How Many Roofs in a City Need Replacing? A Contractor's Method for Estimating the Real Number

Walk into any city of 100,000 households and ask how many of those roofs need replacing, and you'll get a shrug. The honest answer is that nobody knows the exact number, including the manufacturers, the insurers, and the guys who've worked that ZIP code for twenty years. But "nobody knows exactly" is very different from "you can't estimate it." You absolutely can, and the estimate matters more than most owners realize, because it's the difference between hiring a third crew based on a hunch and hiring one based on a defensible number.

A roof-replacement market is a flow, not a stock. Every year a fresh slice of the housing stock crosses from "fine" into "due," a storm pulls some roofs forward out of turn, and a smaller slice gets replaced ahead of schedule for a remodel or a sale. If you can size that annual flow for your city, you can size your sales target, your marketing budget, your crew count, and the number of doors a canvasser has to knock to fill a week. This is a builder's-math problem, and I'm going to walk through the exact arithmetic I use, the data sources that feed it, the mistakes that wreck the number, and how I tighten a city-wide estimate down to the street.

The goal isn't a number to put in a pitch deck. It's a number you'd bet payroll on.

The mental model: roofs are a flow, not a stock

Start with the single most useful idea in this whole exercise. Your city has some total count of roofs, but that total is almost useless to you. You will never sell to all of them, and most of them are nowhere near due. What you sell into is the replacement rate — the share of that stock that ages or storms its way into the buy zone in a given year.

Think of it like a reservoir. The water level is the total housing stock, and it barely moves. What you fish in is the inflow: the roofs crossing into "due" this year. In a steady-state city with no storms and no growth, that inflow is governed by one number — the average service life of the roofs out there. If the typical roof lasts about 20 years, then in any given year roughly one-twentieth of them, around 5 percent, are reaching the end of their useful life. Stretch the average life to 25 years and the natural rate drops to about 4 percent. Compress it to 17 years with cheap three-tab shingles in a hot climate and it climbs past 5.8 percent.

That one fraction — 1 divided by the average roof life in years — is the backbone of every city estimate you'll ever build. Everything else is an adjustment on top of it: storms that pull demand forward, the mix of roofing materials, the age profile of the neighborhood, and the share of homeowners who actually act when their roof is technically due.

Hold that picture. We're going to make it concrete.

Step one: count the roofs you actually have

Before you can estimate a rate, you need a denominator — how many roofs exist in your market. "Roofs" is not the same as "people" or "households," and conflating them is the first place estimates go wrong.

The cleanest public starting point is the U.S. Census Bureau's American Community Survey (ACS), which publishes housing-unit counts for every city, county, and ZIP-code tabulation area. A "housing unit" is close to what you want for residential roofing, but read it carefully:

• Single-family detached homes are one roof, one decision-maker, one sale. This is the heart of the residential replacement market.
• Single-family attached (townhomes, rowhouses) may be one roof over several units or one unit per roof. Treat carefully; an HOA often owns the decision.
• Multi-family buildings (2 to 50+ units) are far fewer roofs than units. A 24-unit apartment building is one roof and one commercial-style decision, not 24 residential sales.
• Mobile and manufactured homes are a different roofing market entirely.

So the move is to pull the ACS table for housing units by units in structure and isolate the 1-unit detached count. That's your honest residential roof count. In a typical mid-sized American city, single-family detached homes run somewhere between 55 and 75 percent of all housing units, but it swings hard by metro — a Sun Belt suburb might be 80 percent detached, while an older Northeastern city core might be under 40 percent. Don't guess the ratio; pull it.

A worked example I'll carry through the rest of the piece: imagine a city with 140,000 total housing units, of which 96,000 are single-family detached (about 69 percent). That 96,000 is our denominator. Multi-family, you'd estimate separately as a commercial line, because the sales motion, the crew, and the deal size are nothing alike.

One more refinement worth the five minutes: county assessor parcel data. Most county assessor and GIS offices publish parcel records with a "year built" field and a land-use code. That gives you both a roof count and an age distribution in one download, which feeds the next step directly. Where assessor data is clean, prefer it over ACS for the denominator, because it's parcel-level rather than survey-estimated.

Step two: pick a defensible average roof life

Now the engine. The natural replacement rate is 1 divided by average service life, so your whole estimate hinges on what "average life" you plug in. This is where contractors either anchor on a brochure number or on hard-won field reality. Use field reality, adjusted by what's actually on the roofs in your market.

Here's a working table of service-life ranges. These are real-world replacement points — when a roof in normal conditions is genuinely due, not the theoretical lab maximum a manufacturer prints on a 50-year limited warranty.

The National Roofing Contractors Association (NRCA) and shingle manufacturers both publish life-expectancy guidance, and it's worth reading, but temper it with three local multipliers that move the real number a lot:

1. Climate and sun load. Asphalt shingles age on a thermal clock. A south-facing slope in Phoenix or central Texas can be cooked years before the same shingle in coastal Oregon. UV and heat-cycling drive granule loss and mat embrittlement. In hot, high-UV markets, knock 2 to 4 years off the table above.
2. Hail and wind history. Storms don't just cause sudden replacements; chronic small hail and wind ages roofs faster between the big events. A market in the hail belt runs a structurally younger replacement age than the same material in a calm climate.
3. Build era and quality. A neighborhood that went up in a single boom year was very likely roofed with the same builder-grade product on the same week. That whole tract ages together and comes due together — which is a gift to a contractor who spots it.

For our example city, say the housing stock is dominated by architectural asphalt with a meaningful slice of older 3-tab, in a moderate-hail, moderate-sun climate. A blended average service life of 22 years is defensible. That gives a natural replacement rate of 1 ÷ 22 = 4.5 percent per year.

Apply it: 96,000 detached homes × 4.5 percent = about 4,300 roofs per year reaching natural end-of-life in this city. That's the baseline flow before storms, before remodels, before anything else.

Sit with that number for a second, because it reframes everything. It is not 96,000 prospects. It's roughly 4,300 roofs a year that the calendar alone is pushing into your market — and that's the pool every contractor in town is fishing in whether they've quantified it or not.

Step three: layer in the age curve (the part most people skip)

The flat-rate method above assumes roofs are evenly spread across all ages — that just as many are 1 year old as 21 years old. Real cities are lumpy. They were built in waves: a postwar wave, a 1970s wave, a late-90s subdivision wave, a mid-2000s boom, and so on. Roofs follow construction with a lag, and re-roofs cluster too, because a tract roofed together gets re-roofed together.

That lumpiness is the single biggest reason a city-wide average misleads you at the neighborhood level. Two subdivisions five miles apart can have wildly different current replacement rates even though they share the same long-run average.

The fix is to build a rough age histogram of the roofs rather than settling for a single average. Two ways to get it:

• Year-built from assessor data, then add the typical roof life to estimate when each home's original roof came due — and recognize that homes past one roof-life are now on their second or third covering, cycling on the same clock.
• ACS "year structure built" tables, which bucket the housing stock by decade. Coarser, but free and fast.

Here's the logic that matters. Take any cohort of homes built in, say, 2003. With a 22-year roof life, the original roofs on those homes came due around 2025. The homes built in 1981 are now (in 2026) 45 years old — they're on their second roof, which itself came due around 2003, and a third around 2025. So the homes that are "hot" in a given year aren't only the ones a single roof-life back; they're every cohort sitting near a multiple of the roof-life from today.

When you actually map this, you find that the city-wide 4.5 percent rate hides a range: some neighborhoods are running at 1 percent this year because their roofs are mid-life, and others are running at 8 to 10 percent because an entire build-wave is coming due at once. The total still sums to roughly your 4,300 — but it is not spread like peanut butter. It's concentrated, and the concentration is where you point crews.

This is the difference between a market number and a targeting number. The city total tells you whether to be in the market at all. The age curve tells you which four subdivisions to canvass first.

Step four: add storm-driven demand on top

Everything so far is the slow, predictable clock. Storms are the fast, lumpy accelerant, and in many markets they're the larger share of replacement volume in any year they hit.

A hailstorm doesn't change the long-run average roof life much, but it yanks a year or two of demand into a single week and adds roofs that weren't naturally due at all. A severe hail event over a populated area can put a meaningful fraction of the roofs underneath it into a replacement conversation — driven by functional damage to the shingle mat and granule layer, the kind a trained eye and a documented inspection identify. The damage isn't always visible from the street, which is exactly why thorough documentation matters.

To size storm demand, you stack two things: how often a damaging storm hits your footprint, and how many roofs sit under a typical track.

• Frequency comes from NOAA's Storm Prediction Center and the National Weather Service storm-events database, which log hail and wind reports by date, location, and magnitude going back decades. Pull your county's hail history and you can see how many days a year produce hail at or above the size that does shingle damage (roughly one inch and up, with damage probability rising fast with stone size).
• Exposure is your own roof-density map. A storm cell a few miles wide passing over a dense subdivision exposes thousands of roofs; the same cell over farmland exposes a handful.

The honest way to talk about storm demand is in odds, not certainties. You don't get to say "this storm replaced 2,000 roofs." You say "a storm of this size over this density historically converts into replacement work at some rate, and here's the band." Treat it as a forecast distribution, not a guaranteed count, and your planning gets a lot more honest.

For our example city in a moderate-hail belt, suppose the county sees a roof-damaging hail event over a populated area roughly once every two to three years, and a meaningful event adds, conservatively, on the order of 1,500 to 4,000 roofs to the replacement pool in the year it lands — concentrated under the track, not spread city-wide. Averaged across the storm cycle, that's perhaps another 800 to 1,500 roofs per year of storm-driven demand layered on top of the 4,300 natural baseline.

The critical operational point: storm demand and age demand don't arrive in the same places or on the same calendar. Age demand is a steady drip you can route against all year. Storm demand is a flash flood under a specific track that you have to mobilize for in days. A shop that plans only for one and not the other is either chronically under-booked between storms or chronically overwhelmed after them.

What actually moves the replacement rate up or down

When you compare two cities and their natural rates differ, it traces back to a short list of physical and economic drivers. Knowing them lets you adjust a rate for a new market without rebuilding the whole model.

• Material mix. A market still heavy with old 3-tab runs a higher rate than one that's mostly newer architectural shingle, simply because the average life is shorter. As a region's housing stock turns over to longer-lived product, its natural rate slowly drifts down.
• Climate severity. High UV, big temperature swings, and humidity all accelerate asphalt aging. A baking inland market and a mild coastal one with identical shingles will have different real service lives and therefore different rates.
• Storm frequency. Beyond the event spikes, chronic small hail and high wind shave years off roofs continuously, lifting the structural rate in storm-belt markets even between named events.
• Build-era concentration. A city that grew in a single decade has a lumpier, spikier rate as those cohorts hit roof-life together; a city that grew steadily over fifty years has a smoother, more predictable flow.
• Economic willingness. A wealthier, owner-occupied market converts due roofs to jobs faster; a market with more deferred maintenance and rentals carries a larger backlog of overdue-but-unreplaced roofs, which can actually become a surge when an event or a tightening insurance market forces the issue.

When you carry an estimate from one city to the next, walk this list and adjust each lever rather than assuming the rate travels unchanged. Two cities of the same population can differ by a full point or more on the natural rate alone, which is the difference between a thin market and a thick one.

A full worked estimate, start to finish

Let me run the entire calculation for the example city in one place, so you can copy the skeleton onto your own market.

Inputs

• Total housing units (ACS): 140,000
• Single-family detached share: 69% → 96,000 detached roofs
• Blended average roof life (field-adjusted): 22 years
• Natural replacement rate: 1 ÷ 22 = 4.5%
• Storm cycle: damaging hail roughly every 2.5 years; ~2,500 roofs added per event under track

Baseline (age/wear) demand

96,000 × 4.5% = 4,320 roofs/year naturally coming due.

Storm demand (annualized)

2,500 roofs per event ÷ 2.5 years between events = 1,000 roofs/year averaged across the cycle. (In the actual storm year it's 2,500-plus and concentrated; in the quiet years it's near zero. The annualized figure is for planning, not for any single year.)

Total annual replacement pool

4,320 + 1,000 = ~5,320 roofs/year in the addressable detached-home market.

From pool to your number

Now translate the market into a business plan. Not every "due" roof gets replaced this year — homeowners defer, sell as-is, or patch. A reasonable act-this-year conversion on the naturally-due pool is well under 100 percent; many roofs limp along years past technically-due. Storm-damaged roofs convert at a much higher rate because there's an event forcing the decision. Blend it and you might assume, say, 60 to 75 percent of the pool actually transacts in a given year across all contractors. Take 70 percent:

5,320 × 70% = ~3,700 roofs actually replaced citywide per year, across every contractor competing in the market.

Your slice

If there are, say, 40 active roofing companies of various sizes serving the city and you're a mid-sized shop, your fair share might be 2 to 4 percent of the transacting market, or roughly 75 to 150 roofs a year — before you do anything special to win more than your share. That's your floor. Everything in sales, marketing, and targeting is about beating that fair-share number.

That chain — housing units → detached roofs → natural rate → storm add → act-this-year conversion → competitive share — is the whole model. Each link is a number you can defend with a source or a stated assumption, which means you can argue about the assumptions instead of the conclusion. That's what makes it a planning tool instead of a guess.

Turning the city number into doors a canvasser can knock

A market estimate that lives in a spreadsheet doesn't book jobs. The next move is converting it into field activity, and the bridge is the concentration you found in step three.

If 4,320 natural roofs are spread across 96,000 homes, a blind door-knock has a 4.5 percent chance of landing on a due roof — and a much lower chance of landing on one that's due and the owner will act and you catch them home. Those are brutal odds, and they're why untargeted canvassing burns out reps.

But you don't have to knock blind. You already established that the due roofs cluster in specific build-wave cohorts. So the real workflow is:

1. Rank neighborhoods by current replacement rate, not by the long-run average. The 2003 subdivision at 22 years old is a target this year; the 2015 subdivision is not, no matter how nice the homes are.
2. Within a hot neighborhood, rank addresses by roof age and condition signal. A roof at the high end of its life range outranks one at the low end.
3. Overlay the most recent storm track so that storm-exposed roofs in those neighborhoods float to the top.
4. Sequence the route so a crew or canvasser hits the densest cluster of high-rank addresses with the least driving between them.

Do that and the canvasser is no longer working a 4.5 percent base rate; they're working a list where a large share of doors are genuinely due. The city number told you the market is worth ~3,700 roofs a year. The ranked list tells one rep which 60 houses to knock on Tuesday.

This is exactly where having roof-age and storm data per address stops being a nice-to-have and becomes the difference between a productive week and a wasted one.

A worked canvass-math example

Let me make the door math concrete, because owners consistently overestimate how many roofs blind knocking finds. Say a rep works a six-hour canvassing block and physically reaches 60 doors. At a blind 4.5 percent due-rate, about 2.7 of those doors sit on a genuinely due roof. But "due" isn't "booked." Apply a contact rate — maybe half the homes answer — then an act-this-year willingness on the due ones, then your close rate on the conversation. Run it through:

• 60 doors knocked
• × 4.5% due → 2.7 due roofs behind those doors
• × 50% home and willing to talk → 1.35 conversations with a due owner
• × 60% act-this-year on a genuinely-due roof → 0.8 real opportunities
• × 35% close → roughly 0.28 jobs per six-hour block

That's about one booked job per three or four full canvassing days on a blind list. It's why reps quit. Now rerun it on a ranked list where, instead of 4.5 percent, 35 percent of the doors are genuinely due because you sequenced into a build-wave cohort that's coming up on its roof-life:

• 60 doors × 35% due → 21 due roofs
• × 50% home and willing → 10.5 conversations
• × 60% act-this-year → 6.3 opportunities
• × 35% close → about 2.2 jobs per block

Same rep, same hours, same close rate — roughly eight times the booked work, purely from where the list pointed them. That multiplier is the entire economic argument for sizing the age curve and storm overlay before anyone laces up their boots. The city-total number justifies being in the market; the targeting math justifies how you spend every canvassing hour inside it.

Direct mail responds to the same lever

The identical logic governs mail. A blanket mailer to 96,000 homes at, say, 60 cents all-in costs about $58,000 per drop and lands on a roof that's due only 4.5 percent of the time. Cut the same budget down to the 8,000 homes in the build-wave cohorts and under the latest storm track, and you mail 8,000 pieces for under $5,000 while reaching a list where a large share are due. Same response rate on a far richer list means a dramatically lower cost per booked job — or, held at the old budget, many more touches on the right homes. Enriching the list with roof-age and storm signals is the move that turns mail from a rounding error into a channel.

Where RoofPredict fits in this workflow

Everything I've described — the housing counts, the age curves, the storm history — you can assemble by hand from public data, and you should understand it that way so you trust the number. The catch is doing it at address-level resolution, across a whole city, and keeping it current. That's the gap RoofPredict is built to close.

RoofPredict scores roofs house-by-house so you can see which ones are due. For each address it estimates a roof-age range from aerial imagery — a range, deliberately, not a precise install date, because no one can read an exact date off a photo and pretending otherwise would be dishonest. On top of that it models storm physics per roof: which specific roofs sat under which hail and wind events, and how that exposure stacks with age. The output is a ranked view of doors, routes, and lists — the roofs the storm wore out plus the roofs aging out, sorted so your crews target the ripe ones first.

It's worth being clear about what this is and isn't. It is a prioritization and enrichment layer. It ranks your territory and enriches your own CRM or mailing list with roof-age and storm signals so your existing prospecting works harder. It is not a lead-buying service handing you pre-sold appointments, and the storm modeling speaks in odds — which roofs are likely to qualify for a closer look — never in proof that a given roof is damaged. You still inspect. You still document. The data tells you where to point.

In the city-sizing exercise above, RoofPredict's role is steps three and four made automatic and current: the age curve and the storm overlay, resolved to the individual address, so the ~3,700-roofs-a-year market turns into an actual ranked list of the specific houses to work this month. The honest limit is that a model gives you a probability-weighted target, not a guarantee — it gets your reps onto the right streets and the right doors; closing is still the job.

Validating the model against ground truth

A model you can't check is a model you can't trust, and there are a few cheap ways to sanity-check the city number against reality before you bet payroll on it.

Permit pull. Most jurisdictions require a permit for a full roof replacement, and many publish permit data. Pull the count of residential re-roof permits issued in the city over the last few years. If your model says ~3,700 roofs transact a year and permits run ~3,000 to 4,000, you're in the right neighborhood. If permits run 800, either your roof-life assumption is too short, your conversion is too high, or — common in some areas — a lot of work is going unpermitted, which is itself useful intelligence about the market. Permit data also reveals seasonality and the storm spikes directly: you'll see the permit count jump in the months after a hail event, which both validates your storm add and tells you the lag between an event and the actual replacement.

Supplier and distributor signal. Roofing distributors know roughly how many squares of shingle move through a metro. A square covers 100 square feet, and a typical detached roof runs 20 to 30 squares. If a distributor rep will share ballpark regional volume, dividing squares by an average roof size gives an independent roof-count estimate to triangulate against your demographic build-up. The two methods coming within shouting distance of each other is a strong signal the number is sound.

Your own historical close data. If you've worked the market for years, you already have a revealed sample. Look at the age and storm profile of the roofs you actually sold last year. If your sold roofs cluster at 19 to 24 years old, that validates your service-life assumption from the field side. If you're closing a lot of 14-year-old roofs, either there's more storm damage in your mix than you modeled or your market runs cheaper, shorter-lived product than you assumed.

Triangulating the demographic model against permits, distributor volume, and your own sold-job history turns a spreadsheet estimate into a number with three independent confirmations behind it. That's the version you take to a lender or a partner.

Commercial and multi-family: a separate count, a separate motion

I've kept the main estimate on single-family detached on purpose, because mixing in commercial and multi-family roofs corrupts the number. They belong in their own line, sized their own way.

The roof clock is different. Low-slope commercial systems — TPO, EPDM, modified bitumen, built-up — run on their own life curves, generally in the 15 to 30 year band depending on membrane, thickness, and maintenance, and they fail differently: ponding, seam failure, and flashing rather than granule loss. The decision-maker is different too: a property manager, a REIT, a facilities department, or an HOA board, with a procurement process and often a bid requirement, not a homeowner deciding at the kitchen table. Deal size is larger and the sales cycle is far longer.

For multi-family residential, the count trap from earlier is the whole story: a 40-unit building is a handful of roofs and one decision, not 40 sales. Size multi-family by building count and roof area, not unit count, and treat the larger properties as commercial-style opportunities.

The practical guidance: build the single-family number as your core market — it's the larger count, the faster cycle, and the motion most residential and storm-restoration shops are built for — and size commercial and large multi-family as a separate, smaller line with its own conversion assumptions, its own sales process, and its own crew capability. Reporting one blended "roofs in the city" number that mixes a kitchen-table decision with a board procurement will mislead every plan downstream of it.

How often to refresh the number

A market estimate isn't a one-time exercise; it decays. Two clocks keep moving under it.

The age curve advances a year every year — last year's not-quite-due cohort is this year's prime target, and a fresh slice of new construction enters the pool at the bottom. That alone argues for re-running the baseline annually, ideally before you set the next year's sales target and marketing budget.

The storm clock moves on its own schedule and demands faster updates. A single hail event reshuffles your entire targeting overnight: the addresses under that track jump to the top of every list regardless of their age. That's not an annual refresh — that's a within-days remap after every significant event. The shops that win storm work are the ones who already know their roof-density map and can overlay a new track on it the same week, rather than starting their analysis after competitors are already knocking.

The practical cadence: re-run the full city baseline once a year, and keep a standing storm overlay you can refresh inside a week of any damaging event. The age side is a calendar appointment; the storm side is a fire drill you've rehearsed.

Reading the assessor and census data without getting fooled

A few traps in the source data swallow good estimates. Knowing them is half the battle.

"Year built" is the structure, not the roof. A 1975 house may be on its third roof. Year-built tells you the original roof timeline and the cohort; it does not tell you the current covering's age. That's precisely why aerial-derived roof age beats assessor year-built for current targeting — the assessor knows when the house went up, not when the roof was last redone.

Housing units overstate roofs in dense areas. As covered, multi-family inflates the unit count relative to roof count. Always split out 1-unit detached before you multiply by a rate.

Vacancy and tenure matter for conversion, not for the pool. ACS reports owner-occupied vs renter and vacancy rates. A high-rental neighborhood has the same roof-aging clock but a different decision-maker (the landlord) and often a slower act-this-year rate. Use tenure to adjust conversion, not the underlying due-count.

New construction dilutes the rate. A fast-growing city is adding brand-new roofs that won't be due for two decades. If 8 percent of your stock was built in the last five years, those homes are dead weight in your near-term pool even though they're in the housing count. Either exclude recent-build cohorts or weight them to near-zero replacement probability.

Survey estimates have error bars. ACS five-year estimates for small areas carry margins of error that can be large for a single ZIP. For a city-wide number it averages out; for a single small neighborhood, treat it as a band, not a point.

A checklist you can run on any new city in an afternoon

When I size a new market, this is the actual sequence. It takes an afternoon the first time and an hour once you've done it twice.

1. Pull total housing units for the city or county from ACS.
2. Pull units-in-structure and isolate 1-unit detached. That's your roof denominator.
3. Pull year-structure-built buckets (or county assessor year-built) to build a rough age histogram.
4. Choose a blended average roof life from the material mix and climate — start from the service-life table, then adjust for sun load and hail history.
5. Compute the natural rate as 1 ÷ average life, and multiply by the detached count for the baseline annual due-count.
6. Pull county hail/wind history from NOAA SPC and NWS storm events; estimate event frequency and a per-event roof add under the track.
7. Annualize storm demand across the storm cycle and add it to the baseline.
8. Apply an act-this-year conversion (adjusted by owner-occupancy) to get roofs actually transacting.
9. Estimate the competitor count and your fair-share percentage to get your floor.
10. Rank neighborhoods by current rate, not average, and identify the build-wave cohorts coming due now.
11. Overlay the latest storm tracks to float exposed roofs to the top.
12. Convert the top cohorts into routes a canvasser or crew can actually work.

Steps 1 through 9 give you the market number you'd bet payroll on. Steps 10 through 12 turn it into next week's schedule.

Common mistakes that wreck the estimate

I've watched a lot of owners build this number badly. The recurring failures:

• Using households or population as the roof count. Population divided by household size is not roofs. Detached-unit count is roofs.
• Anchoring on the warranty number for roof life. A "30-year shingle" does not last 30 years in the field on a hot slope. Use real replacement points.
• Treating the city average as the neighborhood rate. The average hides the lumps, and the lumps are the whole opportunity.
• Counting storm demand as steady. It is not. It's a flood under a track in a storm year and a trickle otherwise. Annualize for planning, but mobilize for the event.
• Ignoring conversion. "Due" is not "sold." A large share of technically-due roofs keep going for years. Your transacting market is smaller than your due pool.
• Forgetting competition. The whole city's transacting roofs are not yours. Start from fair share and earn the rest.
• Letting the number go stale. Roofs age and storms hit continuously. A market estimate from two years ago has the wrong age curve and is missing two storm seasons.

Every one of these pushes the estimate in a predictable direction, so once you know them you can sanity-check a number just by asking which mistakes it might be making.

How the city number drives the rest of the business

The payoff for getting this right is that one estimate feeds every other plan you have to make.

Crew capacity. If your transacting fair-share floor is ~120 roofs a year and one crew installs ~80 to 120 residential roofs annually depending on complexity and season, you can see immediately whether your growth target needs another crew or just better targeting of the existing one.

Marketing budget. If you need to win, say, 200 roofs against a fair share of 120, you need to out-target competitors for ~80 incremental roofs. Knowing the size and concentration of the due pool tells you whether to spend on canvassing labor, direct mail to the hot cohorts, or storm-response readiness — and roughly how much each incremental roof can cost you and still pencil.

Direct mail efficiency. Blanketing 96,000 homes is wasteful when 4,300 are due. Mailing the ranked hot cohorts and storm-exposed addresses can cut the mail volume by an order of magnitude at the same booked-job count. Enriching your mailing list with roof-age and storm signals is exactly the kind of targeting that turns a money-losing mail program into a profitable one.

Hiring and seasonality. The age-driven baseline is steady; the storm demand is spiky. Knowing the split tells you how much of your capacity should be permanent crew versus surge capacity you can stand up after an event.

The number isn't academic. It's the input to four or five decisions that each cost or make real money, which is why "about 5 percent of detached homes a year, concentrated in these cohorts, plus storm spikes" is one of the most valuable sentences an owner can be able to say about their own market.

Putting it together

How many roofs in a city need replacing? For a typical city, start with the single-family detached count, multiply by a field-honest natural rate of roughly 4 to 6 percent a year, add annualized storm demand on top, and discount for the homeowners who won't act this year. In our 96,000-detached-home example that's on the order of 3,700 roofs actually transacting citywide each year — a real, defensible, payroll-grade number, built from public housing and storm data and a couple of clearly-stated assumptions.

The number that sizes the market is the easy part. The hard part — and the profitable part — is that those roofs aren't spread evenly. They cluster in build-wave cohorts coming due now and under storm tracks that landed last season. The contractors who win aren't the ones who know the city total; they're the ones who know which 60 doors to knock on Tuesday. Sizing the market with the method above tells you the prize is worth chasing. Resolving age and storm signals down to the individual address — by hand if you must, or with a tool like RoofPredict that scores roofs house-by-house and ranks your routes — is how you actually go get it.

Build the city number first. Then go find the streets where this year's roofs are hiding.

FAQ

What percentage of roofs in a city need replacing each year?

As a planning baseline, roughly 4 to 6 percent of single-family detached roofs reach natural end-of-life each year, driven mainly by the average service life of the local roofing material. That figure comes from dividing 1 by the average roof life in years: a 20-year average life implies about 5 percent annually, a 25-year average about 4 percent. Storms layer additional demand on top in the years they hit, and a real act-this-year conversion rate means fewer than 100 percent of technically-due roofs actually get replaced in any given year.

How do I count the number of roofs in my city?

Don't use population or household counts. Pull the Census Bureau's American Community Survey table for housing units by units-in-structure and isolate the 1-unit detached count, which is the closest match to the residential roof market. Multi-family buildings have far fewer roofs than units, so they distort the count and should be sized separately. Where available, county assessor or GIS parcel data is even better because it gives a parcel-level roof count plus year-built in one source.

What is the average lifespan of an asphalt shingle roof?

In real-world conditions, 3-tab asphalt shingles typically last about 15 to 18 years, and architectural or dimensional asphalt shingles about 20 to 25 years, with premium products reaching 25 to 30. These are field replacement points, not the longer numbers printed on limited warranties. High UV and heat in hot climates can shorten these by several years, and repeated hail and wind exposure ages roofs faster between major events, so adjust for your local climate and storm history.

How much does storm and hail damage add to replacement demand?

It varies enormously by location and year. Storms don't change long-run roof life much, but a single damaging hail event over a populated area can pull a year or more of demand into a few weeks and add roofs that weren't naturally due. Sizing it requires two inputs: how often damaging hail hits your footprint (from NOAA Storm Prediction Center and National Weather Service storm-event records) and how many roofs sit under a typical storm track. Always express storm demand as a probability band, not a fixed count, and remember it's concentrated under the track rather than spread across the city.

Why does the city-wide average rate mislead at the neighborhood level?

Cities are built in waves, so roofs cluster by age rather than spreading evenly across all ages. A subdivision built in a single boom year was usually roofed with the same product on the same timeline and comes due together. That means some neighborhoods run well above the city average replacement rate in a given year while others run well below, even though they share the same long-run average. The city total tells you whether the market is worth entering; the age curve by neighborhood tells you which streets to canvass first.

How do I turn a market estimate into how many doors to knock?

Rank neighborhoods by their current replacement rate rather than the long-run average, then rank addresses within the hot neighborhoods by roof age and condition signal, then overlay the most recent storm track so exposed roofs float to the top, and finally sequence a route that clusters the highest-ranked addresses with minimal driving. This moves a canvasser off the roughly 4.5 percent blind base rate and onto a list where a large share of doors are genuinely due.

Does year-built from assessor data tell me when a roof needs replacing?

Only indirectly. Year-built tells you when the structure went up and which build-wave cohort it belongs to, but a 45-year-old house may be on its second or third roof, so year-built does not tell you the current covering's age. It's excellent for spotting cohorts coming due, but for current address-level targeting you want an estimate of the actual roof's age, which aerial imagery can provide as a range rather than a precise install date.

How does RoofPredict help size and work a market?

RoofPredict scores roofs house-by-house, estimating a roof-age range per address from aerial imagery and modeling storm physics per roof to show which roofs sat under which hail and wind events. It outputs a ranked view of doors, routes, and lists, and can enrich your existing CRM or mailing list with roof-age and storm signals. It's a prioritization and enrichment layer, not a lead-buying service, and the storm modeling speaks in odds about which roofs likely warrant a closer look. You still inspect and document each roof; the data tells you where to point your crews.

What is a realistic share of the city's replacement market for one contractor?

Start from fair share. Estimate the number of active roofing companies serving the city, and a mid-sized shop's fair share of the transacting market might be a few percent before any special effort. In a market transacting a few thousand roofs a year, that's often on the order of dozens to low hundreds of roofs as a floor. Everything you do in targeting, marketing, and storm response is about beating that fair-share floor, which is why concentration and per-address signals matter so much.

Should roofers help homeowners with insurance claims on storm-damaged roofs?

A roofer can and should inspect thoroughly, document damage with photos, and prepare an accurate, Xactimate-aligned estimate of the repair to their own scope, then hand that documentation to the homeowner. The homeowner files the claim and the insurer decides coverage. A roofer should not, for a fee, negotiate or adjust the claim, interpret policy or coverage, promise a specific payout or approval, promise a deductible will disappear, or represent the homeowner against the insurer, because that is public adjusting and is license-required. Stay on the documentation and accurate-estimate side.