How well will you actually live in this house, on this block?
The ILR is a 0–10 star rating of the lived experience a specific floor plan delivers on a specific parcel — winter sun where you spend your days, stepping straight out from living to the backyard, a home that sits naturally on the land's fall. Every star is backed by deterministic, published math. No opinions, no AI guesswork.
One real plan, 32 ways to place it — only one is the best way to live.
We take a real single-storey plan from our builder catalogue — a different one each visit — and rate it at all 16 compass bearings, for both the standard and mirrored plan, on a Melbourne demonstration lot with survey-backed levels. Tap any dot to see that placement's rating. The engine found the winner deterministically — the same inputs always produce the same stars.
- Solar orientation× 0.55
- 100.0
- Setback & transition× 0.25
- 97.6
- Floor levels× 0.20
- 90.8
Mirroring the plan moves the garage — and the living-zone glazing — to the other side of the lot, so the best left-drive and right-drive placements can rate differently. Tap a side to rotate the dial to its best placement.
Rate a real address — right now.
Pick any Australian street address. We resolve the actual cadastral parcel, read the land's contours where our mirror has coverage, and sweep the same demonstration plan across all 32 placements through the real deterministic engine — no sign-up, no AI guesswork.
Three things that decide how a home lives.
Each factor is scored 0–100 and weighted into the composite. The weights are fixed and published: solar orientation 0.55, setback & transition 0.25, floor levels 0.20.
North light where you spend your days
Winter sun in the living room, a kitchen that is bright at breakfast, bedrooms that are not west-facing ovens in February. This is the single biggest driver of how a home feels to live in.
The math: The existing deterministic livability engine scores every room from its glazed walls, window provision and the lot bearing — 0–100, climate-zone aware, unchanged by ILR.
Walk out level from living to the backyard
Kids running straight from the meals area onto the grass. No awkward flight of steps between the couch and the barbecue, and enough separation from the northern fence that winter sun still reaches the yard.
The math: Three subcomponents: transition quality (−12 points per riser, halved by an alfresco platform), northern-boundary separation against idealSeparationM(lat), and private-open-space utility with the northern-neighbour level trade-off.
A home that works with the slope, not against it
On a sloping block, the right split levels feel effortless — and a slab benched into a hill with a retaining wall and a staircase to the lawn never does. On a flat block, a single level is the honest answer.
The math: Earthworks balance (cut ≈ fill scores 100 × (1 − |net|/(cut+fill))²), entry-transition risers (−10 each), split-level expectation round(fall / (7 × 0.19)), and internal accessibility (100 − 6 per riser).
ILR and PILR — honesty about the evidence.
How a home meets the ground decides how it lives on a slope — so the quality of the elevation evidence decides the rating's designation. Ratings built on built-outcome survey or LiDAR floor levels are a full ILR and carry gold stars. Contour-derived (or no) elevation evidence yields a Preliminary ILR (PILR) with silver stars — never silently upgraded, always labelled with its source and confidence.
Elevation sources follow a strict precedence — survey > LiDAR > contours. A missing source degrades confidence; it never fabricates data.
| Elevation source | Designation | Confidence |
|---|---|---|
| Built-outcome survey | ILR (gold) | High |
| LiDAR | ILR (gold) | Medium |
| Contours | PILR (silver) | Preliminary |
| None | PILR (silver) | Low |
Confidence is additionally bounded by factor coverage: if any factor is unassessed, confidence cannot exceed preliminary, and the star band is capped at 8.0.
Every rule is a closed-form function you can check.
The full methodology is published. Here are the formulas doing the heavy lifting.
Northern separation, from winter sun geometry
How far should the living zone sit from the northern fence so winter sun still lands in the yard? It falls straight out of the winter-solstice noon sun angle at your latitude:
winterNoonAltitude(lat) = 90° − |lat| − 23.44° idealSeparationM(lat) = clamp(1.8 m × cot(altitude), 2.0, 6.0)
- Melbourne (−37.8°): altitude 28.8° → ideal ≈ 3.3 m
- Brisbane (−27.5°): ideal ≈ 2.2 m
- Hobart (−42.9°): ideal ≈ 4.1 m
Continuous in latitude — no hand-maintained city table. Below the ideal, the score falls quadratically: the last metre of separation matters most.
Steps, splits and earthworks
Level changes are converted to 190 mm risers and penalised where they hurt daily life, and rewarded where they are the right answer to a slope:
- Living → backyard: −12 points per riser; an alfresco platform halves the penalty of the flight it interrupts.
- Front door: −10 points per entry riser; a porch landing halves it.
- Split-level expectation: on a fall of f metres, the right number of splits is
round(f / (7 × 0.19))— each split ≤ 7 risers. Deviations cost 15 points each. - Earthworks balance: score = 100 × (1 − |cut − fill| ⁄ (cut + fill))² — a balanced pad beats a scarred site. Under 15 m³ total movement is near-flat: 100 outright.
- Accessibility: 100 − 6 per internal riser; zero steps and at-grade entry earns a +5 bonus.
A calibrated 0–100 → star band table.
The composite score maps to 0–10 stars in half-star steps. The bands are deliberately non-linear, the way NatHERS rates energy: a typical well-designed new home lands at 6–7 stars, and every half-star above 8 is hard-won. 10 stars is reserved for the near-perfect outcome on every factor.
When a factor cannot be assessed, its weight is renormalised across the remaining factors, the effective weights are returned with the result, and the rating is capped at 8.0 stars — a partial-coverage rating is never passed off as a full one.
Deterministic and versioned. Every rating carries its methodology version (ilr-v2). No LLM participates in scoring — every rule is a closed-form, reviewable function — so the same home on the same lot always rates the same, and re-rating under a future methodology never overwrites a v1 record.
| Stars | Score | Stars | Score | Stars | Score |
|---|---|---|---|---|---|
| 0 | 0–11 | 3.5 | 47–51 | 7 | 76–79 |
| 0.5 | 12–17 | 4 | 52–55 | 7.5 | 80–83 |
| 1 | 18–23 | 4.5 | 56–59 | 8 | 84–87 |
| 1.5 | 24–29 | 5 | 60–63 | 8.5 | 88–91 |
| 2 | 30–35 | 5.5 | 64–67 | 9 | 92–94 |
| 2.5 | 36–41 | 6 | 68–71 | 9.5 | 95–97 |
| 3 | 42–46 | 6.5 | 72–75 | 10 | 98–100 |
Lower bound inclusive. Bands are pinned by golden tests — any change is a methodology version bump, not an edit.
See how a home will really live.
Rate a home on the map, or put the ILR behind your own product through the API — the same deterministic engine either way.