Nobody has measured how long it takes to clean a room

To price a cleaning job you need to know how long it takes. Not roughly — per room, because that's how people book: two bedrooms, one bathroom, a kitchen.

So the question was simple. Has anyone measured this at scale and published it? We checked 45 candidate sources across five directions: industry norms, UK and EU standards, government time-use surveys, academic time studies, and public procurement.

The answer is no. Not "it's expensive" — it does not exist.

There is exactly one canonical body of data, and it isn't about homes

ISSA Cleaning Times & Tasks — usually called ISSA 612 for its 612 entries — is the industry's reference work. Every workloading system and every academic paper on housekeeping cites it. It costs $50 to ISSA members, $150 in softcover, $250 for the digital edition.

Its scope is "industrial, commercial and educational facilities". There is no bedroom in it. There is no living room. There never has been, in any edition — the series grew 292 → 540 → 612 entries without ever acquiring a home.

What it does have is fixtures and floor rates, and those transfer: a toilet at 4–5 minutes, a shower cubicle at 1.50, sinks at 2–3, taking out rubbish at 0.50, a 14-inch flat mop at 5,355 sq ft an hour.

There's a wrinkle. The previous edition, ISSA 540, is freely available in full — a PDF posted by a distributor, all 540 numbered entries with minutes and square feet per hour. Unofficial provenance, superseded edition, and substantively about 88% of the current work for nothing.

The UK equivalent, BICSc Standards & Best Practice, is the de facto basis for British commercial cleaning tenders. Its price isn't published; it's free to corporate members; its predecessor sold for £30. We couldn't reach the site at all — our network's web filter classifies it as malware — so every BICSc figure we have is secondhand.

What actually closes part of the gap

Three sources, all free, none of them about British homes.

Aguilar-Escobar et al. 2021, JIEM is the only stopwatch breakdown of a bedroom-plus-bathroom into sub-tasks that appears to exist anywhere: 74 measurements, about 960 data points, 13 tasks. Making the bed takes 4.10 minutes — 23% of the entire room. Bedside tables and surfaces 2.45. Sink, mirror and shower glass together 2.37. Mopping the bathroom floor 1.76. Vacuuming 1.55. Toilet and bidet 0.64. The whole room: 17.85 minutes.

It's a hotel turnover, which is not a lived-in home. No occupants' belongings, no kitchen, no living room. But it is the closest thing in the world to a measured bedroom.

INSTA 800 / Erasmus+, "Factors affecting cleaning time" (May 2022) ran 800-plus stopwatch measurements, minimum twelve repeats per combination of soil level and target quality, across four countries and real buildings. It answers a different question, and answers it well: raising the target quality from AC3 to AC5 costs +133% of the time in corridors and +130% in carpeted offices. A trained cleaner works at 0.26 minutes per square metre against 0.39 for an untrained one — +50%.

Commercial rooms, but a standard-to-deep multiplier is dimensionless and it travels. It is also, as far as we found, the only measured value of it anywhere.

RAL Merkblatt 07.2020 gives about 55 rows of room type against square metres per hour. Toilets and shower rooms: 60–90 m²/h. Offices: 160–230. Corridors: 250–350. So a wet room is 2.5–3.5× slower per square metre than a dry one — and BICSc agrees independently (372 vs 93 m²/h, a factor of 4), and so does APSE (300 vs 75, also 4).

Three unconnected sources converging on 3–4× is the single most solid thing we found. It is also, we should be honest, one coefficient.

The dead ends, which are the interesting part

A whole family of standards that look exactly right and contain no minutes. BS EN 13549, the INSTA 800 standard itself (1,737 SEK), NEN 2075 (€110.60), VSR-KMS, CIMS — these measure cleaning quality. Not one of them contains a duration. DIN 77400 (€130.90) specifies hygiene intervals for German schools: zero time figures. REFA, which sells time-study methodology, takes the position that universal m²/h values don't exist and that published reference values "miss significantly" — an organisation whose business is time studies, arguing against the thing we were looking for.

Public tenders: hypothesis tested, hypothesis refuted. It seemed obvious that British public procurement would carry time-and-frequency schedules. We full-text searched NHS NSoHC 2021/22, NHSScotland SHFN 01-02 (98 pages), and council tenders from Wem and Your Housing Group for "minute", "hour" and "m2". Zero hits. They specify frequency — what gets cleaned and how often — never duration. Large facilities contractors do compute labour hours internally against BICSc or ISSA, but those tables never reach the tender document.

Hospital time studies are methodologically excellent and useless to us. Bed frame 166 seconds, bedside table 83 seconds — for fixtures that don't exist in a flat.

Kaggle and GitHub: nothing. A term collision, in fact — in that world "cleaning" means preparing data.

The government surveys are the biggest data in the world and cannot help

This is the part we expected to save us.

UKTUS 2014-15 has 16,550 diaries from 9,388 people. The American ATUS has run about 14,000 diaries a year since 2003 — over 300,000 in total, public domain. MTUS aggregates more than 1.5 million diary-days across 100-plus surveys in about 30 countries.

They all hit the same wall, and it is a hard one. Cleaning your home is one code. UKTUS "3210 Cleaning dwelling" covers the whole house for the whole day. In ATUS, "020101 Interior cleaning" mentions rooms only as examples in the coder's manual, never as data fields. MTUS at its popular 25-category level merges cleaning with laundry entirely.

There is no room-level breakdown anywhere, and there never will be. That's a limitation of the HETUS coding scheme, not of any sample size.

The finding we didn't expect

We checked six platforms that sit on exactly this data every day — Homeaglow, Handy, Housekeep, Jobber, Housecall Pro, ServiceTitan — plus Thumbtack, Angi and TaskRabbit.

Not one of them publishes any aggregate of job duration.

Homeaglow's own blog post about how long cleaning takes cites the American time-use survey. A company that could answer the question from its own database instead quotes a government diary study that, as established above, does not break down by room.

The franchise networks are the same. Molly Maid, Merry Maids, Two Maids, The Cleaning Authority all run operations manuals with minute-level standards. Those are disclosed only in franchise documents, under NDA.

What we concluded

Buy nothing.

The reasoning, in order of how much money each error actually costs:

An error in splitting minutes between rooms is nearly free, as long as the total is right. Our totals are anchored to strong whole-property hours. At a typical basket, a 15% miss in the split moves a couple of pounds on the minority of bookings where it applies. A $150 commercial standard containing no residential rooms does not fix that.

The expensive coefficient is the deep-clean multiplier, because it multiplies the whole estimate rather than one line. A 20–30% miss there is an order of magnitude worse. And its only measured value is free — the +130–133% from the INSTA 800 report.

And our own measurements beat any purchased reference outright. Three fields — actual start, actual finish, which rooms — and a regression against the whole-property anchor. Two to three hundred completed jobs give per-room coefficients accurate to a few minutes; around five hundred gives a stable deep multiplier. That is more accurate than anything published, because it's our cleaners, British flats and our own checklist, rather than a hotel in Seville and an office in Helsinki.

The one thing worth paying for isn't for sale: the transaction data those six platforms already hold. That's a partnership conversation, not a purchase order.

The only figure we're prepared to defend is $0.

---

We did this to price Leestly. If you want the money side of the same question, that's what the market pays and charges.