micromarkets

Micromarkets · Data methodology

How we measure Australia’s property micro-markets

Every figure on a Micromarkets suburb report is computed by a deterministic pipeline — the same recipe for every suburb, every month, with no hand-editing. This page documents each metric with the same four questions: what it tells you, how we calculate it, a worked example, and what’s worth knowing before you lean on it.

Maintained by the Micromarkets Data Team · refreshed with each monthly data cycle

The pipeline at a glance

Everything you see is built in four layers:

Layer	What it holds
1 · Raw records	Individual sale records, rental (lease) records and live sale listings, collected continuously across Australia.
2 · Monthly observations	Once a month, the raw records are rolled up into one row per suburb × property type × bedroom count × side (sales or rental). Each row stores the trailing 12-month figures as at that month — median price, transaction count, median days on market — plus a count of listings available during the month just finished.
3 · Suburb metrics	The monthly observations are combined into the full metric set: growth rates, months of supply, absorption, yield, ownership premium, velocity, percentile ranks and cycle phases — one set per segment.
4 · The report	Comparable-market scoring, nearby-market joins and ABS Census demographics are layered on, and the result is published as the suburb page you read.

The key design choice sits in layer 2: we keep a history of rolling 12-month windows, one per month, rather than a history of single months. Every headline number and every point on the 5-year charts is a 12-month window — which is what makes suburb-level statistics stable enough to trust (single months in a typical suburb are a handful of sales; their medians jump around for no real reason).

How to read our numbers — three conventions

Three conventions apply across the entire site. Once you know them, every label becomes unambiguous:

1 · “12m” means the last 12 complete months

A metric tagged 12mis computed over the 12 finished calendar months ending with the most recent complete month. If you’re reading in June, the window is June last year through May this year — the current, partial month is never included. Each monthly refresh rolls the window forward by one month.

2 · “Growth” and “change” compare two whole windows (year-on-year)

When 12m is paired with growth or change, the comparison is the current 12-month window versus the previous12-month window — months 1–12 versus months 13–24. It is not “today versus the same day last year”, and not one month versus another. Comparing whole windows means the growth figure inherits the same stability as the levels it compares.

Worked exampleA suburb’s 3-bed houses had a median price of $840,000 over the last 12 months, and $800,000 over the 12 months before that. Price growth (YoY) = (840,000 − 800,000) ÷ 800,000 = +5.0%.

3 · “Listings available” is a snapshot, not a 12-month total

Unlike sales and leases (counted across the whole window), listing availability is a point-in-time stock count: the number of properties listed as available during the most recent complete month. It answers “how much is on the market right now?” — so it’s deliberately fresh rather than smoothed, and it’s the numerator of months of supply.

Two smaller habits worth knowing: we use medians, not averages, everywhere a typical value is needed (one $20m sale can’t drag a median); and segments are never blended — every figure belongs to one suburb × property type × bedroom cohort.

Data sources

Micromarkets combines two independent classes of data:

Property market records.Individual sale, rental and listing records collected continuously across Australia, supplemented with official state-government property data (land registry and valuer-general sales records) where state open-data programs make them available. A sale record carries the price, the listing and sale dates (which give days on market) and the property’s type, bedrooms and location; a rental record carries the advertised weekly rent and days to lease.
ABS Census of Population and Housing (2021). Suburb demographics are derived from the official ABS Census DataPacks at Suburbs and Localities (SAL) geography, ASGS Edition 3 — including the G01 (population summary), G02 (medians), G04 (age), G08–G10 (ancestry, birthplace, arrival), G13–G14 (language, religion), G16–G17 (education, income), G29–G41 (families, households, dwellings, tenure, mortgage, rent, bedrooms), G44–G46 (mobility, labour force) and G60–G62 (occupation, commute) tables, plus the ABS Socio-Economic Indexes for Areas (SEIFA) 2021.

ABS cell values are randomly adjusted by the ABS to protect privacy, and small cells are suppressed. We treat suppressed values as missing rather than zero, and flag suburbs whose population is too small for reliable shares (see sample-quality safeguards).

Coverage & segmentation

We publish reports for 15,000+ Australian suburbs across all eight states and territories. Within each suburb, metrics are computed independently per segment — the combination of property type and bedroom cohort:

Property types: houses and units, measured separately end-to-end. Raw listing categories are normalised first — duplexes, terraces and acreage fold into “house”; apartments, townhouses, villas and studios fold into “unit”.
Bedroom cohorts: 1, 2, 3 and 4 bedrooms, each computed from its own records, plus an all-bedrooms view computed from the full cohort (not by averaging the per-bedroom segments).

A 3-bedroom house in a suburb is never blended with that suburb’s units or its 1-bedroom houses — each segment is its own micro-market, with its own medians, growth rates, supply measures and peer rankings.

Sales metrics, one by one

Median price (12m)

What it tells you: The typical sale price in the segment right now.
How we calculate it: Take every sale recorded in the segment over the last 12 complete months, line the prices up from lowest to highest, and take the middle one.
Example: A segment with 9 sales at $700k, $750k, $780k, $800k, $840k, $860k, $900k, $1.1m, $2.4m has a median of $840,000 — the $2.4m outlier doesn’t move it.
Worth knowing: Obvious data errors (a $1 “sale”, a mistyped extra digit) are filtered during the monthly rollup before the median is taken.

Sales volume (12m)

What it tells you: How active the market is — and how much data sits behind every other sales figure.
How we calculate it: Count every sale recorded in the segment over the last 12 complete months.
Example: 120 sales over the window = an average of 10 per month.
Worth knowing: This count is the sample size for the whole sales side — when it drops below the minimums in safeguards, dependent metrics switch off rather than mislead.

Days on market (DOM)

What it tells you: How long a typical property takes to sell.
How we calculate it: For each sale in the last 12 complete months, count the days from the property first being listed to it selling. Line those day-counts up and take the middle one.
Example: A DOM of 30 means the typical listing sold in about a month.

Listings available

What it tells you: How much stock is on the market right now.
How we calculate it: Count the properties listed for sale in the segment during the most recent complete month. This is a snapshot (convention 3) — not a 12-month total.
Example: 25 properties listed for sale last month.
Worth knowing: Deliberately fresh rather than smoothed, so the figure reacts quickly when stock floods in or dries up. It’s the stock side of months of supply.

Price growth (YoY)

What it tells you: Whether typical prices are higher or lower than a year ago, and by how much.
How we calculate it: Take the current 12-month median price and the median for the 12 months before that. Growth = (current − previous) ÷ previous × 100.
Example: $840,000 now vs $800,000 in the prior window → (840 − 800) ÷ 800 = +5.0%.
Worth knowing: Both sides are full 12-month windows (convention 2), so one unusual quarter moves this figure far less than it moves a quarterly index.

Volume growth (YoY)

What it tells you: Whether more or fewer properties are changing hands than a year ago.
How we calculate it: Take the current 12-month sales count and the count for the 12 months before that. Growth = (current − previous) ÷ previous × 100.
Example: 120 sales now vs 150 in the prior window → (120 − 150) ÷ 150 = −20%.
Worth knowing: Read it with the speed metrics: falling volume with falling DOM usually means scarce stock; falling volume with rising DOM usually means cooling demand.

DOM change (YoY)

What it tells you: Whether the market is getting faster or slower.
How we calculate it: Current 12-month median DOM minus the previous window’s median DOM. Shown in days (negative = selling faster than a year ago) and as a percentage of the previous value.
Example: 28 days now vs 35 days before → −7 days (−20%): the typical sale is a week faster.
Worth knowing: This is the “direction of travel” signal the cycle phase uses.

Rental metrics, one by one

The rental side mirrors the sales side metric-for-metric — same windows, same conventions — computed from rental listings and recorded leases instead of sales.

Median weekly rent (12m)

What it tells you: The typical advertised weekly rent in the segment.
How we calculate it: Take every rental listing recorded in the segment over the last 12 complete months, line the advertised weekly rents up from lowest to highest, and take the middle one.
Example: Median rent of $600/week.
Worth knowing: These are advertised (asking) rents. They lead contract rents slightly — responsive at turning points, but see limitations.

Leased volume (12m)

What it tells you: How active the rental market is — and the sample size behind every rental figure.
How we calculate it: Count every recorded lease in the segment over the last 12 complete months.
Example: 180 leases over the window = an average of 15 per month.

Rent growth (YoY)

What it tells you: Whether asking rents are higher or lower than a year ago.
How we calculate it: Take the current 12-month median weekly rent and the median for the 12 months before that. Growth = (current − previous) ÷ previous × 100.
Example: $600/week now vs $570 in the prior window → (600 − 570) ÷ 570 = +5.3%.

Leasing DOM & change (YoY)

What it tells you: How long a typical rental takes to find a tenant, and whether that’s speeding up or slowing down.
How we calculate it: For each lease in the last 12 complete months, count the days from the listing first being advertised to it leasing; take the middle value. The change is the current window’s median minus the previous window’s, in days and as a percentage.
Example: Leasing DOM of 18 days, vs 22 days in the prior window → −4 days (−18%): tenants are snapping properties up faster.
Worth knowing: Leasing markets run much faster than sales markets — days-to-weeks rather than weeks-to-months — which is why every rental threshold on the site (supply bands, velocity calibration) is set lower than its sales twin.

Rentals available

What it tells you: How much rental stock is on the market right now.
How we calculate it: Count the properties listed for rent in the segment during the most recent complete month — a snapshot, exactly like its sales twin.
Example: 12 properties listed for rent last month.

Months of supply & supply bands

What it tells you

How many months the stock currently on the market would take to sell out (or lease out) at the recent pace — the cleanest single read on whether a market is under- or over-supplied.

How we calculate it

Take the last 12 months of sales (or leases) and divide by 12 → the average number sold per month.
Take the number of properties listed as available last month.
Divide the listings by the monthly sales pace. That’s the months of supply.

Example

120 sales in the last 12 months → 10 per month. 25 properties listed last month. Months of supply = 25 ÷ 10 = 2.5 months — the visible stock clears in about ten weeks, which lands in the “tight” band below.

Worth knowing

The demand side uses the full 12-month pace so one busy month can’t fake a shortage, while the stock side uses last month’s snapshot so the figure reacts fast when listings flood in or dry up. The page also shows the year-on-year change — today’s value minus the value computed a year ago — so you can see whether the market is loosening or tightening.

The raw number is then mapped to one of seven plain-English bands. The thresholds are calibrated against the national distribution of all qualifying segments, and they differ between sides because rental stock turns over much faster — 2.5 months of rental supply is loose, while 2.5 months of sale supply is tight:

Band	Sales (months)	Rental (months)
Severe (extreme shortage)	< 1.71	< 1.11
Very tight	1.71 – 2.19	1.11 – 1.30
Tight	2.19 – 2.64	1.30 – 1.47
Balanced	2.64 – 3.58	1.47 – 1.81
Loose	3.58 – 4.53	1.81 – 2.07
Very loose	4.53 – 6.89	2.07 – 2.53
Saturated (extreme oversupply)	≥ 6.89	≥ 2.53

Band labels are only assigned when the segment recorded at least 20 sales (or 20 leases) in the window — below that, the ratio is too noisy to band honestly.

Absorption rate

What it tells you

What share of the stock on the market gets bought (or leased) each month — months of supply read in the opposite direction.

How we calculate it

Take the last 12 months of sales (or leases) and divide by 12 → the average number sold per month.
Divide that monthly pace by the number of properties listed as available last month, × 100.

Example

10 sales a month against 25 listed properties → 10 ÷ 25 = 40% of the stock absorbed per month. (And 1 ÷ 0.40 = 2.5 months of supply — the two metrics always agree; they’re the same ratio flipped.)

Worth knowing

Absorption reads best when stock is scarce: a value over 100% means the market sells more per month than is visibly listed — stock is being cleared as fast as it appears.

Gross rental yield

What it tells you

The rent a typical property earns in a year, as a percentage of what it costs to buy.

How we calculate it

Take the segment’s 12-month median weekly rent and multiply by 52 → a year of rent.
Divide by the same segment’s 12-month median sale price, × 100.

Example

$600/week × 52 = $31,200 a year. Median price $840,000. Yield = 31,200 ÷ 840,000 = 3.7%.

Worth knowing

Rent and price always come from the same segment — 3-bed-house rent over 3-bed-house prices, never mixed. “Gross” means before vacancy, fees, maintenance, rates, insurance and tax: it’s for comparing suburbs, not for budgeting. Requires at least 15 sales and 15 leases in the window.

Ownership premium

What it tells you

How much more (or less) it costs per week to buy the typical property than to rent it.

How we calculate it

Take the segment’s 12-month median price and assume a standardised purchase: 20% deposit, so the loan is 80% of the price.
Work out the weekly repayment on that loan at 6.0% p.a. over 30 years, principal and interest (standard amortisation).
Compare with the segment’s 12-month median weekly rent: premium = (repayment − rent) ÷ rent × 100.

Example

Median price $840,000 → loan of $672,000 → weekly repayment ≈ $929. Median rent $600/week. Premium = (929 − 600) ÷ 600 ≈ +55%: owning the typical property costs about half as much again per week as renting it. A negative premium means buying is cheaper than renting on these terms.

Worth knowing

The 6.0% rate and 80% loan are deliberate constants, not live market rates — they keep the premium comparable across suburbs and across months. Treat it as a relative measure, not a personal affordability quote. Requires at least 15 sales and 15 leases in the window.

weekly_repayment = (price × 0.80) × r × (1 + r)^n ÷ ((1 + r)^n − 1) where r = 0.06 ÷ 52 (weekly rate), n = 30 × 52 payments

Market velocity & percentile ranks

Market velocity

What it tells you: How briskly the market moves — lots of transactions, sold fast, scores high; few transactions, sold slowly, scores low.
How we calculate it: Combine the segment’s 12-month volume and 12-month median DOM into one score: volume pushes the score up with diminishing returns (doubling sales lifts it ~57%, not 100%), and every extra day on market drags it down on a steepening curve. The raw score is meaningless on its own — it exists to be ranked against other segments (next metric).
Example: Two segments each sell 100 properties a year. One has a DOM of 20 days, the other 60 — the 20-day market scores several times higher, because speed at the same volume is what “brisk” means.
Worth knowing: Sales and rental sides have separate velocity indices — the rental one is built from leased volume and leasing DOM with constants tuned to leasing’s faster timescales.

sales velocity (raw) = 10,000 × volume_12m^0.65 × e^(−(DOM+2) ÷ 40) ÷ (DOM + 2)^2.2

Percentile ranks (state & national)

What it tells you: Where the segment stands against its true peers — “outpaces 92% of comparable markets” — without ever comparing houses to units.
How we calculate it: Pool every qualifying segment of the same property type and bedroom cohort (within the state for the state rank; across Australia for the national rank). Rank the metric in that pool and express the position as 0–100.
Example: A sales-velocity percentile of 92 (state) means this segment moves faster than 92% of the same property-type-and-bedroom segments in its state.
Worth knowing: The same construction produces the percentiles for price growth, volume growth, selling speed (ranked on the negative DOM change, so shedding days earns a higher rank), rent growth, leasing volume growth, leasing speed, and the composite “market heat” figures. Velocity and heat percentiles need 15+ transactions in the current window; growth percentiles also need 20+ in the prior-year window.

Demand cycle phases

What it tells you: A one-glance answer to two questions: is demand strong right now, and is it building or fading?
How we calculate it: Cross two signals. Strength: is the segment’s velocity percentile in the top half of its national cohort (≥ 50) or the bottom half? Direction: is days-on-market shrinking year-on-year (getting faster) or growing (slowing)? The four combinations are the four phases below.
Example: Velocity percentile 78 + DOM down 5 days YoY → in demand, growing. Velocity percentile 31 + DOM down 3 days → softer, firming (still quiet, but improving).
Worth knowing: Needs at least 15 transactions in the window. Sales and rental sides are phased independently — a suburb can be tightening for renters while its sales market softens.

Phase	Velocity percentile	DOM trend (YoY)	Reading
In demand, growing¹	≥ 50 (top half)	≤ 0 (faster)	Strong and still accelerating.
In demand, easing	≥ 50	> 0 (slower)	Strong, but momentum is cooling.
Softer, firming	< 50	≤ 0	Below-average demand, improving.
Softer, weakening	< 50	> 0	Below-average and still deteriorating.

¹ On the rental side the strong-and-accelerating phase is called in demand, tightening — same rule, leasing vocabulary.

Similar markets (comparables)

What it tells you

Which markets behave like this one — useful when you’ve been priced out of a suburb and want its statistical twins, wherever they are.

How we calculate it

Pool every other suburb’s segment in the same state, property type and bedroom cohort.
Score each pairing 0–1 per component: 1.0 for an exact match, falling in a straight line to 0 as the gap reaches 100% of this suburb’s value. (Cycle phase scores 1.0 for the identical phase, 0.5 for the adjacent phase on the same side, 0 otherwise.)
Weight and add the components (table below), subtract a distance penalty of up to 0.05 for peers up to 100 km away, and rank. The top 10 are published.

Example

A peer priced within 10% of yours scores 0.9 on the price component; at the sales weights that contributes 0.9 × 0.35 = 0.315 of its total score.

Worth knowing

Computed twice — once for the sales market, once for the rental market — with different weights. Pairs qualify when both sides have at least 15 sales (sales) or 15 leases (rental) plus complete pricing data. Each published peer also shows its price and DOM difference in five plain-English bands (±5% / ±20% price steps; ±3 / ±14 day DOM steps).

Component	Sales weight	Rental weight
Median price / weekly rent	0.35	0.35
Gross yield	0.15	0.20
Days on market / leasing velocity	0.15	0.20
Ownership premium / rent growth	0.15	0.10
Cycle phase match	0.10	0.15
Distance penalty (max)	−0.05	−0.05

The small distance penalty makes proximity a tie-breaker, not a requirement — a behavioural twin 300 km away still ranks if it matches on the fundamentals.

Nearby markets

What it tells you: How the suburb stacks up against its physical neighbours — the markets a buyer or renter would actually cross-shop.
How we calculate it: Measure the straight-line distance between this suburb’s centre point and every other same-state suburb’s centre point. Keep everything within 25 km, rank by distance, and offer 5 / 10 / 15 / 20 km filters.
Example: “Within 10 km” shows every covered suburb whose centre is at most 10 km from this suburb’s centre.
Worth knowing: Comparisons are always like-for-like: the neighbour’s metrics for the same property type and bedroom cohort you’re viewing. Distances are straight-line (great-circle), not driving distance.

The 5-year tape

What it tells you: How the segment’s price, volume, DOM, yield, rent and supply have travelled over the last five years.
How we calculate it: Plot the last 60 monthly observations for each metric. Every point is itself a rolling 12-month value as at that month (layer 2 of the pipeline) — the price line is “the 12-month median, recomputed each month”, not 60 single-month medians.
Example: The May point on the price line is the median of all sales from June last year to May; the June point is the median from July to June; and so on.
Worth knowing: This is why the lines are readable curves instead of sawtooth noise — and why a turning point appears on the tape a few months after it begins on the ground (a rolling window follows the road, a single month follows every pothole). Thin markets can be missing months: each line carries a completeness flag and a count of how many of the 60 months are present. Gaps are flagged, never silently interpolated.

Census & demographic methodology

All demographic statistics are derived from the ABS 2021 Census DataPacks at SAL (Suburbs and Localities) geography. We compute three kinds of values:

Direct medians and counts — population, median age, median weekly household / personal / family income, median weekly rent, median monthly mortgage, average household size — taken from the ABS tables (G01, G02) as published.
Derived shares and ratios — e.g. renter share, owner-occupier share, apartment share, unemployment rate, public-transport commute share, dependency ratios, rent-to-income and mortgage-to-income ratios, and diversity indices for birthplace, language and religion. Each is a transparent ratio of ABS counts, and every denominator excludes “not stated” responses— renter share, for example, is rented dwellings ÷ occupied private dwellings whose tenure was stated — so a suburb with a high non-response rate can’t have its shares silently diluted. Cross-period ratios are unit-normalised first: the mortgage-to-income ratio converts the ABS monthly mortgage median to a weekly figure (× 12 ÷ 52) before dividing by weekly household income, and the ageing index is residents 65+ per 100 children under 15. The three diversity indices (birthplace, language, religion) are Gini–Simpson-style measures — the probability that two randomly chosen residents differ on the attribute, computed from the ABS distribution shares.
Percentile ranks — every derived metric is ranked against all populated Australian suburbs(national percentile) and against suburbs in the same state (state percentile). A “median household income — 87th percentile (national)” badge means the suburb out-earns 87% of comparable suburbs. Percentiles are computed over the reliable-population universe only (suburbs with 50+ usual residents), so near-empty localities don’t distort the distribution.

Commute-mode shares deserve a special note: public-transport and walk/cycle shares are computed over residents who actually travelled to workon Census day — worked-from-home and didn’t-go-to-work are excluded from the denominator, and reported separately. The 2021 Census fell during COVID lockdowns, which inflated worked-from-home figures in some cities.

SEIFA scores (IRSAD, IER and IEO) are republished from the ABS SEIFA 2021 release with their official national and state deciles and percentiles.

Census data is a point-in-time snapshot from August 2021. Fast-growing greenfield suburbs in particular may have changed materially since. Market metrics on the same page are current to the latest monthly refresh — the two sources have different clocks.

Sample-quality safeguards

Suburb-level data fails quietly: a segment with six sales produces a median that looks as confident as one with six hundred. The pipeline’s answer is hard thresholds — below them, figures switch off rather than mislead:

Threshold	What it gates
15 transactions	Derived metrics: yield, ownership premium, velocity and heat percentiles, cycle phases, comparables eligibility — at least 15 on each side the metric touches.
20 transactions	Supply-band labels, and the prior-year window behind growth percentiles. The report’s tables also grey out segments below 20 as “too thin”.
50 residents	Census reliability: suburbs below 50 usual residents are flagged low-reliability and excluded from the census percentile universe — at that size ABS privacy adjustments alone can swing a share by several points.

Sample-quality labels. Each segment carries a sales and rental sample-quality grade derived from its transaction counts, shown alongside the data.
Outlier clipping for charts.Chart axes clip at the 99th percentile of the national distribution (with sensible floors), so a handful of extreme suburbs can’t flatten the scale for everyone else. Off-scale values are flagged, not hidden.
Completeness flags. The 5-year tapes mark metrics whose monthly history has gaps, instead of silently interpolating.

The Read (narrative summaries)

The written summary at the top of each suburb report is generated by a deterministic, rules-based pipeline from the same metrics documented above. The same inputs always produce the same words; every claim in the prose is traceable to a number on the page. Narratives are regenerated from scratch at every monthly refresh and validated against the underlying data before publication — they are never manually edited, and they make no forecasts.

Limitations

Rolling windows lag turning points. The stability of 12-month windows costs recency: a market that turned three months ago shows the turn gradually, not as a cliff. Read the change metrics and the supply snapshot for the leading edge.
Medians move in steps on thin markets. Near the minimum sample thresholds, a single unusual sale can still nudge a median; treat borderline segments with care even where we publish them.
Coverage isn’t a registry. Recorded sales can lag settlement and some off-market transactions are never advertised; volumes are consistent indicators rather than exhaustive counts. Listing snapshots count advertised availability, not every property that would sell if asked.
Advertised rents aren’t contract rents. Asking rents lead the market and can overstate what tenants ultimately sign at, especially in cooling markets.
Gross yield ignores costs; the ownership premium uses a standardised loan, not your loan.
Census demographics date from August 2021 and age accordingly.
Nothing on Micromarkets is financial, investment or property advice. Figures describe markets, not the future of any individual property.

Attribution & licensing

Census and SEIFA source data: Australian Bureau of Statistics — 2021 Census of Population and Housing; Socio-Economic Indexes for Areas (SEIFA) 2021. © Commonwealth of Australia, licensed under CC BY 4.0. Derived statistics (shares, ratios, indices and percentiles) are Micromarkets transformations of that source data.

Where sale records derive from state-government open-data programs (land registry and valuer-general releases), that source data remains © the respective state and is used under the terms of each program’s open-data licence.