Why the same house can get two different NatHERS star ratings

A builder client of mine on a renovation in Ocean Grove had two NatHERS assessments done on the same plans, six weeks apart, for reasons that weren't even suspicious — one was for a bank finance condition, the other because the original assessor went on leave and handed the file over. The results came back 6.4 stars and 6.9 stars. Same house. Same block. Same drawings. Half a star apart on paper, but half a star is often the difference between "just meets code" and "comfortably clears it" in a lot of state variations.

That gap is not a scandal and it's not evidence the whole system is rigged. But it does surprise people who assume a star rating is a lab measurement, like a fuel economy label. It isn't. It's a simulation built from someone's interpretation of drawings, someone's assumptions about construction details that aren't fully specified, and someone's data entry into modelling software. Every one of those steps has room for two competent, honest assessors to land in slightly different places.

The rating is a model, not a measurement

NatHERS doesn't test your actual house. It runs a simulation — currently built on the AccuRate or similar CSIRO-developed engines — using a 3D representation of the building, the construction materials assumed for each element, the window schedule, the shading, and a standardised climate file for your zone. The assessor builds that model from whatever documentation exists: architectural plans, specifications, sometimes a site visit, sometimes not for as-designed ratings.

Where the plans are precise — window sizes, glazing types, wall construction, insulation batts specified by R-value — there's not much room to diverge. Where they're vague, which is often, the assessor has to make a judgement call. Is that a single or double brick veneer wall if the drawing just says "brick veneer"? What's the actual roof colour if the spec says "colour to be selected"? Does the pergola on the north elevation count as permanent shading or decorative? Two assessors reading the same ambiguous drawing can reasonably tick different boxes, and each box shifts thermal performance a little.

Software version and default assumptions matter more than people think

NatHERS software gets updated periodically, and assessors don't always upgrade on the same day a new build is released. A model built in an older software version with slightly different default assumptions for infiltration rates or window frame conductivity can produce a different star outcome than the identical drawing set run in the current build. This is one of the less talked-about reasons a rating done in 2021 and one done today, for architecturally unchanged plans, sometimes don't match — separate from the point about ratings drifting over time that we've covered in why your NatHERS star rating can change without touching the house.

Default infiltration values are a particular sore point. Unless a blower door test result is supplied, the software assumes a standard air leakage rate for the construction type. That assumed number can be generous or conservative depending on the software version and the assessor's settings, and it has a real effect on the heating and cooling loads the model predicts. This is exactly the kind of gap covered in air sealing before insulation — the model doesn't know how tight your house actually is unless someone measures it.

What a second opinion actually catches

I've reviewed enough assessor reports side by side now to have a rough sense of where the divergence usually comes from, and it's rarely the big-ticket items. Roof and wall R-values from a spec sheet are hard to get wrong. Window U-values and SHGC from a supplier's NFRC or WERS-rated product data are also fairly locked in once entered.

The scatter shows up in three places consistently. First, shading — trees, neighbouring buildings, eaves depth, and whether a nearby structure is modelled as permanent or was left out because it wasn't obviously on the survey. Second, zoning — how the assessor splits the house into thermal zones affects how conditioning loads get distributed and can move the result by a few tenths of a star. Third, construction defaults for anything not explicitly specified on the drawings, which sends the assessor back to the software's built-in assumptions rather than the actual house.

None of this means get a rating, disagree with it, then go assessor-shopping until you land on the number you want. That's exactly the kind of behaviour the accreditation bodies are trying to police, and it undermines the credibility of the whole scheme for everyone. If you think a rating is wrong, the right move is to ask your current assessor to walk through their assumptions on the specific elements you're querying, not to quietly commission a rematch.

Why this matters more at the compliance edge

If your design is sitting at 8.5 stars in a state where the minimum is 6 or 7, a half-star swing between assessors is academic. Nobody cares. But if your design is sitting at 6.1 in a jurisdiction where 6 is the floor, that half-star gap between two reasonable interpretations is the difference between a compliant house and a redesign. I've seen this exact scenario play out on a project near Hastings Lane in Palm Beach, where the first assessment came in just under the line on shading assumptions for a boundary fence that hadn't been built yet, and a second look — with the fence confirmed and modelled as permanent — pushed it over.

My honestly held view here: assessors working close to a compliance threshold should be more conservative with assumptions, not less, because the cost of a false pass (a house that doesn't perform as modelled) falls entirely on the homeowner years later, while the cost of a conservative call falls on the builder now, when it's cheap to fix. Not every assessor works that way, and I don't think that's dishonesty. It's just a different risk tolerance applied to the same ambiguous drawing.

What you can control as the client

The single biggest lever you have isn't the assessor, it's the documentation you hand them. A drawing set with every window schedule item filled in, glazing specified by actual U-value and SHGC rather than "double glazed, TBC", insulation R-values called out per element, and shading structures shown with dimensions gives the assessor almost nothing to guess at. That's the difference between a rating that's a reasonable estimate and one that's basically locked to the built outcome.

If you're building near the compliance minimum in your state, it's worth commissioning a rating early in design development, then again once the drawings are near-final, rather than once at the end. That first pass tells you which elements are moving the needle, which lines up with the point made in why a NatHERS or HERS rating won't tell you what to fix first — the star number alone doesn't tell you where the money's best spent, but comparing two runs against different assumptions will.

For anyone comparing the Australian system against the American HERS Index while researching materials for a build with both jurisdictions in the family (I get this question a fair bit from readers with property either side of the Pacific), the mechanics are similar enough that the same divergence logic applies — a HERS rater working from incomplete plans faces the identical judgement calls, just expressed as an index number instead of stars. We've laid out the structural differences in NatHERS vs. HERS: how Australian and US home energy ratings compare.

Site checks reduce the gap, on-paper ratings widen it

An as-built rating with an actual site visit closes most of this gap because the assessor is checking installed insulation, actual window products, and real shading rather than working from a drawing and hoping construction matches it. We cover what that visit actually involves in what a NatHERS or HERS assessor actually checks on-site, and it's a genuinely different exercise to an as-designed desktop rating. If your rating matters for finance, insurance, or a rebate application under a scheme like the Solar Homes Program in Victoria or the ACT's Sustainable Household Scheme, ask specifically whether it's desktop or site-verified, because the answer changes how much confidence you can put in the number.

The Clean Energy Council's guidance on assessor accreditation, and the NatHERS administrative body's own technical notes, both acknowledge that software updates and assessor judgement introduce a margin of variation — it's built into how the scheme is designed to work, not a flaw nobody will admit to. The US Department of Energy's guidance on home energy assessments makes a similar point about HERS raters and inspection consistency.

None of this is a reason to distrust the rating scheme generally. It's a reason to treat any single star number as an estimate with a margin either side, tighter when documentation is precise and a site visit backs it up, looser when it's a desktop model built from a half-finished drawing set. Ask your assessor what they assumed on shading, infiltration and any element the plans left vague, and you'll usually understand exactly where a second rating might land differently before you ever commission one.