The two books: what Australia’s advertised commercial market can and cannot tell you

Two books, one market

Ask what an Australian commercial asset is worth and you will be answered from one of two records, and almost never told which.

The first is the advertised book: every campaign currently in front of a buyer. It is large, current and public. It is also what most listing-derived market statistics are built from, asking yields, asking $/sqm, months of stock, days on market.

The second is the institutional book: what CBRE, Knight Frank, Colliers, JLL, Cushman & Wakefield and Savills measure each quarter. Cap rates, net face and net effective rents, incentives, vacancy by grade, absorption, supply.

These two books disagree, and everyone who works with both knows it. The usual response is a footnote saying the levels are not directly reconciled. That is true, and it is not enough, because it leaves the reader with no way to judge how far apart they are or in which direction.

So we measured it. We read 1,536 live commercial campaigns and 59 current institutional reports on the same day, and the finding is sharper than we expected.

What the advertised book withholds

Start with what a buyer actually needs. To price an investment asset you need, at minimum, a price and an income. To sanity-check it you want an area, a tenant and a lease term.

Here is how much of the live book gives you each.

Four in five campaigns publish a building area. Fewer than one in four publish an asking price, the rest run as expressions of interest, auctions, or simply "contact agent". Around one in six name a tenant. One in sixteen give a WALE.

And 19 campaigns out of 1,536, 1.24%, publish both a price and an income, which is the only combination from which a yield can honestly be calculated.

Disclosure collapses as value rises

The 1.24% would be a curiosity if it were spread evenly. It is not.

Under $500,000, 61.7% of campaigns publish a price. By $1M–$1.5M it is 45.7%. By $4M–$5M it is 20.0%. Above $5M it is 12.6%.

Above $6M, across the 25 campaigns in the book, it is zero. Not one.

Meanwhile the tenanted share climbs in the opposite direction, from 3.3% under $500k to 26.1% above $5M, and half the campaigns above $15M.

Read those two together and the structure of the advertised market is plain. Small assets are sold to owner-occupiers and private buyers on a published price, often with vacant possession. Large assets are sold to investors on a covenant, through an agent, without a number.

Why this breaks the asking yield

This has a direct consequence for a metric that appears in almost every listing-derived commercial analysis published in Australia, including ours before we checked: the indicative asking yield.

It is usually built one of two ways. Either median asking rent per square metre is divided by median asking price per square metre, two separate medians, drawn from two different sets of properties, which produces a number that belongs to no actual asset. Or it is computed from campaigns that disclose both price and income, which on this book means a sample of 19.

Neither is a cap rate. The first is a ratio of unrelated aggregates. The second is a 1.24% tail, and we now know that tail is not random: it is overwhelmingly the small, cheap, often vacant end of the market, because that is the only part that publishes prices.

So when a regional asking yield reads 8% and a Sydney one reads 4.9%, a large part of that gap is not risk pricing. It is that the regional number is built from sub-$1M freestanding assets with a printed price, and the Sydney number cannot see anything above $6M at all.

Thin markets publish, deep markets do not

The same logic runs across geography, which matters for anyone comparing states.

The Northern Territory publishes a price on 48.0% of campaigns and Tasmania on 40.9%. Western Australia 36.9%, the ACT 34.0%, South Australia 24.4%. The three deepest markets are the three most secretive: New South Wales 19.9%, Victoria 18.0%, Queensland 16.9%.

This is rational behaviour, not obstruction. In a thin market with few buyers, a printed price is how you reach the ones who exist. In a deep market with competitive tension, a price sets a ceiling, so you run an expression of interest and let the buyers set it instead.

But it means a cross-state comparison of asking evidence is partly measuring how willing agents in that state are to print a number. Any regional ranking built on asking data inherits that, and it runs in a consistent direction: thin markets look better documented than deep ones, while being the ones with less reliable comparable evidence behind them.

What the advertised book is genuinely good for

Having spent four sections on the limits, here is the other half. The advertised book is very good at some things, and $/sqm is the best of them, because 81.6% of campaigns publish an area, and area is far harder to be coy about than price.

Reported as quartile bands rather than point estimates, because that is what the evidence supports. Retail asks a median $5,098/sqm on an interquartile range of $3,382–$7,176. Office $5,000 on $3,646–$6,154. Industrial is the cheapest per metre at $3,324 on $2,353–$4,246, which is what you would expect from a sector that buys volume rather than frontage.

The width of those bands is itself the message. An interquartile range that spans roughly two-to-one means "commercial retail in Australia" is not a pricing category. It is a label covering a strip shop, a neighbourhood centre pad and a CBD tenancy, which have almost nothing to do with each other.

Campaign age is the other honest listing metric, with one caveat that must be stated rather than footnoted: this feed is 65 days old and most of these campaigns are still live, so every age below is a floor, not a completed days-on-market.

Even as a floor the ordering is informative, and it matches the disclosure finding almost exactly. The ACT averages 38 days, Tasmania 32, the Northern Territory 33, South Australia 30. New South Wales and Victoria average 19 and 18. Thin markets both publish more and sit longer.

Each sector is sold on a different number

Office is the most likely to publish a price, at 31.2%, and among the least likely to name a tenant, at 8.2%. That is the signature of a vacant-possession-heavy book, consistent with what the occupier data says about backfill space.

Childcare is the mirror image and the most instructive row in the table. Just 1.5% publish a price. But 64.2% name the tenant and 22.4% give a WALE, three times the disclosure rate of any other sector on both counts.

Retail and industrial sit in between, both around a quarter disclosing price. Retail names a tenant nearly three times as often as industrial (23.1% against 8.9%), which is the difference between selling an income stream and selling a shed.

The other book: what actually trades

Everything above is the advertised market. None of it is a cap rate, and no amount of care with listing data will turn it into one. For that you need the second book.

Across 59 current reports, the clearest finding is that for the first time in this cycle the three major sectors are not moving together.

Retail is still compressing. Over the year to Q1 2026, CBRE has sub-regional centre yields 20 basis points tighter, large format 17, regional 16, neighbourhood 6, compression through a hiking cycle, which Colliers attributes to "resilient non-discretionary income" against a 4.35% cash rate.

Industrial has stopped. Colliers has national prime at 5.72%, three basis points tighter over twelve months, which is to say unchanged; underneath it Adelaide moved 17.2 basis points wider over Q2, Perth 12.5, Melbourne 15. Sydney held.

Office is still repricing, and further than is usually quoted. Colliers reports office yields 206 basis points wider than their previous trough. Knight Frank has Sydney prime 150 to 225 basis points off its 2022 tights.

And the widest gap in the market is now inside a single sector. Sydney CBD prime office held at 5.70% for a third consecutive quarter while Melbourne CBD prime softened 13 basis points over the quarter to 7.02%, a record 132 basis point spread between two prime CBD markets in the same country.

Where the houses disagree

The institutional book has its own problem, and it is the mirror of the advertised book’s. It is precise, but it is not singular.

Sydney industrial vacancy in Q2 2026, as published by six houses within weeks of each other: Knight Frank 2.8%, CBRE 3.5%, Cushman & Wakefield 3.7%, Colliers 3.9%, Savills 5.58%, JLL 5.8%.

The highest reading is more than double the lowest. These are not revisions. Each is correct inside its own method, what differs is the stock universe, the size threshold, whether sublease counts, whether speculative stock under construction counts, and where the precinct boundaries fall.

Which means quoting "the vacancy rate" is as unsafe as quoting "the asking yield", for a different reason. The safe reading is grade inside one dataset: Cushman has Sydney CBD office Premium vacancy at 8.9% and A-grade at 16.6%, a 770 basis point spread with no methodological argument available, because it is one house, one quarter, one series.

The number in neither book

There is one figure that decides the economics of most commercial leases in Australia right now, and it appears in neither an asking price nor an asking rent.

Melbourne prime industrial incentives run 21–26% outside the City Fringe. Sydney 17.5–23%. Cushman puts Sydney pre-lease incentives at 20–25%. Against Perth at 8.75–10% and Adelaide at 7.5–12.5%.

A landlord can hold a face rent perfectly steady and move the incentive, and the advertised rent will not blink. Adelaide shows the whole mechanism inside one market: prime face rents grew 3.1% year-on-year to $149/sqm, incentives went from 8.9% to 10.2%, and net effective rental growth came out at 1.5%. Half the headline, in one of the tighter incentive markets in the country.

This is the single largest blind spot in any purely listing-derived view of the market, and it cannot be patched from listing data, because the incentive is precisely the thing that is never advertised.

Sector by sector, both books

Industrial

Advertised: 417 campaigns, 24.5% disclosing a price, a median asking $3,324/sqm, and only 8.9% naming a tenant. Institutional: national prime yields flat at 5.72%, vacancy at or near its peak, national prime rents across $200/sqm for the first time and up 4.7% year-on-year, but with Sydney prime face rents unchanged for seven straight quarters while Brisbane ran 10.4%.

The two books agree that industrial is normalising rather than breaking. They disagree about where, and only the institutional book can tell you that the growth has relocated to Brisbane and Perth.

Retail

Advertised: the largest book at 567 campaigns, 25.2% disclosing a price, the highest median asking rate at $5,098/sqm and the highest tenanted share of the major sectors at 23.1%. Institutional: regional centres at 99%+ occupancy nationally, Sydney regional centre vacancy at 1.1%, no new retail supply completed in Sydney in Q2 2026, and $3.9bn traded in the quarter, up 14%.

Here the books point the same way, which is worth noting because it is not the norm. A deep, actively tenanted advertised book and a tight, well-bid institutional market are consistent readings of the same recovery.

Office

Advertised: 269 campaigns, the highest price-disclosure rate at 31.2% and the lowest tenanted share at 8.2%. Institutional: national CBD vacancy 14.9%, Melbourne 18.9–20.5% depending on the house, Brisbane tightest at 10.2%, and a record Melbourne–Sydney prime yield spread.

The advertised book’s signature here, priced, empty, sitting the longest of any sector at 25 days, is what a market clearing backfill space looks like from the listing side. The institutional book explains why: the flight to quality means Premium runs at 8.9% vacancy while A-grade runs at 16.6%, and the stock being advertised is disproportionately the latter.

Practical implications

• Never accept an asking yield as a cap rate. Ask which fields it was built from. If it is two unmatched medians it belongs to no asset; if it is matched, ask for the sample size, because on a book like this it will be around 1%.
• Treat price disclosure as information, not an obstacle. A published price on a $6M asset is unusual enough to be worth asking about. A withheld price on a $600k asset is equally unusual in the other direction.
• Do not compare asking evidence across states without adjusting for disclosure. NT and Tasmania will always look better documented than NSW and Victoria, and it says nothing about value.
• Use listing data for area, tenure and time, not for yield. $/sqm bands, tenanted share, WALE availability and how long stock has been standing are all measured over the whole book and are robust.
• Get the incentive before you believe a face rent. On the actual lease, not the market range, Brisbane Trade Coast prime incentives rose 350 basis points year-on-year even as rents there grew.
• Use the research houses for direction and grade, not for a single level. Their levels disagree by more than most people assume; their directions rarely do.
• Rebuild net passing income from the lease. Income-weighted WALE, passing against market rent, recoverable against non-recoverable outgoings, then incentives, vacancy allowance, capex and purchaser costs. The lease remains the asset.

Method, definitions and limitations

The advertised book. 1,536 live commercial campaigns held in the sourcing feed, read on 16 September 2026. The feed was reset on 13 July 2026, so the window is 65 days and nothing here reaches further back.

The institutional book. 59 current reports from CBRE, Knight Frank, Colliers, JLL, Cushman & Wakefield and Savills, indexed into roughly 2,000 individual signals, each carrying its metric, sector, geography, grade, period and the sentence it was read from. Only live, non-superseded signals are used. Nothing is modelled, smoothed or interpolated; where two houses disagree, both are reported.

The cash rate was verified directly against the RBA on the date of writing: 4.35%, after three consecutive 25 basis point increases in February, March and May 2026, then held in June and August.

Limitations.

• The listing feed is one source, not a census of Australian commercial campaigns. State comparisons partly reflect how well its saved searches cover each state.
• No achieved-price evidence is presented. This feed records campaigns, not settlements. An asking-versus-achieved comparison across two different populations would not measure a discount and is not attempted.
• Building areas are as disclosed and mix GFA, NLA and agent-reported measures, so $/sqm is a screening band and not a constant-quality index.
• Campaign ages are right-censored by a 65-day window and understate true marketing duration.
• Institutional levels and listing levels are not numerically reconciled anywhere in this report. That they cannot be is the report’s subject, not a defect in it.
• Sector labels on the listing side come from the portal’s own classification and are coarser than the institutional sub-sector taxonomy.

General information only, prepared for a professional audience. Not financial, investment or valuation advice, and not a recommendation to acquire or dispose of any asset. Figures are as reported by the named research houses for the periods stated and may be revised by them. We are not registered valuers.