Coliving Occupancy Benchmarks 2026: What "Good" Actually Looks Like

Before comparing anything, be clear about which of three measures is on the table, because operators quote whichever flatters them and rarely say which one they used.

Physical occupancy is beds filled divided by beds available, measured on a date. It is the easiest to compute and the easiest to game — pick the right Tuesday in October and a mediocre building looks full. Financial occupancy, sometimes called economic occupancy, is rent actually collected divided by rent that would be collected at full asking rate. It quietly absorbs discounts, incentives, rent-free periods and arrears, which is why it is almost always lower than physical occupancy and almost never the number in the marketing deck.

Bed-night occupancy is the honest one: bed-nights sold divided by bed-nights available across a period. It captures the four days a room sat empty between a move-out and a move-in, which the other two hide entirely. In a product where the average stay is six to nine months rather than twelve, those turnover gaps are the difference between a good year and a flat one.

The spread matters more in coliving than in the sectors it gets benchmarked against. Student accommodation lets on a synchronised 51-week cycle: one move-in weekend, one move-out weekend, and a summer the operator prices separately. Coliving lets rolling, with stays that end in every month of the year, which means turnover gaps are continuous rather than seasonal. A PBSA operator can quote a snapshot and be roughly right. A coliving operator quoting a snapshot is quoting the one measure that structurally hides their biggest leak.

The gap between the three in the same building is routinely five to eight points. So when a benchmark quotes 93% with no definition attached, it is not a benchmark. It is a mood.

There is no listed pure-play coliving operator, which means there is no coliving occupancy figure anywhere that has been through an audit. The most-cited industry number — 93.4% across the global portfolio for 2026 — comes from an annual survey in which participating operators report their own performance. That is not dishonest, and it is genuinely the best voluntary dataset the sector has. But self-selected respondents grading their own homework produce a number with a known direction of error: operators having a bad year are less likely to fill in the survey.

Adjacent sectors do have audited numbers, because listed companies publish them under regulatory obligation and an auditor signs the accounts. Purpose-built student accommodation and UK build-to-rent are the two closest comparables to coliving in operating model: all-inclusive rent, amenity-heavy, professionally managed, short-cycle leasing. Here is how the sector's self-reported figure sits against them.

The shape of that chart is the finding. Coliving's self-reported average sits below professionally managed build-to-rent and below the premium student operator — and it is the only bar on the chart that nobody has audited. If the reporting bias runs the way self-selected surveys usually run, the true figure sits lower still. Every business plan that assumes coliving fills faster than institutional rented product because it is a better experience is arguing against the only evidence available.

The most useful line on that chart is the bottom one. Unite Students guides to 94–96% occupancy for 2026/27. Hello Student, a brand inside the same listed group, guides to at least 87%. Same country, same academic year, same management team, same auditor. Nine points of difference.

The difference is not operational competence. It is position: different price points, different cities, different building ages, different distance from campus. Which is precisely why a single industry average is close to useless as an underwriting input. If a group that files audited accounts runs a nine-point internal spread, the honest version of a sector benchmark is a range, not a point.

Single-operator volatility over time makes the same case. Unite went from 97.5% to 95.2% in one academic year and then guided down again — an operator with two decades of leasing history, in a market where new supply is running roughly half of pre-pandemic levels. Movements of two to three points a year are normal in this product class. If your model has occupancy flat for five years, your model has a bug.

Annual averages hide the months that actually decide the year. In UK city-centre coliving the pattern is consistent enough to plan against: enquiry volume peaks late August through early October as the academic and graduate-job cycles move people, holds through to a secondary January peak, then thins from late May. The soft window is June to mid-August — the months when a room that goes empty tends to stay empty.

That produces a shape most models miss. A building averaging 90% across the year is often running 95% in October and 82% in July. If your debt service is level and your occupancy is not, the annual average tells you nothing about whether you clear your covenants in the third quarter. Model monthly, and stress the trough rather than the mean.

The practical consequence is a leasing rule, not a marketing one: control when your tenancies end. A twelve-month tenancy signed in July ends in July, and you have engineered your own worst month into every subsequent year. Operators who shift break dates and renewal windows toward the strong months buy occupancy points that cost nothing, which is the cheapest form of the lever there is.

Market maturity moves the curve too. In a city where coliving is an established category — where renters search for it by name rather than discovering it — lease-up runs materially faster and the summer trough is shallower, because demand is not purely tied to a relocation event. In a market where you are the category's first serious operator, budget for both a longer lease-up and a deeper trough, and do not benchmark yourself against portfolio averages drawn mostly from mature markets.

Use bed-nights. Not beds, not a snapshot date, not the average of twelve month-end readings.

Occupancy = bed-nights sold ÷ bed-nights available, over the period. Bed-nights available is beds multiplied by days in the period. A ten-bed house across a 30-day month has 300 bed-nights available. If eight beds ran the whole month, one bed was let from the 12th, and one sat empty, you sold 240 + 19 + 0 = 259. That is 86.3% — not the 90% you would report by counting nine occupied beds on the last day of the month.

Three rules keep the number honest. Count a bed as available from the day it is legally lettable, not the day you finish the marketing photos. Count arrears as sold — occupancy measures physical use; collection is a separate line and hiding a collections problem inside occupancy is how operators lose lenders. And never annualise from a strong quarter: report a rolling twelve months alongside the current period, so seasonality is visible instead of averaged away.

Then produce the financial version alongside it. Rent collected ÷ rent at full asking rate, same period. If your physical occupancy is 94% and your financial occupancy is 86%, you do not have an occupancy business — you have a discounting business with good attendance, and the fix is priced, not marketed.

One more discipline worth adopting before a lender asks for it: report your denominator. Two operators can both claim 91% while one excluded three rooms under refurbishment from beds available and the other did not. Stating bed-nights available alongside the percentage makes the claim checkable, and a checkable claim is worth more in a credit committee than a higher unverifiable one.

Occupancy is a ratio with no money in it. A building at 97% because the rent card was cut 15% is losing to a building at 88% at full rate, and the occupancy benchmark will tell you the opposite.

Revenue per available bed — RevPAB — is total room revenue divided by bed-nights available. It moves when occupancy moves and when rate moves, which means it cannot be gamed by discounting your way to a number. It is the same discipline hotels adopted with RevPAR forty years ago, for exactly the same reason: occupancy alone rewarded managers for giving rooms away.

Worked through: a ten-bed house over a 30-day month has 300 bed-nights available. At 94% occupancy on a £780 monthly rate cut to £663 to fill it, room revenue is roughly £6,230 and RevPAB is £20.77. At 88% on the full £780, revenue is roughly £6,860 and RevPAB is £22.87 — ten percent more money from six points less occupancy. The occupancy-only dashboard shows the first building winning. The bank account shows the second.

Track both and look at the pair. Occupancy up and RevPAB up means real demand. Occupancy up and RevPAB flat or down means you bought the occupancy, and you should know the price you paid. Occupancy down and RevPAB up means you traded some volume for rate, which is often the right call in a soft market and almost always looks like failure on an occupancy-only dashboard.

This is also the metric to put in front of investors. Anyone underwriting the asset seriously will discount an occupancy claim they cannot audit; a RevPAB series with the bed-night denominator shown is the version that gets taken at face value.

These are the ranges we use in models, and they are deliberately wider and lower than the ones circulating in the sector.

Lease-up, months 0–6: assume 45–70% average across the period for a first building in a market where you have no existing brand. A single house in a city where you already run three fills faster; a first building in a new city almost never does. The honest planning assumption is that lease-up costs you a full quarter of revenue you will never recover, and the model should carry it as a cost, not a delay.

Year one, post lease-up: 80–90%. The gap between the bottom and the top of that band is almost entirely turnover management — how many days a room sits empty between two tenants — rather than demand. Compressing that gap is the cheapest occupancy you will ever buy, and it is where we would send you next.

Stabilised, year two onward: 88–94% is a good building. Above 94% sustained, check the rent card — you may be underpriced rather than exceptional. Below 85% sustained is a positioning problem, not a marketing problem, and more advertising spend will not fix it.

Underwrite the bottom of each band and treat the top as upside. Every operator we have watched get into trouble did the reverse.

Do not benchmark a single house against a portfolio average. A ten-bed house loses ten points of occupancy when one room turns badly; a thousand-bed portfolio loses one. Variance at small scale is structural, not a performance signal, and reacting to it with discounts is how small operators wreck their rent card.

Do not compare across definitions. If your number is bed-nights and theirs is a snapshot, you are not looking at the same quantity, and the comparison will consistently make you look worse than you are.

And do not accept an occupancy figure — including any figure on this page — without asking who produced it and how. The audited numbers here can be checked against filed results in about two minutes, which is why we used them. The self-reported one cannot be checked at all, which is why we labelled it. That distinction matters more than the numbers themselves.

If your number is below the band you want, the sequence that moves it is renewals first, then turnover-gap compression, then waitlist activation, and price last. We have written that up as a separate practitioner playbook: the coliving occupancy rate playbook covers each lever with the order to attempt them in and an honest account of the ones that did not work for us.

If you are still underwriting rather than operating, model the trough month rather than the annual average, take the bottom of each stage band above, and carry lease-up as a cost rather than a delay. Those three choices remove most of the distance between a coliving model and what the building actually does.