Docs

How accurate is the estimate?

As of September 15, 2026, we've measured 52 real homes across Denver, Michigan, New York, Indianapolis, Phoenix, Atlanta, and Columbus, over two rounds of testing (September 14 and 15, 2026). 77% of those measurements (40 of 52) came straight from Google's own measured 3-D roof data. Checked against another widely used instant-estimate tool, on the same real addresses, our roof size came out a median of 14.8% apart, with 18 of 35 comparisons (51%) landing within 15% of each other.

Where the number comes from

Every estimate starts with measuring the roof from above — no one visits the house for this step.

When Google has 3-D roof data for a house — the same data behind Google's own solar tools — we use Google's own measured roof shape and slope. That's the strongest measurement we can get from a photo, and it's the most common case.

When Google doesn't have that data for a house, we measure the building's outline from above instead and multiply it by a standard slope factor, so a homeowner still gets a number.

Building outlines now cover every address in the country. When the satellite photo of a home is older than the outline data we have on file, we always price the newer, larger outline — never a smaller one — so a home that's grown since its last photo was taken doesn't get under-measured. If a homeowner has added a second building, they can tap it on the photo to include it in the measurement.

Either way, the price a homeowner sees is always the roofer's own rates multiplied by that roof size. We never make up a price on our own.

The test

Here's exactly how we check our own work:

Results

Test date Metro(s) Homes measured From Google's 3-D data Median difference Within 15%
September 14, 2026 Denver, Michigan, New York, Indianapolis 23 23 of 23 12.3% (14 comparisons) 9 of 14
September 15, 2026 Phoenix, Atlanta, Columbus 29 17 of 29 16.2% (21 comparisons) 9 of 21

On the September 15 run, our newer building-outline data kicked in for 8 of the 29 homes measured that day — all newer-built homes. On those 8, our number landed a median of 11.1% from the other tool's — closer than that day's overall 16.2%.

Here's every home we measured and compared, grouped by metro:

15% line Denver 1 12.8% Denver 2 2.6% Denver 3 10.4% Denver 4 1.4% Denver 5 11.9% Michigan 1 40.1% Michigan 2 42.2% Michigan 3 12.7% Michigan 4 8.1% New York 1 65.6% New York 2 6.5% New York 3 10.9% New York 4 32.5% New York 5 20.9% Phoenix 1 16.1% Phoenix 2 19.1% Phoenix 3 14.8% Phoenix 4 50.7% Phoenix 5 16.9% Phoenix 6 11.2% Phoenix 7 10.9% Phoenix 8 11.0% Phoenix 9 10.8% Atlanta 1 12.6% Atlanta 2 16.2% Atlanta 3 1.5% Atlanta 4 12.0% Atlanta 5 20.7% Atlanta 6 22.3% Columbus 1 20.4% Columbus 2 21.2% Columbus 3 25.7% Columbus 4 54.7% Columbus 5 74.0% Columbus 6 0.1% Percent difference from the other tool's measurement (each bar = one real home)

Why two tools can differ

When one tool counts a detached garage or a second building on the same property and the other tool doesn't, the two numbers will differ — that's a difference in what got counted, not a wrong measurement. The same is true when the satellite photo on file is older than the home itself. Both tools let a homeowner confirm the building on the photo before trusting the price, which is exactly how a mismatch like this gets caught.

Other instant-quote tools

We also checked our roof-size numbers against eRoofQuote, another public instant-quote tool, on the same 30 addresses in three cities. It answered every address, and its number matched ours exactly on 65% of the comparable addresses. Both tools measure from satellite the same way, which is why they land close together. Other public roof-quote tools we looked at either ask you to pass a robot check before showing a price or don't show a homeowner any result without a contractor, so we could only run a direct side-by-side against one of them. Details and the full table: the 9/17 comparison.

What an estimate is, and is not

Every number our tool shows is a ballpark estimate, not a signed price. Measuring from satellite gets closer than guessing from a few questions about a home, but the measured size can still vary a little from what's actually on the roof. The final price is always confirmed at a free in-person inspection, where a roofer can see things a photo can't — like decking damage or extra layers.

Run it yourself

You don't need any of our tools to check this — a browser, a stopwatch, and a home address will do.

  1. Open our instant quote tool and another instant-estimate tool's demo in its own browser tab.
  2. Pick one real home address. Type it into both and pick it from the suggestion list each time.
  3. Start a stopwatch the moment you click "go" on each one, and stop it the instant a price appears.
  4. Note when each tool asks for your name, email, or phone number — before the price, or after.
  5. If a roof size shows up anywhere, write it down so you can compare.

Still stuck? Reply to any email from us.

← How the measurement works Next: Limits & fair use →