Aerial Imagery for Roofing Sales Leads: A Field Playbook for Picking the Right Doors

Most roofing companies already pay for aerial imagery. They just use it for the wrong half of the job. The reports show up after a homeowner says yes, when a rep needs squares and pitch to build an estimate. That is fine. But the imagery you buy to measure a roof you already sold is the same data that could have told you which roof to walk up to in the first place.

That gap is where a lot of money leaks out of a roofing sales operation. Reps knock 100 doors to set 3 appointments. The canvassing manager picks a neighborhood off a hunch, or because a competitor's yard sign showed up there. Storm season hits and the whole team floods a ZIP code that took graupel and pea-size hail while the real damage sat two miles east. Aerial imagery, used early and used right, tightens every one of those loops.

What follows is a field playbook, not a product brochure. It covers what aerial imagery actually is, what it can and cannot tell you about a roof from above, how to combine it with hail and wind data, how to turn pixels into a route a 22-year-old can run on a Tuesday afternoon, and where the whole approach quietly breaks. I have tried to be honest about the limits, because the contractors who win with this stuff are the ones who know exactly where it stops working.

What "aerial imagery" actually means for a roofer

When people in roofing say "aerial imagery," they are usually lumping together four different things that come from very different places and cost very different amounts. If you do not separate them, you will overpay and over-trust.

Satellite imagery. This is what powers the base layer in most consumer mapping apps. Resolution at the street level is typically somewhere around 0.3 to 0.5 meters per pixel for the best commercial constellations, and often worse. That is enough to see a roof outline, count obvious facets, and spot a tarp or a skylight. It is not enough to see hail bruising, granule loss, or a cracked shingle. Satellite passes also age. The picture you are looking at might be 18 months old, which matters a lot after a storm.

Fixed-wing aerial imagery. Companies fly planes on a grid and capture high-resolution orthophotos, often at 5 to 15 cm per pixel, sometimes from multiple oblique angles (north, south, east, west) so you can see the sides of a structure, rather than only the top-down. This is the workhorse behind most paid roof-measurement reports. Oblique angles are the reason a report can give you pitch and identify dormers and chimneys without anyone climbing a ladder.

Drone (UAS) imagery. This is your own aircraft, flown over a specific property, producing centimeter-level detail and, if you fly a grid, a 3D model you can measure off. Drones are the only aerial option that gets close enough to document actual damage, granule loss, or storm bruising. They are also the most regulated: commercial flights in the US require a remote pilot certificate under the FAA's Part 107 rule, and you have airspace and line-of-sight constraints to respect. Drones are an inspection and documentation tool, not a prospecting-at-scale tool.

Derived data layers. This is the category most people miss. From imagery you can derive roof outlines, square footage, facet counts, predominant pitch, and rough complexity. Combine that with parcel data, year-built records, and storm history and you get something far more useful for sales than a pretty picture: a structured list of properties with attributes you can filter and rank.

For prospecting and lead selection, you live mostly in the satellite, fixed-wing, and derived-data world. Drones come in once a homeowner has agreed to an inspection. Keep that boundary clear, because confusing "I can measure it from a plane" with "I can prove damage from a plane" is the single most common mistake, and it is the one that gets reps making claims they cannot back up.

A quick resolution reality check

The table is the whole argument in miniature: the higher you fly, the more roofs you can scan but the less you can prove about any one of them. Prospecting wants breadth. Inspection wants depth. Use the right altitude for the job.

The three jobs aerial imagery does in a sales pipeline

Aerial data earns its keep in three distinct places. Treat them separately, because the data quality you need is different for each.

1. Targeting — deciding which neighborhoods, blocks, and addresses to work at all.
2. Qualifying — at a specific door, deciding whether this roof is worth a rep's 12 minutes.
3. Measuring — once you have a yes, producing accurate squares, pitch, and waste for an estimate.

Most vendors sell you the third job and let you assume it helps with the first two. It does, but only if you build a workflow around it. Below, each job gets its own treatment.

Job one: targeting — finding the blocks worth working

Targeting is where aerial imagery, combined with two or three other public layers, beats gut feel by a wide margin. The goal is to walk into a canvassing morning with a ranked list instead of a vague map.

The signals that actually predict a due roof

A "due" roof is one near or past the end of its serviceable life, or one a storm just aged out prematurely. From above and from public records, these are the signals that correlate with that, roughly in order of strength:

• Storm exposure history. Did this specific roof sit under damaging hail or high straight-line wind in the last few years? This is the strongest near-term signal because it changes a roof's condition overnight. More on sourcing this below.
• Roof age (as a range). Asphalt shingle roofs in the US commonly carry 3-tab or architectural laminate shingles with field service lives that vary widely with climate, ventilation, and quality. The National Roofing Contractors Association is explicit that a warranty term is not a service life and that real-world longevity depends on installation and conditions. So you are never going to get an exact install date from the sky. What you can do is bracket it.
• Visible wear from imagery. On good ortho imagery you can sometimes see differential weathering, streaking, patched sections, or a roof that has clearly been partially replaced (mismatched color across facets). These are coarse signals but real.
• Roof complexity and material. Cut-up roofs with many valleys and penetrations fail at the details first. Material type changes the sales conversation entirely (a tile or metal roof is a different product and a different buyer than three-tab asphalt).
• Property and ownership context. Owner-occupied versus rental, length of ownership, and home value band all affect close rate and ticket size. None of this is in the imagery, but it joins to the same address.

Notice that none of these, alone, tells you a roof needs replacement. Targeting is about stacking probabilities, not certainty. You are trying to raise your hit rate from, say, 3 in 100 doors to 8 or 10 in 100. That is a doubling or tripling of rep productivity, and it compounds across a season.

Estimating roof age from aerial imagery (and why it is a range, never a date)

This deserves its own discussion because it is the most misunderstood part of the whole field. People hear "roof age from imagery" and imagine the software reads an install date. It does not. Here is what is actually happening and how to use it without lying to yourself or a homeowner.

The honest version of roof-age estimation works off historical imagery and visual condition:

1. Change detection over time. If you have imagery from multiple years for the same parcel, you can sometimes see when a roof changed color or pattern, which often marks a replacement. If the roof looked dark and uniform in a 2014 capture and the same in 2024, you have not learned much. If it visibly changed between two captures, you have bracketed a replacement to that window.
2. Weathering signature. A newer asphalt roof reads darker and more uniform; an older one shows lighter, streaked, granule-thinned facets, especially on south- and west-facing slopes that take more sun. This is a soft signal and varies by shingle color and region.
3. Records cross-reference. County permit records sometimes log a reroof. Coverage is wildly inconsistent by jurisdiction, but where it exists it is gold for bracketing age.

Stack those and you get a range: "this roof is most likely 15 to 22 years old." You never get "this roof was installed on March 4, 2008." Anyone selling you a hard install date from imagery is overselling. Treat the range as a filter input, not a fact you repeat to a customer. The right way to use it on a doorstep is, "From the aerial record it looks like your roof has some age on it, mind if I take a closer look?" not "Our data says your roof is 19 years old."

Building a target list, step by step

Here is a concrete workflow a canvassing manager can run for a given metro. Adapt the tools to whatever stack you have; the logic is what matters.

1. Draw your service polygons. Define the ZIP codes or drive-time radii you actually want to work. Do not skip this; reps waste hours driving between scattered doors.
2. Pull the parcel layer. Get the address points and basic attributes (year built, lot, owner-occupancy where available) for those polygons.
3. Join roof attributes. Bring in roof outline, square footage, facet count, and predominant material/condition where your imagery provider derives it.
4. Overlay storm history. Add the hail and wind exposure layer (covered in the storm section). This is the single highest-value join.
5. Score and rank. Assign weights. A simple, defensible starting model: storm exposure 40%, roof-age range 30%, visible condition 15%, complexity/material fit 10%, owner-occupancy 5%. Tune the weights against your own close data after a few weeks.
6. Cut routes. Slice the top-scoring addresses into geographically tight walking routes of, say, 40-60 doors per rep per shift.
7. Brief the reps. Each rep gets the route and the one-line reason each block scored, so they knock with context, not blind.

The difference between this and the old way is that every door a rep approaches has at least one reason behind it. Reps feel it immediately. Knock rates and morale both climb when canvassers stop feeling like they are wandering.

Job two: qualifying at the door

Targeting gets a rep to the right block. Qualifying happens in the 30 seconds before and during the knock. Aerial imagery helps here in a quieter way: it lets the rep walk up already knowing the roof's shape, slope, story count, and obvious features.

Why does that matter? Three reasons:

• Credibility. A rep who can say, "You've got that lower slope over the garage and the dormer on the front, those valleys are usually the first to go," sounds like a roofer, not a salesperson. Knowing the roof before knocking earns the next two minutes of attention.
• Speed. The rep can pre-judge ticket size and complexity. A simple 20-square gable and a cut-up 45-square hip with three valleys are very different jobs and very different conversations.
• Safety and honesty. The rep is not pretending to have inspected a roof from the sidewalk. They are saying, plainly, that the aerial record and the storm history suggest a closer look is worth it. That framing keeps everyone honest.

A practical move: load the route in a tool that shows each property's top-down image and key roof stats on the rep's phone. When the door opens, the rep already has the picture. They are not fumbling, and they are not guessing.

There is a subtler payoff here too. A rep who walks the block already knowing which homes have the bigger, more complex roofs can self-prioritize within a route. If two doors are no-answers and the next is a 50-square hip with three valleys, that is the door to circle back to. Aerial context lets a rep allocate their own attention toward the homes most worth a second knock, instead of treating every house on the street as equal.

What you must never claim from the sky

This is the guardrail that keeps you out of trouble. From aerial imagery alone, at prospecting altitude, you cannot see hail bruising, granule loss at the mat, soft hits, or any sub-surface condition. So a rep must never tell a homeowner "you have hail damage" or "your roof needs to be replaced" based on imagery. The honest, durable script is condition-neutral: the roof has age and/or storm exposure that warrants an inspection, and the inspection is where damage gets documented. The roofer documents conditions and provides an estimate; the homeowner's insurer is the party that decides coverage; the homeowner owns the claim. Keep your reps inside that lane and the aerial program is an asset, not a liability.

A door script you can actually defend

Here is a script skeleton that uses the aerial context without crossing any line. The point is to give the rep something to say that is both effective and true.

• Opener (context, not claim): "Hi, I'm with [Company]. We've been working roofs in this area because a storm came through a few weeks back and a lot of the roofs on these blocks have some age on them. Mind if I take two minutes to take a look at yours?"
• If asked how you know: "We use aerial and weather records to figure out which streets are worth checking, it tells us a roof is probably worth a closer look, not whether it actually has anything wrong. That's what the inspection is for."
• Setting the inspection: "I'd just get up there, photograph the condition top to bottom, and walk you through exactly what I find, good or bad. No obligation. If your roof is fine, I'll tell you it's fine."
• What the rep never says: any specific age in years, "you have damage," "insurance will pay for this," "this'll be a free roof," or anything about a deductible.

Notice the script gives the rep cover to walk away from a good roof. That honesty is more than ethical; it is what builds the referral reputation that outlasts any single canvassing season. Reps who promise damage at the door get short-term yeses and long-term complaints.

Handling the skeptical homeowner

The sharper homeowners will push back, and your reps should have honest answers ready:

• "How do you know my roof is old?" — "I don't know its exact age. The aerial record suggests it's got some years on it, which is why I'd want to actually look before saying anything."
• "Are you one of those storm chasers?" — "Fair question. We're a local [or licensed in-state] company. I'm not here to file anything on your behalf or promise you a payout, just to document the roof's condition and give you an estimate if you want one."
• "My roof looks fine from down here." — "It might be. A lot of hail and wear doesn't show from the ground. If it's fine when I get up there, that's the best outcome and it costs you nothing to know."

These answers work because they are true. A rep who has been trained to speak in ranges and odds never gets caught in a claim they cannot support, because they never made one.

Job three: measuring for the estimate

Once there is a yes, aerial measurement reports do the heavy lifting most contractors already know. A good report gives you total area, facet-by-facet breakdown, predominant and secondary pitch, ridge and hip and valley and eave and rake linear footage, and a waste table. The value is twofold: you build accurate estimates without sending someone up a ladder for takeoffs, and you standardize the numbers your crews and your suppliers work from.

A few things pros get wrong even here:

• Trusting waste factors blindly. The report's suggested waste is a starting point. Cut-up roofs, certain shingle patterns, and your own crew's habits push real waste up or down. Track your actual waste against report estimates by job type and adjust.
• Ignoring pitch on labor and price. A 12/12 is a different labor day than a 4/12, and steep-slope and access charges are where margin quietly disappears if your estimating template treats all pitches the same.
• Stale imagery on the measurement. If a property changed (an addition, a recent partial reroof) after the imagery capture, your measurement is wrong. Always reconcile against what the inspector sees on site.
• Skipping the on-site verification entirely. Aerial measurement does not replace the physical inspection; it front-loads it. The inspection still confirms condition, decking, ventilation, and the details a plane cannot see.

Measurement is the most mature use of aerial data and the least differentiated. Everyone has it. The edge is in jobs one and two, where most contractors are still flying blind.

How aerial measurement changes your estimating math

It is worth being concrete about where aerial measurement actually saves money, because the savings are not only in time. A manual takeoff on a cut-up roof can take a senior estimator 30 to 60 minutes, plus the risk of a transposed number that turns into a material shortage or an over-order. Pulling a report instead frees that estimator to write more proposals and reduces the variance in your numbers. Standardized squares also let you negotiate supplier pricing against consistent quantities and lower the odds of a job going short mid-tear-off, which is where crews lose a half-day waiting on a delivery.

A simple way to see the payoff: if a report costs you a modest per-property fee and saves 30 minutes of skilled estimating time plus one avoidable material re-order per ten jobs, the report pays for itself on volume alone, before you count the deals an estimator closes with the freed-up hours. Run that math for your own shop with your own labor rates rather than trusting a vendor's pitch deck.

Data quality, freshness, and the questions to ask any vendor

Aerial-driven prospecting lives or dies on data quality, and most contractors never interrogate it. Before you sign with any imagery or targeting vendor, get straight answers to these:

• How old is the imagery, and how often is it refreshed? Capture cadence varies enormously by region. Dense metros get reflown more often than rural counties. After a storm, ask specifically whether they have post-event imagery or are still serving pre-storm captures.
• What is the ground resolution? A vendor quoting "high resolution" without a number is hiding something. Ask for centimeters or inches per pixel.
• Where does the roof-age estimate come from? If they claim age, ask whether it is from historical change detection, weathering analysis, permit records, or a model, and ask for the confidence range. If they cannot give you a range, be skeptical.
• What is the source of the storm layer? Is it ground reports, radar-derived grids, or both? Radar fills gaps but is a model; ground reports are sparse but observed. Good vendors use both and tell you which is driving a given score.
• How is the storm field joined to the roof? Per-parcel intersection is what you want. "This ZIP got hit" dressed up as targeting is not worth paying for.
• Can you see the reasoning per address? You want to know why a door scored high, both to brief reps and to debug the model when hit rates disappoint.
• What is the coverage in your specific service area? National coverage claims often have soft spots. Verify your counties, not the marketing map.

If a vendor gets cagey on freshness or refuses to express age as a range, that tells you how much to trust the rest of the pitch.

The freshness trap, in practice

Staleness causes two distinct failures and it is worth separating them. On the measurement side, stale imagery gives you wrong squares when a property added a section or did a partial reroof since capture, which you catch at the on-site visit. On the targeting side, stale imagery is more insidious: a roof that was replaced 18 months ago after the last capture still reads "aged" in your data, so you waste a knock on a brand-new roof. The fix for targeting is to weight recent captures more heavily and to treat any address where a rep reports a fresh roof as a signal to flag and suppress that parcel. Feed those corrections back; a roof that got replaced is the cleanest negative example your model can learn from.

Putting storms into the picture

For storm-restoration contractors, storm exposure is the dominant targeting signal, and it is also the most abused. Flooding a ZIP code because "there was a storm" is how teams burn a week on roofs that took nothing. The fix is to model exposure per roof, not per region.

Where storm data actually comes from

A few authoritative public sources underpin most storm-exposure data, and it is worth knowing them so you can sanity-check any vendor:

• NOAA Storm Prediction Center storm reports. The SPC compiles preliminary local storm reports including hail and wind, with locations and magnitudes. These are reports, not a continuous field, so coverage is uneven (more reports where more people are).
• NWS Storm Events Database. A longer-term archive of severe weather events maintained by the National Centers for Environmental Information. Good for history, lagged for real time.
• Radar-derived hail estimates (MESH and similar). Radar products estimate maximum expected hail size across a grid. This fills the gaps between ground reports and gives you a continuous map, but it is a model estimate, not a measured hailstone at a given roof.
• IBHS research on hail and wind. The Insurance Institute for Business and Home Safety publishes research on how hail size and impact relate to actual roof damage. Useful for calibrating what hail size actually threatens a given roof type.

The key mental shift: ground reports tell you that it hailed somewhere nearby; radar grids estimate how big across an area; neither tells you what happened to one specific roof. Damage at a given address depends on the hail size and wind speed that hit that roof, the angle, the roof's material and age, and its orientation. A 1.25-inch hail event will do very different things to a 3-year-old impact-rated roof and an 18-year-old worn 3-tab.

Modeling exposure per roof instead of per ZIP

The better approach intersects the storm field with each individual roof and weights by what that roof actually is. Conceptually:

1. Take the estimated hail size and wind speed at the roof's exact coordinates from the gridded model.
2. Adjust for the roof's vulnerability: material, estimated age range, and slope orientation relative to the storm's approach.
3. Produce a per-roof exposure score, not a blanket "this ZIP got hit."

This is the difference between sending a crew down a street where every third roof actually has a story, versus a street where the radar tile was red but the hail core clipped the next subdivision over. The honest framing on this is important: a storm model gives you odds, not proof. It tells you which roofs are more likely worn or damaged, so your inspectors spend their day on the right doors. The inspection still does the proving. Never present a storm model as evidence of damage to a homeowner or anyone else; present it as the reason you prioritized a closer look.

A worked targeting example

Walk through a realistic scenario. A supercell tracks across the north side of a metro on a Thursday evening. By Friday morning you have:

• SPC reports showing 1.5-inch and 1.75-inch hail at three points along the track.
• A radar MESH grid showing a corridor of estimated 1.25 to 2-inch hail about two miles wide.
• Your parcel-plus-roof database for the whole metro.

The lazy play is to dump every rep into the reddest ZIP. The disciplined play:

1. Clip your roof database to the MESH corridor.
2. Within the corridor, rank roofs by estimated hail size at the address times a vulnerability factor (older range, asphalt, and west/south orientation score higher).
3. Pull the top, say, 1,500 addresses and cut them into 25 routes of ~60 doors.
4. Hold back the edges of the corridor where MESH was under ~1 inch, since damage there is far less likely on most roofs; work those only if the core dries up.

You have now turned a vague "storm hit the north side" into 25 routes ranked by probability of a real conversation. Your first-day inspection-set rate is the proof of whether the model is calibrated; track it and tune.

Where RoofPredict fits

Most of what is above can be assembled by a sharp operations person with a GIS background, a measurement vendor, and a few public weather feeds. The reason a product exists in this space is that stitching those layers together per roof, for a whole metro, on the morning after a storm, is a lot of plumbing that most roofing companies do not want to build or maintain.

RoofPredict is built specifically for jobs one and two above: which roofs are due, house by house. It produces a roof-age range per address from aerial imagery and pairs it with storm physics modeled per roof, then ranks doors and routes so crews knock the roofs a storm actually wore out and the roofs simply aging out. The output is a prioritized canvassing list, not a pile of imagery you have to interpret.

What it is not, stated plainly so there is no confusion:

• It is not a lead-buying service. You are not purchasing homeowners who raised their hand. You are getting a ranked view of which existing roofs in your service area warrant attention, so your own reps work smarter.
• It does not prove damage. The roof-age figure is a range, never an install date, and the storm model gives odds, not evidence. Documentation of actual conditions still happens at the physical inspection, by your team.
• It does not touch claims. The roofer documents conditions and provides an estimate; the insurer decides coverage; the homeowner owns the claim. The product's job ends at "here are the doors most worth your time."

Used the way it is meant to be used, it collapses the GIS-plus-weather-plus-imagery assembly into a route your canvassing manager can hand out before coffee. It is most valuable to storm-restoration teams who need to move fast after an event and to retail roofers who want their age-out canvassing to stop being random. It is least valuable to a one-truck operation that closes everything by referral and never canvasses; those folks do not need targeting at all.

The broader point stands with or without any product: the contractors who win the targeting game are the ones who model exposure per roof and keep the prospecting layer honestly separated from the proof layer.

Wiring aerial targeting into your CRM and canvassing app

A ranked list is only useful if it lands in the tool your reps already use. The integration work is unglamorous but it is where adoption is won or lost. A few patterns that hold up:

• Push the score and the reason into the CRM, not the address alone. A rep should see, on the property card, the score band and the one-line reason ("storm corridor, older roof range, west-facing slopes"). A bare list of addresses with no context gets ignored within a week.
• Make per-door outcomes a required field. No-answer, not-interested, inspection-set, not-a-fit, already-replaced. If logging is optional, it will not happen, and without it you cannot tune. Build it into the disposition the rep must select to advance to the next door.
• Close the loop nightly. Feed yesterday's outcomes back so the next day's routes reflect what reps learned, especially fresh-roof flags that should suppress a parcel.
• Respect contact rules. Honor do-not-knock lists, local solicitation ordinances, and any state-specific canvassing registration. Aerial data tells you which doors are worth knocking; it does not override the rules about whether you may knock. Check municipal solicitation permit requirements in each city you work.

The teams that get value treat the data and the CRM as one system with a feedback loop, not as a spreadsheet someone emails out on Monday.

Does any of this apply to commercial roofs?

Mostly the targeting logic transfers, but the economics and the data shift. Commercial low-slope roofs (TPO, EPDM, modified bitumen, built-up) do not weather the way residential asphalt does, so the visual age signature from imagery is weaker. What aerial imagery does well on commercial is footprint and area, which on a large flat roof is a meaningful estimating input, and identifying units, penetrations, and obvious ponding or patched sections from a top-down view.

Storm exposure still matters, but hail behaves differently on a membrane than on shingles, and wind uplift on a large low-slope roof is its own failure mode. For commercial prospecting, the strongest plays are area-based filtering (find the big roofs in your service radius), age cross-reference from any available records, and storm-corridor overlay, with the understanding that the per-roof condition read from the sky is coarser than on a steep-slope asphalt neighborhood. The inspection carries even more of the weight on commercial, and the buyer (a facility manager or property owner) responds to different framing than a homeowner. Treat commercial as a related but distinct workflow, not a copy-paste of the residential canvassing model.

Building the operating rhythm

Data is only as good as the routine it lives inside. Here is a weekly and post-storm rhythm that keeps an aerial-driven sales program sharp.

Normal (non-storm) weeks

• Monday: Operations refreshes the age-out target list for the week's planned territories. Pull the top-scoring blocks where roofs are aging out, weighted by your historical close rate by neighborhood.
• Tuesday-Friday: Reps run ranked routes. Each logs outcome per door: no-answer, not-interested, inspection-set, not-a-fit. This is the feedback loop. Without per-door outcomes, you cannot tune the scoring.
• Friday: Manager reviews hit rate by route and by score band. Are high-scored doors actually setting more inspections? If not, the model needs adjustment or the script does.

Post-storm weeks

• Hour 0-12: Storm passes. Operations pulls SPC reports and the radar hail grid, clips the roof database to the damage corridor, and generates ranked routes.
• Hour 12-48: Reps deploy to the highest-probability corridor doors first. Speed matters in restoration, but precision matters more; the team that knocks 40 right doors beats the team that knocks 200 random ones.
• Day 2-14: Work outward from the core as the high-probability doors get covered. Keep logging outcomes; a storm is a fast, rich calibration event for your model.

The metrics that tell you it is working

Track these and you will know within two or three weeks whether the program earns its cost:

The one number to obsess over is hit rate by score band. If your top-scored doors do not set inspections at a meaningfully higher rate than your bottom-scored doors, your scoring is not working and you should fix the weights before you scale the spend.

What pros get wrong

After watching a lot of teams adopt aerial-driven prospecting, the same mistakes recur. Avoid these and you are ahead of most of your market.

Treating a range like a date. Saying "your roof is 19 years old" when your data supports "roughly 15 to 22 years" is both wrong and risky. Speak in ranges and "worth a closer look."

Confusing storm odds with proof. A red radar tile is not damage. Reps who say "the storm damaged your roof" before anyone has inspected it are writing a problem for the company. Odds get you to the door; the inspection does the proving.

Flooding ZIPs instead of modeling roofs. The whole point of the data is precision. If you are still working entire ZIP codes after a storm, you are paying for a scalpel and using it as a hammer.

Skipping the feedback loop. Teams that buy the data but never log per-door outcomes can never tune their scoring. The model that ships is never the model that is calibrated to your market; your outcomes calibrate it.

Letting reps make insurance or coverage promises. No "free roof," no deductible talk, no "insurance will definitely cover this." The roofer documents and estimates; the insurer decides; the homeowner owns the claim. Aerial data does not change that lane, and crossing it invites trouble with state insurance regulators.

Over-investing in resolution you cannot act on. You do not need centimeter drone imagery to target. You need it to document once you have a yes. Buying drone-grade everything for prospecting is wasted money.

Letting the imagery go stale. A measurement off two-year-old imagery misses a recent addition or partial reroof. Always reconcile aerial numbers against the on-site inspection before the contract.

A starter implementation checklist

If you are standing this up from scratch, here is a sequence that gets you to value without boiling the ocean.

1. Pick one territory. Do not roll out metro-wide. Choose one to three ZIP codes you know.
2. Get your roof attribute layer. Source roof outlines, area, facet count, and material/condition for that territory.
3. Wire in storm history. Set up access to SPC reports and a radar hail grid for your region, or use a provider that has already joined these to addresses.
4. Build a simple score. Start with the five-factor model above. Keep it transparent so you can explain why a door ranked where it did.
5. Cut three routes and run them. Have reps log every door outcome.
6. Review after one week. Compare hit rate across score bands. Adjust weights.
7. Tighten the door script. Make sure every rep speaks in ranges and "worth a look," never in dates or damage claims.
8. Then scale. Expand territory only after the metrics in one area look right. Scaling a broken model just spreads the waste.

The entire program can be live in a couple of weeks. The slow part is not the technology; it is the discipline of logging outcomes and tuning, which is exactly the part most teams skip.

The honest bottom line

Aerial imagery does not find leads. It finds probabilities, and probabilities are enough to roughly triple a canvasser's productivity if you handle them honestly. The roofs that are aging out and the roofs a storm just wore out are knowable from above as ranges and odds, never as certainties. Your reps' inspections turn those odds into documented facts, and your honesty about where the data stops is what keeps the whole operation clean.

Buy the imagery for the half of the job you have been ignoring. Target the right blocks, qualify with context at the door, measure for the estimate, and keep the prospecting layer and the proof layer cleanly separated. Do that, and the same data you already pay for stops being a post-sale formality and starts being the reason your team knocks fewer doors and books more roofs.

FAQ

Can aerial imagery tell me a roof's exact age?

No. Aerial imagery can bracket a roof's age into a range, for example 15 to 22 years, by comparing historical captures, reading weathering signatures, and cross-referencing permit records where they exist. It cannot read an install date. Treat the range as a targeting filter, and on a doorstep speak in terms of the roof having age, not a specific year.

Can I prove hail or storm damage from aerial imagery?

Not from prospecting-altitude satellite or fixed-wing imagery. Those sources show roof shape and gross condition, not hail bruising, granule loss, or soft hits. Damage gets documented during a physical inspection. Storm data and imagery tell you which roofs are more likely affected so you prioritize the right doors; the inspection does the proving.

What is the difference between satellite, fixed-wing, and drone imagery for roofing?

Satellite imagery is lowest resolution (roughly 30 to 50 cm per pixel) and best for broad targeting. Fixed-wing aerial gives 5 to 15 cm per pixel with oblique angles, which is what most paid measurement reports use. Drone imagery is centimeter-level and is the only one detailed enough to document actual damage, but it requires a certified pilot and is for inspection, not mass prospecting.

How does storm data get matched to a specific roof?

Public sources like NOAA Storm Prediction Center reports give point observations of hail and wind, and radar-derived grids estimate hail size across an area. To get per-roof exposure you intersect that storm field with the roof's exact coordinates and weight it by the roof's material, estimated age, and orientation. Ground reports tell you it hailed nearby; the per-roof model estimates what likely reached that specific roof.

Do I still need to inspect a roof if I have aerial data?

Yes, always. Aerial data front-loads the work by giving you measurements, shape, and a probability that a roof is worth attention. The physical inspection confirms actual condition, decking, ventilation, and damage, and produces the documentation behind any estimate. Aerial measurement also goes stale, so on-site reconciliation catches additions or recent partial reroofs the imagery missed.

How accurate are aerial roof measurements for estimating?

Aerial measurement reports from good oblique imagery are accurate enough to build estimates without manual ladder takeoffs, providing area, pitch, and linear footage of ridges, valleys, and eaves. Accuracy degrades if the imagery is outdated or if the roof changed after capture. Suggested waste factors are starting points; track your real waste by job type and adjust your template.

Is using aerial imagery for lead targeting the same as buying leads?

No. Buying leads means purchasing homeowners who have already raised their hand. Aerial targeting ranks the roofs already in your service area by how likely they are to be due, so your own reps canvass smarter. You are improving the productivity of your own sales effort, not paying per contact for someone else's list.

What can my reps legally and ethically say at the door based on aerial data?

Reps should stay condition-neutral: the roof has some age or recent storm exposure that warrants a closer look, and they would like to inspect it. They must not state a specific roof age, claim damage exists, promise insurance coverage, mention deductibles, or offer a free roof. The roofer documents conditions and provides an estimate, the insurer decides coverage, and the homeowner owns the claim.

How do I know if an aerial-driven targeting program is actually working?

Track hit rate by score band: your top-scored doors should set inspections at a meaningfully higher rate than your bottom-scored doors. Also watch doors-knocked-per-inspection-set falling over time and inspection-to-contract rate holding steady. If high scores do not outperform low scores, fix the scoring weights before expanding territory or spend.

How fast should I deploy after a storm, and where?

Move within the first day or two, but lead with precision over volume. Clip your roof database to the radar hail corridor, rank addresses by estimated hail size times a vulnerability factor for age, material, and orientation, and send reps to the highest-probability doors first. Knocking 40 right doors beats knocking 200 random ones, and you work outward as the core gets covered.