Azimuth vs magnetic north: the compass error that skews solar quotes

By Sam Whitfield · September 23, 2026 · 8 min read
Solar panels installed on a home roof
Photo by Vivint Solar on Unsplash

I was on a roof in Frankston a few years back, compass app open, watching a sales rep's "north-facing" pitch turn into something closer to north-north-west once we actually checked it. Nobody had lied. The rep had eyeballed the roofline off a satellite photo and called it north because it looked north-ish. That gap, maybe 12 degrees, doesn't sound like much until you run it through a production model and watch the projected annual output shift by a few percent. On a system sized to offset a specific bill, a few percent is real money over 20 years.

This is the bit of solar orientation nobody explains properly at the kitchen table: there's a difference between magnetic north (what a compass points to) and true north (what the sun actually cares about), and depending on where you live, that difference can be anywhere from a couple of degrees to well over 15. Get it wrong and every downstream calculation, your quote's payback estimate, your sizing, your east/west split decisions, is quietly built on the wrong number.

Why magnetic north isn't the number you want

Earth's magnetic field doesn't line up with its rotational axis. The magnetic north pole wanders, slowly, year to year, and at any given point on the surface a compass needle points toward it rather than toward true (geographic) north. The angle between the two is called magnetic declination, and it's not a fixed global constant, it varies by location and drifts over time.

In parts of southeastern Australia the declination sits somewhere around 11-13 degrees east, meaning a compass reading of "north" is actually pointing east of true north by that amount. In the US it swings the other way in places, west of true north on the eastern seaboard, east of it out west, with a line of near-zero declination running roughly through the Midwest. Solar irradiance modelling, and the azimuth figure your installer should be using, is calculated against true north, not the number your phone's compass spits out.

The practical effect: a roof that "reads" as due north on a compass might actually be closer to north-north-east or north-north-west in solar terms. For a roof already close to due north the error barely matters, cosine losses near zero azimuth are small. But for roofs already sitting at an angle, say 20-30 degrees off true north, adding another 10-15 degrees of measurement error can push the effective orientation meaningfully further from optimal, and that's where the output estimate starts drifting from reality.

How much output actually rides on this

Orientation loss isn't linear and it isn't dramatic near the peak. A panel facing due north (in the southern hemisphere) at the ideal tilt for your latitude captures close to the maximum available annual irradiance. Rotate that panel 15 degrees off true north and you typically lose somewhere in the low single digits of annual output, sometimes less, depending on latitude and roof pitch. Rotate it 30 degrees off and the loss becomes more noticeable, often in the range of 5-8%, though the exact figure depends on the specific site's tilt, latitude and shading.

None of that is catastrophic on its own. The problem is compounding. If your installer used magnetic north instead of true north to model your quote, and your roof already sits at an angle to true north, the modelled azimuth in the software might be wrong by the full declination value, on top of whatever your roof's actual pitch already costs you. I've seen quotes where the combined error moved the projected output enough to change the payback-year figure by close to a year. That's the kind of number that looks fine on a glossy PDF and then quietly disappoints once the first full year of bills comes in.

If you want a fuller sense of what a payback estimate is actually built on and where it tends to hide assumptions, our piece on what "clear-sky payback years" on a solar quote actually hide is worth reading alongside this one. Orientation error is one more input sitting inside that black box.

Checking it yourself before you sign

You don't need surveying equipment. The simplest fix is to stop relying on a compass app at all and instead pull up satellite imagery, Google Maps or Google Earth both display true north at the top of the frame by default, and align your roof plane against that instead of trusting a magnetic reading. It takes two minutes and it's the same reference point any competent designer should already be using in their modelling software.

If you do want to use a compass, most smartphone compass apps have a setting to toggle between "true north" and "magnetic north" display, buried in settings, easy to miss. Switch it to true north and the app will apply the local declination correction automatically using its GPS position, which solves the problem in one tap.

Worth asking your installer directly, too: "did you model this system's azimuth off true north or magnetic north?" It's a fair question and a good one for separating installers who actually run proper design software from ones eyeballing a satellite photo. Our guide on how to choose a solar installer without getting burned covers other questions worth asking at the same stage, and this one belongs on that list.

Where this interacts with roof shape and split orientation

Declination error matters more on some roofs than others. A simple gable roof with one dominant north-facing plane is fairly forgiving, small orientation errors barely move the needle when the roof is already close to ideal. Where it bites harder is on hip roofs, roofs with multiple smaller planes, or anything already compromised by a less-than-ideal build orientation, the kind of roof where every degree of azimuth is doing real work in the production estimate.

It also matters if you're weighing up a split east/west array instead of a single north-facing one. We've written before about whether east/west solar is actually a bad idea, and the honest answer is "it depends on your load profile," but that comparison only holds up if the east and west azimuths fed into the modelling software are accurate in the first place. If the installer's east-facing string is actually 12 degrees more northerly than modelled, the production curve shifts, and so does the case for or against that layout.

A quick contrarian note on precision

I'll say something that might sound like it contradicts everything above: for most straightforward roofs, chasing the last couple of degrees of azimuth accuracy is a waste of energy. The irradiance loss curve near the top is genuinely flat, the difference between a roof modelled at 355 degrees true north and one modelled at 5 degrees off is negligible in kilowatt-hours. Where I get fussy is specifically when a roof already sits well off true north, 25 degrees or more, because that's where the declination correction stops being a rounding error and starts changing the number on the page. If your roof is close to ideal already, don't lose sleep over a couple of degrees either way. If it's already awkward, get it right.

This is also a reasonable moment to sanity-check the rest of the sizing exercise while you've got the compass out. If you haven't already gone through the full process, our guide on how to size a home solar system without overpaying for panels you won't use walks through the load-matching side of things that orientation feeds into.

What to actually ask for in the paperwork

A properly run solar design isn't guesswork dressed up as software output. Ask your installer for the azimuth figure they used per string or per roof plane, in degrees, and ask whether it was derived from true north. Reputable design tools, the kind that model shading, tilt and azimuth together to produce an annual kWh estimate, all take true north as the reference input; if your installer can't answer the question or looks blank when you ask, that's a fair signal to get a second quote. The Clean Energy Council's installer accreditation standards in Australia and comparable state licensing requirements in the US both assume a level of technical competence that includes getting this right, it's not an exotic ask.

For a US reader, the National Renewable Energy Laboratory's PVWatts calculator is a genuinely useful free tool for cross-checking a quote's production estimate against an independent model, and it explicitly asks for azimuth relative to true south (the northern hemisphere equivalent of our true north question) rather than a compass reading. Running your own address through it before you sign anything costs nothing and takes ten minutes.

None of this is really about being paranoid over a specification sheet. It's about not letting the smallest, cheapest-to-fix error in the whole process, a compass reading, quietly erode a decision you're locking in for the next two decades. Grab the phone, check the settings, toggle to true north, and look at the roof again before anyone signs off on a number.

Tagged

Common questions

How do I find my magnetic declination without an app?
Search for your suburb or town plus "magnetic declination" and a government geomagnetic reference (Geoscience Australia or the US NOAA National Centers for Environmental Information both publish current declination tools) will give you the local value in degrees and direction.
Does declination change over time?
Yes, slowly. The magnetic pole drifts a small amount each year, so declination values are usually quoted with a reference year and a small annual change rate. For a solar quote it's not something you need to track precisely, just get the current value at the time of design.
Is this the same issue as roof tilt?
No, tilt is the vertical angle of the panel off horizontal and azimuth is the horizontal compass direction it faces. Both feed into the same production model, but declination only affects azimuth, not tilt.
Should I walk away from an installer who used magnetic north?
Not necessarily on that alone, but it's a fair prompt to ask more questions about their design process generally. Combine it with the other checks in our installer vetting guide before deciding.
About the author
SW
Sam Whitfield
Solar & batteries · California, US

Sam spent six years installing residential solar before moving to writing full-time, and covers panels, inverters and home batteries.

Former licensed solar installer (C-46, California).

More from Sam Whitfield