Satellite vs Drone vs Street View: Which Imagery Actually Works for Roof Prospecting?

The question isn't "which imagery is best." It's "best for what."

Whatever the imagery source, start where hail fell: the free US hail map has every NOAA report, updated daily.

Scoring 4,000 roofs in a ZIP code and inspecting one roof you're about to quote are completely different jobs. Using the wrong imagery source for either one wastes money — or worse, leaves you with a prospecting method that tops out at 12 roofs a day because it requires a pilot and a flight plan.

Here's the short version: drones win the single-roof close-up. Street view adds ground-level context. Satellite is the only modality that scales to neighborhood-level prospecting. The rest of this post explains why, what each source actually shows AI, and the workflow that stacks all three without overpaying for any of them.

The three imagery sources side by side

Source	Coverage	Cost	Freshness	Resolution	What it's actually for
Satellite	Every roof in the country	Priced per processed area, not per roof	Refreshed periodically; storm overlays compensate for lag	~30–50cm	Prospecting at scale — rank thousands of addresses by condition
Drone	One roof at a time	Per-flight pricing from a provider, or your own labor and licensing	On-demand	Sub-centimeter	Inspection, damage documentation, homeowner-facing reports
Street View	Front elevation only	Free	Often years stale	Ground angle	Supplemental signals — gutters, fascia, visible sag on steep faces

The column that matters most for prospecting is coverage. If you can't score 500 roofs before lunch, it's not a prospecting tool — it's an inspection tool. That single constraint eliminates two of the three sources before the conversation goes any further.

Drone imagery — the closer, not the prospector

Drone-plus-AI tools have gotten genuinely good at a specific job: generating inspection reports that hold up under insurance scrutiny. Measurement reports, hail strikes flagged with GPS coordinates, a PDF your adjuster and the homeowner can both read. That's real value.

What drone imagery can't do is prospect. The constraint is structural: every roof is a separate flight. One ZIP code is maybe 800–1,200 residential parcels, and there is no version of that math that ends in a canvassing list by Friday. Even if the budget existed, the calendar doesn't.

There's also the permission problem. Flying over private property for commercial purposes generally requires the homeowner's consent, and you can't fly 800 addresses in a ZIP without someone calling the FAA. Drone prospecting isn't just expensive — it's structurally impossible at scale.

Verdict: Use drones after the appointment exists, not to create it. If you're using drone footage to build your canvassing list, you're using the wrong tool.

Street view — useful signals, wrong angle

Google Street View (and Bing Streetside, Apple Look Around) gives you something drones and satellite don't: a ground-level human perspective on the property. That angle reveals things overhead imagery misses — fascia rot, soffit staining, gutters pulling away from the fascia, visible patching on steep front faces, general property maintenance signals that correlate with deferred roof replacement.

Those are real signals. A house with sagging gutters, peeling paint on the fascia, and a tarp over one corner of the front slope is a different prospect than a house with identical satellite scoring but immaculate curb appeal. Street view adds that context.

The hard limits:

• Flat and low-slope roofs are invisible. Street view shows you the front elevation. If the roof pitch is under 4:12, you're seeing sky and maybe a few inches of drip edge.
• Image age is a real problem. In suburban and rural areas, Street View imagery is often several years old. A roof that looked marginal in an older capture might have been replaced since. You're prospecting on stale data.
• One elevation only. The back of the house — where a lot of storm damage concentrates on north-facing slopes — is invisible unless there's a rear alley with Street View coverage.

This is where Roofbird handles street view differently than a human with a browser tab. It reads the overhead tile and the ground-level façade together into one combined assessment, and it compares each roof against its immediate neighbours from a wider centred view. If the houses either side have visibly been re-roofed and this one hasn't, that gets flagged, because a street where the neighbours have already gone new is the strongest social proof in the trade. And because street-level imagery is newer than the aerial on roughly 85% of roofs, Roofbird checks every shortlist against the newer picture before you see it. Anything already replaced is moved down and said out loud. Houses that genuinely cannot be judged — behind tree canopy, set too far back from the road — are labelled as exactly that instead of being mixed in with the confident ones. Where the aerial and the street photo disagree, the newer one wins and both capture dates are shown. A roof whose own description says it was recently replaced never reaches the top of a list.

Street view still works as a secondary filter when you're doing it by hand. After satellite scoring surfaces your top 20 addresses in a neighborhood, spending 90 seconds per address in Street View to sanity-check the front elevation is worth doing. Using Street View as your primary prospecting method — manually scanning addresses one at a time — tops out at maybe 30–40 roofs per hour and gives you incomplete data on every single one.

Can AI score a roof from Street View alone? Partially. Front elevation signals (gutter condition, visible patching, fascia state, general property upkeep) are readable by computer vision. But you're missing the overhead view entirely, which means you can't assess field condition, ridge line integrity, valley wear, or any rear-slope damage. Street view AI scoring is a supplemental signal, not a replacement for overhead imagery. For a look at how the tools that combine both cameras stack up, see the comparison of AI vision tools that assess roof condition for prospecting.

Satellite — the only modality that scales to prospecting

Satellite imagery covers every residential address in the country, and it's the only source you can point at a whole ZIP without booking anything. That coverage-to-cost ratio is why satellite is the only imagery source that makes prospecting at scale possible.

At 30–50cm resolution, AI trained on roofing-specific datasets can extract:

• Material classification — asphalt shingle vs. metal vs. tile vs. flat membrane
• Surface degradation patterns — granule loss, oxidation discoloration, visible cracking on flat roofs
• Patch evidence — color discontinuities that indicate prior repair work
• Approximate age banding — newer roofs reflect differently than aged ones; AI can sort roofs into rough age brackets even without permit data
• Storm damage probability — when satellite scoring is overlaid with NOAA hail event data, the model can flag roofs in the damage footprint that show surface anomalies consistent with impact damage

Roofbird uses exactly this stack — satellite imagery scored by AI, overlaid with storm event data, surfaced as a ranked territory heatmap with individual address scores. The output is a prioritized canvassing list: here are the 40 addresses in this ZIP most likely to need a new roof in the next 18 months, sorted by score. You work the list top-down.

Two things changed the game here. First, you no longer have to draw a polygon and hope. Search the area you work — a town, a suburb, a ZIP, or a street — and Roofbird reads the neighbourhoods in view and ranks them before you commit to anything. For each block it reports the housing era, what share of roofs read as weathered, how many homes are actually there (counted from building footprints, not estimated from area), whether hail has hit in the last 12 months, and how walkable the block is. Scattered acreage on long driveways gets called out as bad canvassing even under a perfect storm, because a morning is measured in doors per hour. Click a suggestion and it draws the area for you; you can still drag the corners or draw your own polygon from scratch. When nothing in view clears the bar, it says so plainly rather than offering three neighbourhoods you should skip.

Second, every lead carries two numbers instead of one. NEED is how badly the roof needs replacing — condition and roof age only. NOW is whether anything is forcing a decision on this specific house: visible active damage like a tarp, missing shingles or exposed decking, hail damage on this roof, a purchase in the last 18 months, patching already attempted, or neighbours who have re-roofed. Neighbourhood hail deliberately does NOT raise the NOW score, because every home in the scan shares it and it can't separate one door from the next. It belongs in the pitch, not the ranking. That's why most homes in a settled neighbourhood honestly have no trigger, and Roofbird says so rather than inventing urgency to look busy.

See your territory scored from satellite →

Honest limits of satellite scoring:

• No under-deck condition. Satellite tells you about surface condition. Decking rot, structural issues, and interior leak damage require a physical inspection. Satellite score ≠ inspection report.
• Tree occlusion. Heavy canopy coverage can obscure portions of a roof. AI flags occlusion rather than guessing; you account for it on the door knock.
• Refresh lag. Satellite imagery isn't live. Most commercial providers refresh at intervals that vary by region, and rural areas lag urban ones. Storm event overlays compensate for this — a hail event from last month is flagged even if the imagery predates it — but a roof replaced recently might still show its old score. It's a prospecting tool, not a permit database.

For a full breakdown of what AI reads at satellite resolution and how scoring models are trained, see the full guide to satellite imagery for roofing.

The stack that actually wins in 2026

None of these three sources is a complete system by itself. The roofers who are prospecting most efficiently in 2026 use all three — each for its specific job:

Step 1: Satellite to build the ranked list. Search your target ZIP or county and let Roofbird rank the blocks before you draw anything. Then let satellite AI scoring surface the top 40–60 addresses by condition, age band, and storm event overlap. This is your canvassing list for the week. Time: minutes.

Step 2: Street view to sanity-check the top 20. Before you load up the truck, spend 60–90 seconds per address on your highest-priority targets. You're looking for obvious disqualifiers (brand-new roof, commercial property mislabeled as residential) and confirming the curb-appeal signals align with the satellite score. Cut anything that doesn't pass. Time: 20–30 minutes. Roofbird does this pass for you on every shortlist, which is the version of this step that doesn't eat your morning.

Step 3: Door knock the filtered list. You're showing up with a specific reason — "your roof is in the age range where we're seeing a lot of insurance-eligible wear in this area" — not cold-pitching. The score gives you the conversation opener, and the property record gives you the rest: owner name, whether they live there or it's a rental, estimated market value and confidence, year built, last sale date and price, beds and baths, living square footage, lot size, stories, garage, annual property tax, and mortgage lender. All of that shows free, along with an equity estimate, because equity is what decides whether someone can say yes. Only the DNC-scrubbed phone and email sit behind a one-click unlock. Where a hail event is on record, Roofbird also shows roughly how long the homeowner has left to file an insurance claim, based on the typical notice deadline in that state. Treat that as an estimate and verify with the carrier, not as a legal deadline.

Step 4: Drone to document and close. You've got an interested homeowner. Now fly the roof (or have it flown), generate the inspection report, document the damage with GPS-tagged photos, and hand the homeowner a PDF they can submit to their adjuster. The drone does its actual job: turning an interested prospect into a signed contract.

Each modality does one job. Contractors who try to use satellite imagery to close (it's not granular enough) or drones to prospect (it's not scalable enough) are using the wrong tool for the wrong stage. The stack works because it doesn't ask any single source to do more than it's built for.

For a step-by-step guide to building the ranked list in Step 1, see how to build a roofing canvassing list with AI.

FAQ

Q: Can AI score a roof from Google Street View alone?

It can extract front-elevation signals — gutter condition, fascia state, visible patching on steep faces, property upkeep indicators. What it can't do is assess field condition, ridge integrity, valley wear, rear-slope damage, or flat/low-slope surfaces. Street view AI scoring is a useful secondary signal layered on top of overhead imagery. As a standalone prospecting method, it's incomplete and slow.

Q: Is drone imagery more accurate than satellite for assessing roof condition?

Per-roof, yes. Sub-centimeter drone resolution picks up individual hail strikes, granule loss in specific zones, and flashing gaps that 30–50cm satellite imagery can't resolve. But accuracy isn't the bottleneck in prospecting — coverage and cost are. You don't need sub-centimeter accuracy to decide which neighborhood to canvass next week. You need to score 500 roofs before lunch.

Q: How fresh is the satellite imagery these tools use?

It varies by provider and region. Commercial satellite providers refresh urban areas more frequently than rural ones, and the gap between the two can be substantial. Storm event overlays compensate for refresh lag: a hail event that happened last month is flagged against addresses in the damage footprint regardless of when the imagery was captured. The more useful question is whether the tool tells you the capture date it judged from. Roofbird shows it on every lead, and where the aerial and the street-level photo disagree it shows both dates and trusts the newer one.

Q: Do I need a Part 107 license to use drone inspection software?

The software, no. The flying, yes — if you're operating commercially in the US. Part 107 is the FAA certification required for commercial drone operations. It's a written test, not a flight test, and takes most people 10–20 hours of study. If you're hiring a drone inspection company, they handle licensing. If you're flying yourself for commercial roofing work, you need the cert. This is another reason drones don't scale for prospecting — the regulatory overhead alone makes high-volume flights impractical.

Q: How much does it cost to score a neighborhood?

Scanning new ground and ranking every roof in it is free, at any size, and so is asking questions about roofs already read — searching, sorting, and saving named lists costs nothing and spends nothing. Credits are only spent on acting: reading a roof properly, which buys both photographs from above and from the road AND the property record in one go, is 1 credit per 8 roofs. The owner's name, phone and email is 1 credit per 2 roofs. Opening a house completely is 1 credit. The free trial is 10 credits, no card, no expiry, and after that you buy one-off packs — Starter $49 for 49 credits, Crew $99 for 125, Hunter $199 for 300, Pro $399 for 800. Bought credits never expire and nothing renews. The full breakdown is on the pricing page.

---

Prospect from orbit. Close on the roof. Try Roofbird.

Your territory is already scored. The ranked list is waiting. See how it works →