Industries / Construction and property / Janitorial and facility services
Routes and schedules
Routes and schedules decide who cleans what, how often and at what hour: how long each task should take, how many cleaners a building needs, which shift they work, and whether the building's own staff or a contractor does the work.
6
Milestones achieved, from 1963 to 2016.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1963
A time for every task
ISSA, the cleaning industry association then called the National Sanitary Supply Association, introduced its first cleaning times booklet in 1963. It gives a standard number of minutes for each task, such as vacuuming 1,000 square feet, so a manager can add up the minutes on a route and set staff hours. Later editions grew to 292 tasks in 1990 and 612 in 2014. In the 1994 edition, vacuuming 1,000 square feet with a 12-inch upright took 22.80 minutes and with a backpack vacuum 6.00 minutes, which pushed crews toward backpacks. APPA, the association of college facility managers, folded ISSA's data into its own staffing tables in its 2012 edition.
Credit
ISSA (National Sanitary Supply Association); John Walker, editor of the 1990s and 2000s editions; Ben Walker, compiler of the 2014 edition
1988
Councils must tender cleaning
Part I of the UK's Local Government Act 1988, titled "competition", listed "cleaning of buildings" among the "defined activities" that local councils had to put out to compulsory competitive tendering (CCT), in which a council's own cleaning team bid against private contractors for the work. The Local Government Act 1999 ended these rules from 2 January 2000 and replaced them with a duty called best value.
LessonIn 20 councils studied for the Equal Opportunities Commission, building cleaning jobs fell 29% in the first round of tendering, from 34,370 to 24,342, and every lost job was part-time, by the count of the Centre for Public Services as reported by the union UNISON in 1995.
legislation.gov.uklegislation.gov.ukeuropean-services-strategy.org.uk
Credit
UK Parliament
1990
Janitors march in Century City
In the 1980s many new Los Angeles office towers switched to non-union cleaning contractors, and janitors' pay fell from about $12 an hour in 1982 to $4.25 by 1986, by the UCLA Labor Center's account. SEIU Local 399's Justice for Janitors campaign bargained for whole districts at once rather than building by building: it signed a master agreement for all downtown buildings in April 1989, then turned to ISS, the Danish-owned contractor that cleaned much of Century City. On 15 June 1990 police attacked a peaceful march there. Video of the attack spread, the city held an inquiry, and the janitors reached a settlement with ISS. SEIU says more than 225,000 janitors in more than 30 US and Canadian cities are now its members.
LessonWhen a building owner can swap one cleaning contractor for a cheaper one, a contract with a single firm protects little, so the campaign organized every contractor in a district.
Credit
SEIU Local 399 and its Justice for Janitors campaign
1992
Five levels of clean
APPA published Custodial Staffing Guidelines for Educational Facilities in 1992, by its own account the first work to tie the floor area one custodian cleans to the level of cleanliness that results. It defines five appearance levels, from Level 1, Orderly Spotlessness, to Level 5, Unkempt Neglect. In the 2012 edition one custodian can keep 12,300 square feet of carpeted office at Level 1 or 116,800 square feet at Level 5. Universities use the scale to staff buildings: in 2024 the University of Illinois found that its staffing across 223 buildings matched "a very high level 5", and a pilot that raised staff in five buildings from 15 to 35 moved Lincoln Hall from a graded Level 5 to Level 3.
Credit
APPA; Alan Bigger, editor of the 2012 edition
2015
Cleaning in daylight
Office cleaning is usually done at night, which keeps lights, elevators and air handling running after tenants leave. 11 Stanwix, an office tower in Pittsburgh, moved its cleaning into the working day so the building could shut down at night. The Pittsburgh 2030 District, a group of building owners that tracks energy use, reported annual savings of 620,000 kWh, a 5.5% cut in energy use, and over $104,000 a year, for an outlay of about $9,000.
LessonThe District's case study names telling tenants and staff in advance as the key to the switch, since cleaners now work around people.
Credit
the owners and managers of 11 Stanwix; the Pittsburgh 2030 District
2016
Janitorial firms must register
California's Property Service Workers Protection Act (Assembly Bill 1978), signed on 15 September 2016, barred anyone from running a janitorial business in the state from 1 January 2018 without registering with the Division of Labor Standards Enforcement, at $500 a year. It covers janitors whether they work as employees, independent contractors or franchisees. It also ordered the state to write a four-hour training on sexual harassment, sexual violence and human trafficking for janitors and their supervisors, in every language spoken at home by at least 2,000 janitors in the state.
Credit
Assemblymember Lorena Gonzalez; California Legislature
Watch for
Schedules set by sensors
ABM, one of the largest US cleaning contractors, says demand-driven cleaning, in which sensors that count people decide when a space gets cleaned instead of a fixed timetable, "can reduce labor costs by 20–40%", and that adopters "consistently report double-digit labor efficiencies". The article cites two articles in IFMA's Facility Management Journal and names no client, building or method.
Achieved whena building owner or an independent researcher publishes cleaning hours and inspection scores for named buildings before and after the switch, compared with buildings that kept fixed schedules.
Claimed by
Angela Culver and Kayla Oliver, ABM, March 2026
Open
01
Open cleaning times, measured
An open table of how long common cleaning tasks take, measured on video in working buildings with the cleaners' consent: vacuuming carpet with each type of machine, mopping, emptying bins, wiping desks, cleaning a toilet. Each task gets a median and a spread across workers and buildings, not one fixed number, with the clips behind it.
The briefCredit · unclaimed02
Fixed versus sensor schedules
A public trial comparing a fixed cleaning schedule with one driven by people counts, on the same floors in alternating periods, logging cleaner hours, inspection scores and occupant complaints, with the data released.
The briefCredit · unclaimed
Open 01 · the brief
Open cleaning times, measured
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open table of how long common cleaning tasks take, measured on video in working buildings with the cleaners' consent: vacuuming carpet with each type of machine, mopping, emptying bins, wiping desks, cleaning a toilet. Each task gets a median and a spread across workers and buildings, not one fixed number, with the clips behind it.
Why it is still open
The usual reference, ISSA's Official Cleaning Times, is sold, and ISSA's own history of it says the times for two vacuuming tasks did not change from the 1999 edition to the 2014 edition. ISSA lists the other common sources of these rates, all from trade bodies: APPA's staffing guidelines, BSCAI's production rate recommendations and BOMA's Cleaning Makes Cents, which give one rate per task as a guideline, not a spread from measured work. Published time-and-motion studies are of hospital rooms, such as one of 40 hours of discharge-room cleaning by ten workers (Cadnum and others, 2026). Web and GitHub searches ("cleaning times", "open dataset janitorial cleaning task times") found no open, measured table.
Built with what exists
Phone or body-worn cameras; an open video-labeling tool such as CVAT; a model that reads video, such as Google's Gemini, to mark where each task starts and stops, checked by a person; floor plans to measure the area cleaned; a public spreadsheet under a CC BY license.
First step
With a contractor's and its cleaners' consent, film one night's route on one office floor, label the tasks by hand, and compare the hand labels with the model's.
Done looks like
At least 30 tasks, each timed at least 50 times across at least ten buildings and 20 consenting cleaners, published with medians and spreads in minutes per 1,000 square feet or per fixture, and set beside ISSA's published times for the same tasks.
Open 02 · the brief
Fixed versus sensor schedules
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public trial comparing a fixed cleaning schedule with one driven by people counts, on the same floors in alternating periods, logging cleaner hours, inspection scores and occupant complaints, with the data released.
Why it is still open
Sellers publish savings without methods. ABM's 20–40% figure (March 2026) names no client, and Butlr, a sensor maker, says demand-based cleaning saves $1,400 per restroom a year without saying how it measured that. The practice is in use: the Norwegian School of Sport Sciences in Oslo sends cleaners where 147 door sensors and feedback buttons show need, and its sensor supplier says the system saved 40.5 cleaning hours in sensor-fitted spaces in October 2020, but that is the supplier's case story, with no comparison. Searches found no university or independent field trial of demand-based cleaning. Occupancy-driven control has been tested this way for lighting and air conditioning, as in a 92-day trial on one office floor in Shenzhen that switched between sensor control and manual control (published 2026), but not for cleaning.
Built with what exists
People counters, either commercial thermal sensors or a Raspberry Pi with a low-resolution time-of-flight sensor; Google's open-source OR-Tools to plan routes from the counts; APPA's five appearance levels or the INSTA 800 standard to score rooms; a shared timesheet.
First step
Put two door counters on one restroom corridor and one block of meeting rooms, log a month of counts, and count how many fixed cleaning visits land on rooms nobody used.
Done looks like
At least 12 weeks on at least two floors in an A-B-A design (fixed, then sensor-driven, then fixed again), with cleaner hours, room scores from an inspector who does not know which schedule is running, and complaint counts, published with the data and with savings given in hours per 1,000 square feet and a confidence interval.