How accurate are satellite-based solar estimates?

Instant solar estimates feel almost too easy: type an address, get a roof broken into planes with a panel count and an annual energy figure seconds later. If you're relying on that number to make a decision, the fair question is how much to trust it. The short answer is that satellite estimates are remarkably good for a first pass and genuinely wrong in a few predictable situations. Knowing which is which is the whole game.

What the data actually is

Google's Solar API, which powers many instant estimates, is built on two layers. Building Insights returns a structured model of your roof: it splits the roof into segments and reports each one's area, pitch (tilt), and azimuth (direction), then computes a maximum panel count and a modeled annual kWh for the building. Data Layers provides per-pixel annual sunlight, essentially a sun-and-shade heat map showing how much light each patch of roof receives across a year.

Critically, the roof geometry isn't guessed from a flat photo. It's derived from LIDAR and multi-angle aerial imagery, which capture height and slope. That's why the model can report a pitch and an azimuth rather than just an outline, and it's why these estimates are far better than eyeballing a top-down picture.

Where satellite estimates are reliable

On a home with recent, high-resolution coverage and a straightforward roof, the results hold up well. The azimuth is usually dependable because a roof's facing direction rarely changes. Pitch and segment area are typically close enough to plan around. And because the sunlight model reflects real surroundings, the relative comparison between roof planes, which face is best, where shade bites hardest, is trustworthy even when an absolute number is a little off.

This is exactly the kind of information that answers the early questions homeowners have. Is a given plane worth using? Which direction should the panels face? Our guides on whether your roof is good for solar and tilt and azimuth lean on the same geometry the API reports.

Where it falls short

The model is only as current as its imagery, and that's the biggest source of error. A few situations reliably trip it up:

• Imagery age. Aerial data can be one to several years old. If you re-shingled, added a dormer, or removed a tree last spring, the model may not know. A brand-new roof or new construction can be missing or wrong entirely.
• New or changed obstructions. A recently added skylight, vent, satellite dish, or HVAC unit won't appear, and each one removes usable area.
• Trees. Tree canopy is the hardest case. Trees grow, get trimmed, or come down, and dense overhang can both hide roof area and distort the shade model. The sunlight heat map may reflect a canopy that's since changed.
• Complex roofs. Many small facets, curved sections, or unusual shapes get simplified, so panel counts on intricate roofs are rougher.
• Coverage gaps. Some addresses simply have low-quality or no high-resolution data, which lowers confidence across the board.

There's also what the satellite fundamentally can't see. It doesn't know your electricity usage, your rafter spacing or structural condition, your electrical panel capacity, or your jurisdiction's fire setback rules. Those determine whether the theoretical panel count is buildable and how big your system should actually be. To connect a production estimate to your real bill, see estimating solar production in kWh and how to calculate solar offset.

How to use the number well

Treat a satellite estimate as a high-quality hypothesis, not a verdict. It's excellent for deciding whether to pursue solar at all, for comparing roof planes, and for getting an honest ballpark before you talk to anyone. Run your own address through the free instant estimate tool to see the panel count and production figure it produces, then sanity-check it against what you know: Is the roof shape it drew right? Has anything up there changed since the imagery? Are the trees still there?

When it's time to commit, a designer takes over. They confirm the roof plan against as-built drawings or a measurement, apply your local setbacks, verify structure and electrical, and size the array to your usage rather than to the roof's ceiling. That's the difference between an estimate and a stamped design, and it's the work the design studio exists to do. You can see how a confirmed, permit-ready set comes together in the permit anatomy walkthrough. The satellite gets you to the doorstep fast; a human walks you through it.

Educational reference, reviewed 2026-07. A design aid, not a substitute for a licensed electrician or PE. Confirm the enforced NEC edition and local amendments with your AHJ.