Industries / Construction and property / Janitorial and facility services / Facility management
Help desk
The help desk is how occupants report a problem, by phone, email, form or app, and how each request is sorted, sent to the right trade, fixed and closed.
6
Milestones achieved, from 1991 to 2026.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1991
Too hot, too cold
The International Facility Management Association (IFMA) first surveyed its members on the top office complaints in 1991, when "too hot" ranked first and "too cold" second; the two swapped places in 1997, and "too cold" was still first in 2003. In IFMA's 2009 study Temperature Wars, based on 473 members, 94% said they received "too cold" complaints and 91% "too hot", with indoor air quality a distant third at 25%. Asked how they respond, 90% said they check the temperature where the complaint came from, 87% check that the HVAC is working and 75% adjust thermostats.
Credit
International Facility Management Association
1998
Complaints traced to controls
Clifford Federspiel of Johnson Controls examined unsolicited complaints from 23,500 occupants in 690 commercial buildings. Too hot and too cold were the most common complaint of any kind, and his analysis found that they came mostly from poor control and from faults in the heating and cooling, not from people preferring different temperatures. He estimated that cutting such complaints could save about 20% of the labor cost of HVAC maintenance. A companion paper gave a model that predicts, from how closely a building's controls hold temperature, how often hot and cold calls will come in and what answering them costs.
LessonA hot or cold call usually points to the controls or the equipment, not the person, so the call log can be read as a record of faults.
Credit
Clifford Federspiel, Johnson Controls
2006
Few buildings please most
The Center for the Built Environment at the University of California, Berkeley, runs a standard online survey of office occupants. Its analysis of 34,169 responses from 215 buildings in the US, Canada and Finland found that 80% or more of occupants were satisfied with the temperature in only 11% of the buildings, and with the air quality in 26%; across all occupants, more were dissatisfied with temperature (42%) than satisfied (39%). A 2005 study by Wang, Federspiel and Arens found that buildings with lower survey satisfaction logged more unsolicited complaints in their maintenance databases.
Credit
Charlie Huizenga, Sahar Abbaszadeh, Leah Zagreus and Edward Arens, Center for the Built Environment; D. Wang, Clifford Federspiel and Edward Arens
2015
Set for an average man
Boris Kingma and Wouter van Marken Lichtenbelt of Maastricht University Medical Center pointed out that indoor climate rules rest on a comfort model developed in the 1960s whose standard value for metabolic rate, the heat a body gives off, is based on an average man and may overstate a woman's by up to 35%. They argued for using measured metabolic rates instead.
LessonSome "too cold" calls come from the setting itself, not from a fault in the equipment.
Credit
Boris Kingma and Wouter van Marken Lichtenbelt
2021
Repair records faked for bonuses
Balfour Beatty Communities, one of the largest landlords of privatized US military housing, pleaded guilty to major fraud and agreed to pay $65 million, $33.6 million in criminal fines and over $31.8 million in restitution, plus three years of probation and an outside compliance monitor. The Justice Department said its employees altered data in property management software and destroyed or falsified resident comment cards so the company would meet the metrics that earned performance bonuses; a 2019 Reuters and CBS investigation had found two sets of maintenance records at some bases. GAO had told Congress that a common yardstick for these landlords was how quickly they responded to a work order, not whether the problem was fixed. In December 2019 Congress gave military tenants a legal right to an electronic work-order system in which they can request repairs and track the work.
LessonA response-time figure that the contractor records itself can be gamed; check the repair, not the timestamp.
airandspaceforces.comwebarchive.library.unt.edulaw.cornell.edu
Credit
US Department of Justice; Air Force Office of Special Investigations; Reuters and CBS News; US Congress
2026
Closed before the fix
GAO analyzed a year of maintenance work orders from all four US military services. All require emergencies to be fixed within 24 hours, and the Army, Navy and Marine Corps allow 5 to 7 days for urgent work and 30 days for routine work. The data showed 68% to 89% of emergency orders closed within a day, but GAO judged the data too unreliable to say how much work was done on time: Air Force records held about 2.4 million work orders with missing dates, and 17 of the 37 customer organizations GAO interviewed said orders were sometimes closed before the work was finished. The same report put the Defense Department's maintenance backlog at an estimated $285 billion in fiscal year 2025.
Credit
US Government Accountability Office
Watch for
A voice agent takes calls
Facilio, which sells maintenance software, says its help-desk agent answers a UAE facilities firm's phone lines in Arabic and English, checks each request, creates the work order and passes only unclear cases to a person. It reports that the agent handled 1,110 of 1,387 calls from January 2026 end to end (80%), took about 30% of calls outside office hours and cut the time from call to dispatch from 3.6 hours to 0.9 hours.
Achieved whenthe facilities firm or an auditor publishes a sample of calls with the agent's work orders, including requests it mis-sorted, set wrong priorities for or failed to log.
Claimed by
Facilio, in an undated case study online in November 2026
Open
01
Comfort calls with zone data
An open, anonymized data set of at least 5,000 "too hot" and "too cold" work orders from at least ten buildings in at least three climates, each paired with the room's measured temperature and its setpoint for the two hours before the call. Add a first analysis: what share of calls came while the room was inside its setpoint range, and what share traced to a fault.
The briefCredit · unclaimed02
Recheck closed work orders
A tested method, and a first result, for checking closed repairs in the field: draw a random sample of 200 work orders closed in the past 30 days at a consenting campus or housing estate, revisit each, record whether the fault is fixed, partly fixed or not fixed, and compare the recorded close date with what the occupant says.
The briefCredit · unclaimed
Open 01 · the brief
Comfort calls with zone data
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open, anonymized data set of at least 5,000 "too hot" and "too cold" work orders from at least ten buildings in at least three climates, each paired with the room's measured temperature and its setpoint for the two hours before the call. Add a first analysis: what share of calls came while the room was inside its setpoint range, and what share traced to a fault.
Why it is still open
Federspiel examined complaints from 690 buildings in 1998, but only an abstract is on record and a search found no release of the data. The ASHRAE Global Thermal Comfort Database II, led by Berkeley's Center for the Built Environment and the University of Sydney, pairs 109,033 questionnaire answers with measurements taken in the same room at the same time, but they are answers to surveys, not calls to a help desk. Gunay, Bucking and Dutta (2020) mined four years of work orders from four office buildings in Ottawa, found "too hot" calls peaking at 2 p.m. when indoor temperatures peaked, and wrote that the work must be extended to larger portfolios in several climate zones. Pampana and colleagues (2024) released university work orders with weather data but no room readings. A search found no open data set that pairs complaint tickets with room sensor data.
Built with what exists
Work-order exports from a campus maintenance system; building automation trend logs exported as CSV; the Brick schema to name sensor points the same way across buildings; Python with pandas; a low-cost text model to sort free-text calls into hot, cold and other.
First step
Ask one campus facilities office for a year of hot and cold work orders and the matching room trend logs for two buildings.
Done looks like
The paired data set published on Zenodo with at least 5,000 calls from at least ten buildings, and a short report giving the in-range share with confidence intervals for each building.
Open 02 · the brief
Recheck closed work orders
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A tested method, and a first result, for checking closed repairs in the field: draw a random sample of 200 work orders closed in the past 30 days at a consenting campus or housing estate, revisit each, record whether the fault is fixed, partly fixed or not fixed, and compare the recorded close date with what the occupant says.
Why it is still open
In August 2026 GAO found US military work-order data too unreliable to measure on-time repair, and 17 of 37 customer organizations it interviewed said orders were sometimes closed before the work was done. GAO had earlier reported that oversight of privatized military housing measured how fast a landlord responded, not whether the problem was fixed. A search found no published field study that measures the share of closed work orders whose repair was actually complete. Scotland makes social landlords report the share of reactive repairs completed right first time. When Edinburgh's internal auditors tested the council's figure, 71% for 2022/23, they found in a report of 30 January 2024 that recalls were not recorded well enough to report it accurately, and that 1,050 of 36,994 repair lines had no completion date or one earlier than the issue date.
Built with what exists
A random sampler over the maintenance system's export in Python; KoboToolbox forms on a phone for revisit records and photos; a second checker repeating 10% of revisits. Revisits are visual checks only: no opening of panels or equipment.
First step
Agree access with one campus or housing provider and pilot 20 revisits.
Done looks like
200 closed orders revisited, the shares fixed, partly fixed and not fixed reported by trade and priority, the agreement between the two checkers on the repeated 10%, and the method and anonymized results published.