Robot Labor Cost Avoidance: Auditing the Claim for Finance

The Number Your CFO Will Ask For

There are three different labour numbers in circulation and they are routinely conflated. The first is avoided headcount, the number of people not hired. The second is avoided hours, the labour content removed from the schedule. The third is avoided cost, the money that does not leave the bank.

Only the third survives a finance review, and it is the smallest of the three. Avoided headcount is the number vendors quote because it is the largest and the easiest to compute, but a vacancy that is never filled does not always reduce the cost base: it may reduce output, it may be covered by overtime, or it may be absorbed by an agency at a higher rate. The cost actually avoided depends on how the work was being covered, not on how many positions were notionally removed.

The practical test is to state the saving in the same units as the existing cost line. If the site currently spends a monthly figure on cleaning labour, the business case discusses a monthly figure, not a headcount. Everything that follows is an exercise in getting from one to the other honestly.

It is worth saying why this matters beyond a finance argument. A business case that overstates the saving is approved, deployed, and then fails at the first quarterly review when the actual spend has not moved as promised. The failure is rarely the robot; it is the arithmetic that was used to justify it, and it damages the credibility of every subsequent automation proposal by the same team.

Building the Burden Rate That Survives Finance Review

The burden rate is the fully loaded hourly cost of labour, and it is the multiplier most business cases either omit or understate. A site comparing a robot’s hourly cost against a wage rate is comparing against roughly two-thirds of the real number.

The components below are additive and most of them are discoverable from payroll and from the operations manager. The ranges are indicative for outsourced commercial cleaning and in-house facility teams in mature markets; they vary enormously by region and by employment model, and should be replaced with the site’s own figures wherever they exist.

Component	What it covers	Typical range (% of wage)
Base wage	Paid hourly or salaried rate	100% (the baseline)
Employer on-costs	Social contributions, pension, insurance, payroll taxes	15–35%
Paid absence	Holiday, sickness, statutory leave	8–14%
Supervision	Team leader and management time apportioned	8–15%
Recruitment and turnover	Advertising, agency, onboarding, repeat hiring	5–15%
Training and certification	Initial and refresher, including paid training time	2–6%
Equipment handling	Issue, maintenance, loss of consumables and PPE	2–5%

Work an example. A cleaner on a base rate of 14.00 per hour, with on-costs at 25 percent, absence at 10 percent, supervision at 10 percent, turnover at 8 percent, training at 3 percent and equipment handling at 2 percent, lands at a burden rate of roughly 22.10 per productive hour. The base rate understates the cost by about 58 percent.

Two refinements matter more than the exact percentages. First, the burden should be applied to productive hours, not to paid hours, because paid time includes breaks and travel that the robot does not eliminate. A schedule of 40 paid hours may be 33 productive hours, and dividing cost by 40 rather than 33 understates the hourly cost of the work being automated.

Second, the comparison should use the burden rate that applies to the specific labour being displaced. Agency cover, overtime and weekend rates carry different multipliers from base-rate weekday labour, and a site that covers its cleaning with a mix of all three has a blended rate that is higher than the payroll figure suggests. Using the cheapest rate in the mix is the most common way a business case is inflated without anyone intending to inflate it.

Vacancy and Turnover: Why Modelled Savings Disappear

This is the section that most business cases omit and most post-deployment reviews discover.

The standard modelling assumption is that the robot displaces labour that is currently being paid for, so the saving equals hours removed multiplied by the burden rate. On a fully staffed site that is approximately true. On a site running with vacancies, it is not, because the labour being displaced was not being performed by a paid employee in the first place.

Run a worked example on a site that is 15 percent under establishment. The establishment is 20 full-time equivalents at 42,500 each, so the budgeted labour cost is 850,000. Actual headcount is 17, so actual spend is 722,500. The site is not currently spending 850,000; it is spending 722,500 and absorbing the gap through longer intervals between tasks, deferred deep cleans and some overtime.

Now introduce a robot fleet that removes the equivalent of four full-time roles. If the model assumes the establishment cost is displaced, it claims a saving of 170,000 a year. The actual cash effect is different, and depends on how the site responds.

If the four roles removed are taken from the three vacancies plus one filled post, the immediate cash saving is the cost of that one filled post, around 42,500, plus whatever overtime was being used to cover the vacancies. The remaining three roles were never being paid for, so removing them saves nothing in cash; it merely formalises a gap that already existed.

The real benefit in that scenario is not cost but restored service quality: intervals return to specification, deferred work gets done, and overtime falls. That is a genuine and often substantial benefit, but it is not a labour cost saving and should not be presented as one. Finance will see through it at the first variance review, because the labour line will fall by 42,500 and not by 170,000.

Site state	Naive modelled saving	Realistic cash saving	Where the rest goes
Fully staffed	Hours removed × burden rate	Close to the model	Small residual into supervision and quality
10–20% vacancy	Hours removed × burden rate	Only the filled posts displaced	Service quality restoration, overtime reduction
Heavy agency cover	Hours removed × base rate	Larger than expected per hour	Agency margin and premium disappear entirely
Growing site	Hours removed × burden rate	Avoided future hiring	Benefit is deferred cost, not reduced cost

The agency row is the happy exception and it is worth knowing. Agency labour often costs 35 to 60 percent more per productive hour than directly employed labour, and all of that premium disappears when the hours are no longer needed. A site covering a third of its cleaning with agency staff has a much stronger case than one fully staffed at base rates, even though the headline headcount is smaller.

Turnover compounds the same distortion. A site with 40 percent annual turnover is not paying a stable wage bill, it is paying a wage bill plus a permanent recruitment and training overhead. The robot removes the hours and therefore the turnover that attached to those hours, and the turnover component of the burden rate is a real saving rather than a modelled one. This is one of the few places where the honest saving is larger than the naive headcount figure, which is why the burden rate build-up matters in both directions.

What to Baseline Before the First Unit Arrives

A saving that cannot be measured afterwards will be argued about instead. Six measurements taken before deployment convert the business case into an auditable claim, and all six are cheap to collect.

• Productive hours by task type per week. Not paid hours. The split between routine, periodic and reactive work matters, because a robot usually displaces routine work and leaves reactive work intact.
• The actual blended labour cost per productive hour, including agency and overtime, from payroll rather than from budget.
• Vacancy rate and overtime hours, weekly, for the three months before deployment. These two numbers determine how much of the theoretical saving is real.
• Service level achieved, measured the way the contract or the internal standard measures it. If intervals were being missed, this is the baseline that lets the quality benefit be demonstrated rather than asserted.
• Turnover and training spend for the preceding twelve months, so the reduction is visible later.
• The task-level time standard: how many square metres of floor, rooms or deliveries per productive hour. This is the conversion factor between robot performance and labour displacement, and without it the whole calculation is guesswork.

The task-level time standard is the one most often missing, and it is the most important. It connects a robot specification to a labour number: a unit that covers 1,800 square metres per hour, running a six-hour effective shift, displaces 10,800 square metres of task per night, which at a manual productivity of 350 square metres per productive hour is roughly 31 hours of labour. That chain of arithmetic is the entire business case, and every link has to be measured rather than assumed. The robot-side figure is discussed in deployment throughput planning, and the manual baseline in the cleaning labour cost model.

Baseline windows of eight to twelve weeks are adequate for most sites. Shorter windows catch a seasonal anomaly, longer windows delay the project for diminishing returns.

Where the Saving Lands in the P&L, and Why That Decides the Project

A business case can be arithmetically correct and still be abandoned, and the reason is usually that the saving accrues to a different budget from the one carrying the cost.

The robot is typically a capital or service charge against a facilities or operations budget. The labour it displaces sits often in a departmental budget, sometimes in a contracted service line, and occasionally in a shared overhead allocation. If the operations director pays for the robots and the department keeps its headcount budget, the operations director has incurred a cost and received no relief, and the project becomes a political problem regardless of its return.

Structure	Who pays	Who benefits	Practical fix
Central capex, departmental labour	Facilities or central	Operating departments	Transfer the released budget, or charge back the service
Outsourced cleaning contract	Facilities	Contractor margin	Renegotiate the specification and the price together, not the price alone
RaaS from the operating budget	Facilities opex	Contracted service line	Cancel or reduce the equivalent service scope at the same time
Site pays, group reports	Site	Group consolidation	Report the site-level benefit in the site P&L, not only at group

The outsourced row captures the most common disappointment. A site that keeps its cleaning contract at the same price while adding robots has added a cost and bought improved service, not a saving. The saving only materialises if the specification is renegotiated downward at the same time, which requires the contract to be reopened and the frequency and scope re-based on the robot’s actual coverage. Doing this at renewal is far easier than mid-term, which is why the sequencing of deployment against contract dates is a commercial decision and not just an operational one. The tender-side mechanics are covered in cleaning contract tender and robot pricing.

The practical recommendation is to name, in the business case itself, the budget line that will reduce and the date by which it will reduce. A case that cannot name that line is a case for service improvement, which may still be worth approving, but it should be approved on those grounds rather than on a saving that will not appear.

Redeployment, Not Replacement: The Honest Version

The phrase is used so often that it has lost meaning, and the versions that survive scrutiny are narrower than the versions that get presented to staff.

The claim is usually that displaced hours are redeployed to higher-value work, so the saving is capacity rather than cash. That is a legitimate outcome, and it is worth quantifying rather than asserting. A team of seventeen that previously spent 70 percent of its time on routine floor work and now spends 40 percent has gained a substantial capacity for periodic deep cleaning, inspection and reactive response. That capacity is a real benefit if the site has work that was previously being deferred, and it is worth listing the deferred work explicitly, because that list is the benefit.

It is not a real benefit if the deferred work does not exist. On a site already operating to specification with no backlog, redeployment is a euphemism for either headcount reduction or slack, and presenting it as a benefit without saying which invites cynicism from the team and scepticism from finance.

Two practical points make redeployment credible. First, redeployed hours need a destination with a defined output: a named task list, a frequency, a standard. Without that, the capacity evaporates into the schedule and the benefit is unmeasurable in practice. Second, the change needs the same care as any other process change, because the people affected are being asked to do different work with different skills, and the failure mode is quiet under-use rather than open resistance. The change management side is treated in human-robot collaboration and change management, and the training structure in staff training and certification.

The version that holds up in audit is: the routine labour content falls by a measured number of hours; those hours are redeployed to a named list of tasks with a measured output, or released; and the cash saving is claimed only for the portion that was actually being paid for. That last clause is the whole discipline.

Recalculating at Year Three, When Wages Have Moved

A labour saving is a comparison between two cost curves, and both of them move. The business case is settled at signature and then quietly decays or improves depending on which curve rises faster.

Manual labour cost tends to rise with wage inflation, statutory on-cost changes and minimum wage adjustments, which improves the case over time. The robot’s cost per hour falls slightly as the capital is depreciated but can rise with service fees, parts inflation and price escalation clauses at renewal. The net direction depends on the ratio, and the ratio is the number worth tracking.

Scenario	Wage inflation	Service cost inflation	Effect on case
Base	3.5% p.a.	3.5% p.a.	Neutral; payback holds
Wage-led	6.0% p.a.	3.5% p.a.	Improves; payback shortens by months
Service-led	3.5% p.a.	7.0% p.a.	Deteriorates; watch the escalation clause
Stagnant wages	1.5% p.a.	3.5% p.a.	Case decays; extension becomes marginal

Work the magnitude. A case with a five-year payback and a neutral assumption gains roughly eight to ten months of payback improvement under the wage-led scenario, and loses a similar amount under the service-led one. Over a fleet of any size that is a material sum, and it is the reason the price escalation clause at renewal deserves more attention than the headline service fee. The mechanics are set out in contract renewal and price escalation.

The recommendation is to recalculate the case annually with actuals, using the same burden rate method, and to treat the recalculation as a routine management task rather than a post-mortem. A case that is recalculated annually will be defended; a case that is computed once and referenced for three years will be quietly dropped.

An Audit Trail a CFO Will Accept

The whole argument reduces to one property: every number in the case should be traceable to a document that predates the deployment.

The payroll burden rate comes from payroll. The vacancy rate comes from the establishment report. The task-level time standard comes from a measured baseline study, ideally the site survey done before quotation. The robot throughput comes from the acceptance test, not the datasheet. The displaced hours come from the schedule. Each of these is a small piece of evidence, and together they are the difference between a saving and a claim.

It is worth running the case in the same format finance uses for other capital requests, with the burden rate shown, the vacancy adjustment shown, and the cash saving separated from the service-quality benefit. Presenting both, honestly labelled, is far more persuasive than a single inflated number, because it demonstrates that the arithmetic was done rather than borrowed. The discounting mechanics belong in the ROI framework and the payback convention in payback period calculation; both are worth aligning with whatever convention the finance team already uses, because a mismatch in method invites a rejection on form rather than substance.

The upstream budget question, including how to phase capital across years, is covered in budget planning, and the cost of the downtime that partly offsets the labour saving is in downtime cost and OEE.

If you are building a labour case for a specific site, we can provide the task-level time standards for our cleaning and delivery platforms, measured on real deployments, so the productivity link in the chain is evidence rather than assumption. The quote form reaches a technical contact directly.