At some point the figure lands on the table. A tender coming in a third below what your own cleaning costs. Or the reverse: a model calculation from the administration showing in-house cleaning to be cheaper than the current contract. And because both cannot be true at once, the argument begins.
We are regularly asked to review such calculations or to build them afresh. What stands out is almost never an arithmetical error. It is the items that never made it into the table at all, and for each of the two options they sit somewhere different.
Why the simple comparison does not work
A cleaning contract is an invoice with a figure at the bottom. In-house cleaning is a bundle of staff costs, material costs, administrative effort and risks, spread across several budget lines and appearing nowhere as a total.
Anyone wanting to compare the two has to construct the second side first. Four blocks belong in it.
Staff costs. Pay including premiums and allowances, the employer's social security contribution, occupational pension provision, the annual bonus payment, overtime, capital-forming benefits. Plus, often forgotten, the cost of cover for leave and sickness, and initial and continuing training.
Material costs. Cleaning products and consumables, cleaning equipment and machines, workwear and protective equipment, laundering of textiles.
Operational overhead. Management and administrative staff, accounting, controlling, building supervision and inspection, purchasing, travel costs, premises and storage costs, insurance, depreciation.
Internal cost allocation. What other departments do for the cleaning function: HR, the treasury, the municipal cash office, the audit office, IT, occupational health, the staff council.
The fourth block is missing almost every time, and a cost accounting system capable of showing it at all does not exist in many local authorities.
Conversely, with contracting out, what remains in-house by way of contract management is regularly underestimated. A contract does not look after itself. Acceptances, complaints, variations, adjustments when areas change, and eventually the re-tender. Enter that as zero and you have made the same mistake in the other direction.
The pay question: closer together than everyone assumes
It is worth looking closely here, because the widespread assumption does not hold.
In the building cleaning trades, wage group 1, covering routine and interior cleaning, has carried an hourly rate of 15.00 euros since 1 January 2026. This minimum wage has been declared generally binding and applies to every business in the sector regardless of membership of an employers' association. The collective agreement between the trade union IG BAU and the Federal Guild Association runs until 31 December 2026.
On the municipal side there is the TVöD, the collective agreement for the German public sector. Cleaning duties fall there into pay group 1 or 2 depending on how they are cut; what is decisive is the work actually assigned rather than a blanket allocation of all staff to one group. Once somebody looks after a round on their own responsibility, coordinates with users and has been trained beyond a brief induction, pay group 2 regularly applies.
And here it comes down to the increment, which is precisely where most model calculations fail. Enter the starting rate at increment 1 and you are costing a team you do not have. After six years an employee stands at increment 4, after fifteen at the top, and an established municipal cleaning workforce consists predominantly of such employees.
The table values for pay group 2 in the municipal sector have stood since 1 May 2026 at 3,027.12 euros at increment 3 and 3,101.04 euros at increment 4, and at 3,433.49 euros at the top increment. At 39 hours a week, that is around 169.5 hours a month, this corresponds to 17.86 euros at increment 3, 18.30 euros at increment 4 and 20.26 euros at the top.
Against the 15.00 euros in the trades agreement that is around nineteen to twenty-two per cent more, and that is basic pay alone.
Add the collectively agreed extras and the gap widens. The annual bonus payment of 85 per cent of a month's pay works out at a good seven per cent when spread across the year, and the employer's share of the supplementary pension at a further five to eight per cent depending on the scheme. For an employee at increment 4 you therefore land at around 20.50 euros an hour before social security, materials and overhead are considered at all. Social security contributions arise on both sides and largely cancel out in the comparison.
That establishes the starting point: the difference in basic pay is a good fifth, and considerably more on a full-cost basis. Three further items carry it.
Incremental progression. It runs automatically: increment 2 after one year, increment 3 after three, increment 4 after six, increment 5 after ten and the top increment after fifteen years. Between the starting rate and the top increment in pay group 2 lie around twenty-four per cent. Grading into a low group therefore has its effect mainly on new appointments, while the cost of an existing workforce rises over the years without any decision being taken.
Set against that is something that features again below: the same workforce brings experience and knowledge of the building, and both reduce the cost of training and of errors.
The annual bonus payment. It is collectively agreed under the TVöD and paid in November. The building cleaning trades agreement has no equivalent. That is a full cost block frequently missing from administrative calculations, because it does not appear in the monthly pay.
The supplementary pension. Occupational pension provision in the German public sector is financed predominantly through a levy, the larger part of which the employer bears. At the VBL, the employer's share of the levy is 5.49 per cent of pensionable pay. At the municipal supplementary pension schemes the rate differs; the Rhineland scheme, for instance, states a total levy rate of 7.86 per cent for 2026 in its western accounting group.
Which rate applies to you depends on your scheme and accounting group and is set out in its statutes. What holds in every case: the item is large enough to turn a result, and it is missing from model calculations more often than you would suppose. Private providers likewise bear pension costs, though in a different form and usually at a lower level, and theirs sit inside their tender price.
Absence. That the sickness rate in the public sector runs higher than in the private sector is not an assertion but something that can be quantified. The Scientific Institute of the AOK health insurance fund records a sickness rate of 7.5 per cent for public administration in 2023, against 5.4 per cent in the services sector and 6.6 per cent across all sectors.
The comparison by occupation shows, however, that cleaning taken on its own also sits above the average: 6.8 per cent in building cleaning, 6.9 per cent in property and personal protection. Musculoskeletal conditions account for a considerable share of it.
A municipal cleaner therefore carries both factors at once, the sector and the occupation, and on top of that a workforce that is as a rule older. Cost in-house cleaning using the trades sector figure and you are entering absence too low.
Two qualifications belong with this. The electronic certificate of incapacity for work has since 1 January 2022 also captured short absences that frequently never reached the insurer before. Figures from before 2022 are therefore not comparable with current ones, and part of the visible increase is purely a matter of recording. And a sector average does not replace a figure of your own: absence depends on age structure, physical demands and the organisation of the individual operation. What belongs in the calculation is your own absence over the past three years, separated into planned and unplanned, and the associated cost of cover.
Fail to reflect these items and you arrive at a result making in-house cleaning look cheaper than it is. We have seen more than one economic appraisal in which the authority's own staff were simply entered at the sector minimum wage.
The productivity rate matters more than the hourly wage
A point that regularly gets lost in the discussion: whether a cleaner manages 180 or 260 square metres an hour affects the cost more than a euro's difference in pay.
That can be worked out. For 1,000 square metres you need five hours at a productivity rate of 200 square metres per hour and four at 250. That is twenty per cent less time for the same area. A euro more on an hourly wage of 15 euros amounts by contrast to just under seven per cent, and even with on-costs the gap remains clear. Argue about wage figures while taking the productivity rate unchecked from a table and you are arguing about the smaller quantity.
Useful orientation comes from the guidance note of the RAL quality association for building cleaning on productivity figures in routine cleaning. For administrative buildings it gives 160 to 230 square metres per hour for offices, 60 to 90 for sanitary areas, 130 to 200 for stairwells and 250 to 350 for manually cleaned corridors. For schools, classrooms are added at 180 to 300, and for day nurseries group rooms at 120 to 220.
The association itself states that standardised values cannot exist and that its figures claim no binding force. More important than the table, therefore, is the worked example it supplies, because that shows what comparability actually hangs on.
An office of 18 square metres, with floor cleaning two and a half times and waste emptying five times a week, yields a productivity rate of around 350 square metres per hour. Add the desk, the window sill and the telephone and it falls to 293. Include the proportionate set-up and travel time and it lands at 226.
The same room, the same person, a difference of more than fifty per cent. Argue about productivity rates without setting the scope of services alongside them and you are arguing about figures that have nothing to do with each other.
Every rate taken from elsewhere therefore needs a statement of what it covers: floor cleaning only or every service in the room, with or without set-up and travel time, which interval, which quality level. And defensible values still arise only for the specific building, through a site inspection with a time study. For cases of doubt the association recommends the same thing we do: a trial clean by the tenderers under consideration and a post-costing once the contract has started.
One particularity of the public sector should be planned for. Raising productivity rates alters working processes and workload, which brings the staff council into play. How far its participation rights reach in the individual case depends on the relevant state staff representation act and is to be clarified within the organisation. For planning purposes that means: a cost advantage that would be arithmetically achievable is not thereby schedulable. We have seen appraisals where implementation began a full year later than intended.
What speaks for in-house cleaning
The question of whether in-house cleaning is qualitatively better is readily asked and rarely answered cleanly. Municipal audit reports mostly find no evidence for it, and there is a reason: what they measure is the state of cleanliness at the time of the audit, and that says little about causes.
What such an audit reflects poorly is the effect of a stable workforce. Where cleaners are permanently employed and have looked after the same rooms for years, something arises that no productivity table captures: they know the building, they know the users, they see when something is not right, and they report it. Turnover is lower, training falls away, arrangements are made quickly.
We say this more plainly than many economic appraisals do: in-house cleaning frequently produces better results in practice, and the reason lies in belonging. Somebody who is part of the organisation's own workforce, who is known in the building and who will still be working there in five years works differently from somebody who will be deployed in another building the day after tomorrow. That is not a statement about the diligence of individual employees at a contractor. It is a statement about what integration does to a sense of responsibility.
Quantifying it is hard. Anyone wanting to bring it into the calculation regardless can attempt it through the items it influences: training effort per new member of staff, the number of staff changes in the building, the volume of complaints, cases of damage to surfaces.
One precondition belongs with it, though, and it is not met everywhere. The effect arises where the cleaning is managed within the organisation, where somebody is responsible and where the staff are genuinely integrated. Where in-house cleaning runs along on the side, with no site supervision and no point of contact, it does not arise.
It is not an argument against the figures. In the appraisals we have produced or reviewed, contracting out predominantly showed the lower arithmetical cost, and anyone claiming the opposite for a particular building has to demonstrate it. The question of what a local authority gets for the difference belongs in the same appraisal, though. Otherwise you optimise a figure and lose something in the process that only shows up years later and is then hard to recover.
The reverse holds just as much: contracting out is more demanding to manage. That is rarely down to the price, which was formed in competition. It is down to the fact that the service does not stay right by itself and drifts across the contract term where nobody in the organisation looks regularly.
How we go about it
First: describe the service before valuing it. Which areas, which type of use, which intervals, which quality level. Until that is settled you are comparing two different services and wondering at the difference. Frequently that is precisely the real finding: the in-house cleaning delivers something other than what would be tendered, and nobody has ever written it down.
Second: establish the time required, do not estimate it. From areas, use and interval follows a requirement in hours. The benchmarks for it have to be adapted to the building, because an open-plan office and an area of the same size with twenty single rooms take different lengths of time.
Third: calculate with productive hours, not with paid hours. This is the error occurring most often in administrative calculations and having the largest single effect. From an employee's paid annual hours come off leave, public holidays, sickness, training and instruction, and time for meetings and organisation. What remains are the hours in which cleaning actually happens. Only full costs divided by that figure give the cost per productive hour, and only that quantity is comparable with a tender price. Treat paid and productive hours as the same and you cost in-house cleaning too cheaply by a double-digit percentage.
Fourth: cost both sides in full. For in-house cleaning by the four blocks above; for contracting out, the market price to be expected plus the contract management remaining in the organisation. Where items are not known, work with disclosed approximations rather than leaving them out.
Fifth: place the result honestly. An economic appraisal delivers a figure, but rarely an unambiguous one. Frequently the two options lie closer together than expected, and other things then decide. What makes sense, therefore, is not to calculate a single point value but two or three scenarios with disclosed assumptions on productivity rate, absence and management effort. What you want to see is the point at which the decision tips.
There is a third route that is often missing from the appraisal: splitting. Sensitive areas in-house, general area cleaning contracted out. Or deep cleaning and special services external, routine cleaning internal. More complicated to cost and frequently the most workable solution in practice.
When the appraisal is worth doing
Not every year. It makes sense before a re-tender, after a significant increase in floor area, following a step change in pay rates, or where the question is on the table anyway.
And particularly where you think you already know the answer. We have produced appraisals confirming what the client suspected, and others producing the opposite. In both cases the result mattered less than the fact that it was substantiated, before the council, the oversight body and the audit office.