rather than four separate purchases — delivered with phone numbers matched by name to the owner on title and scrubbed against the federal do-not-call registry. Scout puts the same score on a map, one street at a time.
What it does not do is tell you a roof age nobody recorded. Where the evidence is a permit, you get a date. Where it is imagery, you get a condition. Where it is neither, you get a prior, labelled as one.
Frequently asked questions
Is roof age recorded anywhere?
Not as a field, anywhere, at national scale. County assessors record the year a house was built and often the roof material; almost none record when the roof was last replaced. Roof age is always inferred, and the only question is what it is inferred from — which is why two vendors can show you different roof ages for the same house and neither is lying.
How old is a roof before it needs replacing?
Among homes where two or more roofing permits are observed, the median interval between them is 14 years, across 309,255 observed intervals — with a quarter at 7 years or shorter and a quarter at 20 or longer. That sample skews short, because a roof lasting 30 years rarely shows both permits inside a coverage window, so treat 14 years as a floor on typical roof life rather than an average.
How can I tell a roof’s age from the outside?
You can grade its condition, which is related but not the same thing. A roof that is visibly cupping, streaked or patchy is telling you something real; a roof that looks fine may be a twenty-year-old architectural shingle in a mild climate or a five-year-old three-tab that hailed last spring. Condition is a good tiebreaker on top of a recorded date and a poor substitute for one.
What does a roof replacement cost?
The national median declared valuation on residential roofing permits issued since January 2024 is $15,000, running from about $10,900 in Kansas to $18,400 in New York across the highest-volume states. Declared permit valuations sit below real contract prices in many jurisdictions, so treat those as conservative floors rather than quotes.