From Spreadsheets to Smart Scheduling: A Cleaning Company's Digital Journey

Most cleaning companies that move to cleaning scheduling software expect a better looking grid. The grid was never the problem. The problem is the number of decisions a weekly roster holds and how few of them anybody checks before the week goes out.

This blog is a small research paper. One question. The published data that answers it. The finding. And what it means for a contractor on Monday morning. Every figure is taken from a named source or calculated from one, with the working shown in a numbered table.

Check the week before it goes out and not after it goes wrong. Operify AI is cleaning scheduling software that tests every assignment against the contract and the rules before you publish.

Key Findings#

• 1,500 decisions a week. An illustrative estate of 300 operatives on 40 sites working five shifts a week creates 1,500 roster assignments a week and 78,000 a year. Each needs at least two statutory checks under the Working Time Regulations [1].
• 86 per cent gave a wrong result. An audit of 50 spreadsheets in live use, covering 270,722 formulas, found that 86 per cent produced at least one wrong result [2].
• 6 to 38 wrong entries a week. Published cell error rates of 0.4 to 2.5 per cent, applied to the roster above, give between 312 and 1,950 wrong entries a year [2][3].
• 10,092 cover hours a year. At the ONS sickness absence rate of 2.9 per cent for elementary occupations [4], the estate needs 10,092 hours of cover a year. Every 10 per cent of those hours moved from overtime to planned relief is worth £8,477.
• 59 per cent called in at short notice. Living Wage Foundation polling found that 59 per cent of workers whose hours vary had been called in with less than a week of notice [5]. The government plans rights to reasonable notice and short notice payments for 2027 [6].

1. The Question#

How many roster decisions does a cleaning contractor make each week, and how many of them can a spreadsheet check before the week is published?

The question matters because of who the sector employs. The British Cleaning Council reports that 67 per cent of workers in cleaning activities work part time and that more than 83 per cent of businesses in the cleaning activities sub industry are micro businesses with fewer than ten people [7]. Part time work means more shifts for every contracted hour. More shifts means more decisions.

There is no official measure of how the sector schedules. The ONS Management and Expectations Survey found that 61 per cent of UK firms adopted specialised software in 2023 [8]. That survey covers only firms with 10 or more employees, which leaves out most cleaning businesses. So we built the answer from the decisions themselves rather than from a survey.

2. The Data and the Method#

We used eight published sources for the volume and the risk, and four statutory rates for the cost. Each source answers one part of the question.

Table 1. Sources used and what each one answers

Ref

Source

What it contains

Used for

1.1

Working Time Regulations 1998, as amended [1]

Regulations 4 and 10

Checks per assignment

1.2

Powell, Baker and Lawson 2009 [2]

Audit of 50 operational spreadsheets and 270,722 formulas

Error rates

1.3

Panko 2000 [3]

Field audit cell error rates of 0.4 to 2.5 per cent

Error range

1.4

ONS sickness absence 2024 [4]

2.9 per cent for elementary occupations

Cover hours

1.5

Living Wage Foundation 2024 [5]

Polling of more than 2,000 UK workers

Notice of shifts

1.6

GOV.UK Employment Rights Act timeline [6]

Zero hours measures planned for 2027

Future cost

1.7

British Cleaning Council 2026 [7]

Workforce and business structure

Sector context

1.8

ONS Management and Expectations Survey 2023 [8]

Technology adoption in firms with 10 or more employees

Benchmark

We applied those rates to an illustrative estate of 300 operatives across 40 sites. Each operative works 25 hours a week over five shifts, which reflects a part time workforce. The estate is a model and not a customer. Costs use the April 2026 National Living Wage of £12.71 [9], with employer National Insurance at 15 per cent above the secondary threshold [10] and the minimum employer pension contribution of 3 per cent [11]. Statutory holiday of 5.6 weeks [12] leaves 46.4 productive weeks a year.

3. What the Data Shows#

3.1 A Weekly Roster Is 1,500 Decisions#

The estate produces 1,500 assignments a week, or 37.5 for every site. Each one places a person at a site for a set time. Regulation 4 says working time must not exceed an average of 48 hours a week over a 17 week reference period. Regulation 10 gives an operative 11 consecutive hours of rest in each 24 hours [1]. That is at least 3,000 statutory checks a week, before competence or site induction is even considered.

The 48 hour check also needs history. A rolling 17 week average across 300 operatives means 5,100 operative weeks of hours in view every time the week is published. A spreadsheet holds that history only if somebody keeps it, and checks it only if somebody builds the formula.

3.2 Spreadsheets Fail at a Measured Rate#

Powell and colleagues audited 50 spreadsheets in live operational use. 86 per cent contained at least one error that produced a wrong result, and 94 per cent contained an error on a wider definition. The cell error rate was 0.87 per cent for wrong results and 1.79 per cent on the wider definition [2]. Hard coded values made up 37.7 per cent of error instances, and reference errors 32.9 per cent. Those are the two errors a copied roster template invites. Last week pasted over this week. A row added that the total never picks up.

Field audits summarised by Panko found cell error rates between 0.4 and 2.5 per cent [3]. Table 2 applies that range to the estate.

Table 2. Modelled wrong entries on the illustrative estate

Ref

Cell error rate

Source

Wrong entries a week

Wrong entries a year

2.1

0.4 per cent

Butler field audit in Panko [3]

6

312

2.2

0.87 per cent

Powell wrong results [2]

13

679

2.3

1.79 per cent

Powell wider definition [2]

27

1396

2.4

2.5 per cent

Lukasik field audit in Panko [3]

38

1950

Modelled. Rates applied to 1,500 assignments a week and 78,000 a year. The rates were measured on formula cells in general business spreadsheets, and their use on roster entries is an assumption.

Even the lowest audited rate gives six wrong entries every week. Not every wrong entry becomes a missed shift, because some are caught. But they are caught by a person reading a grid, and the rates above are what people reading grids achieve.

3.3 Late Notice Pushes Cover Onto the Dearest Route#

Cover is where the spreadsheet meets the group chat. At 2.9 per cent absence [4], the estate loses 10,092 productive hours a year, or about 194 hours a week. Each hour is covered by planned relief or by overtime, or not at all.

Table 3. The price of a cover hour at April 2026 rates

Ref

Line

Working

Value

3.1

Wage per productive hour

£12.71 times 52 weeks divided by 46.4 weeks

£14.24

3.2

Planned relief hour

Row 3.1 plus 15 per cent NI and 3 per cent pension

£16.81

3.3

Overtime hour at time and a half

Row 3.2 times 1.5

£25.21

3.4

Premium on every late hour

Row 3.3 minus row 3.2

£8.40

3.5

Cover hours a year

348,000 productive hours times 2.9 per cent

10092

3.6

Value of moving 10 per cent of cover to relief

1,009.2 hours times row 3.4

£8477

Calculated at marginal rates because a cover hour sits above the NI secondary threshold and the pension lower limit. Overtime at time and a half is an assumption.

Whether an hour lands on relief or overtime depends mostly on how early the gap is seen. Living Wage Foundation polling found that 59 per cent of workers whose hours vary had been called into work with less than a week of notice, and 13 per cent with less than 24 hours [5]. That is the message in the group chat at 5.47 in the morning.

For zero hours and low hours workers, the government plans a right to reasonable notice of shifts and a payment for shifts cancelled or cut short at short notice. GOV.UK puts both in 2027, with timings to be updated after consultation [6]. Today, a late change costs a premium. From 2027, it may cost a payment as well.

Six wrong entries a week is the floor and not the ceiling. See how cleaning scheduling software refuses to publish a week with an uncovered SLA job and checks working time at the moment of assignment.

4. What It Means for a Contractor#

The digital journey is usually told as paper, then spreadsheet, then app. The data points to a better measure. How many of the 1,500 weekly decisions does the system check rather than a person.

Table 4. What each stage checks before the week is published

Ref

Check

Paper and group chat

Spreadsheet

Contract driven scheduling

4.1

Empty slot

By eye

By eye or a formula somebody built

Publish blocked while an SLA job has a gap

4.2

Working time limits

Not checked

Only if a formula is built and kept current

Checked at assignment with pending claims counted

4.3

Certificate and signed RAMS

Not checked

Not checked

Task cannot start until both are valid

4.4

Cover request

Group chat

Group chat

Open shift in a claims pool on the record

4.5

Proof the shift happened

Signature sheet

None

Clock in at a scanned checkpoint that works offline

The final column describes Operify AI as set out on its website. It makes no claim about any other product.

The order of the move matters more than the speed of it.

• Load the contract first. Hours and SLA jobs and frequencies go in before any names. A roster checked against nothing passes every test.
• Move the checks before the screens. Working time and certificate checks remove more risk than a new layout does.
• Put cover on the record. An open shift that operatives claim shows the gap days before a group chat would.
• Measure delivery at the point of work. A roster is a plan. A clock in at a checkpoint is a record.
• Compare delivered hours with priced hours. The roster matters because the contract was priced on those hours.

In Operify AI, the scheduling and roster module expands the contract into recurring tasks and shifts, and will not publish a week with a hole in it. The mobile app records the clock in with no signal. The health and safety gate blocks a task when a certificate lapses. The gap between hours priced and hours delivered then shows as margin leakage each week rather than in the year end accounts.

5. What This Does Not Show#

• The error rates are general. They were measured on formula cells in business spreadsheets between 1995 and 2009, and not on cleaning rosters [2][3].
• The estate is illustrative. 300 operatives on 40 sites is a model and not a customer.
• Absence is not sector specific. The 2.9 per cent rate covers all elementary occupations [4].
• Overtime may be paid at plain rate. Where it is, the premium in Table 3 closes, and the cost of late cover moves into missed shifts and, from 2027, into short notice payments.
• The polling is not cleaning specific. It covers all UK workers whose hours vary [5].
• The ONS survey leaves out micro firms. It covers firms with 10 or more employees only [8].

6. Conclusion#

Moving from spreadsheets to smart scheduling is not a change of screen. It is a change in who checks the decisions. On the illustrative estate, a spreadsheet leaves at least 3,000 statutory checks a week to a person, and the audited error rates suggest 6 to 38 entries a week will still be wrong. Finding a gap late costs £8.40 on every cover hour that goes to overtime, and from 2027 it may carry a payment too.

A contractor that moves the checks into the system first gets the benefit first. The rest of the journey follows from that.

Bring us one contract and one live week. In thirty minutes, we will rebuild it in cleaning scheduling software, with every check running, and show you what it catches. Book a demo.

FAQ#

What is cleaning scheduling software?#

Software that builds the roster from the contract and checks each assignment before the week is published. The useful difference from a spreadsheet is not the grid. It is the checks that run without a person. See the scheduling and roster module for how that works.

How many roster decisions does a cleaning contractor make?#

An illustrative estate of 300 operatives on five shifts a week makes 1,500 assignments a week and 78,000 a year. On 40 sites, that is 37.5 decisions per site every week.

How common are spreadsheet errors?#

An audit of 50 operational spreadsheets by Powell, Baker and Lawson found that 86 per cent produced at least one wrong result, and 0.87 per cent of formula cells were wrong.

Why does a spreadsheet roster miss working time breaches?#

Because it only checks what somebody built into it. The 48 hour limit is an average over 17 weeks, so the check needs 5,100 operative weeks of history on a 300 operative estate.

What does late cover cost?#

At April 2026 rates, a planned relief hour costs about £16.81, and an overtime hour at time and a half about £25.21. Every late hour costs £8.40 more than an early one.

How many cover hours does a 300 operative estate need?#

At the ONS absence rate of 2.9 per cent for elementary occupations, it needs 10,092 hours a year, or about 194 hours every week.

Will short notice shift changes cost more from 2027?#

The government plans a right to reasonable notice of shifts and a payment for short notice cancellation for zero hours and low hours workers. GOV.UK puts both in 2027, with timings to follow consultation.

How many UK firms use specialised software?#

The ONS found that 61 per cent of UK firms with 10 or more employees adopted specialised software in 2023. There is no equivalent figure for micro firms, which make up more than 83 per cent of cleaning activities businesses.

Where should a cleaning company start the move from spreadsheets?#

With the contract. Load hours and SLA jobs first, then the working time and certificate checks, then cover, then clock in. The mobile app closes the loop with a record of delivery.

How much does Operify AI cost?#

Pricing is built around operatives and sites and the modules taken rather than a per head licence, so cost stays steady through turnover. The figure is given on the demo call. See pricing or email hello@operifyai.co.uk.

References#

[1] UK Government. The Working Time Regulations 1998, as amended. Regulation 4 maximum weekly working time and regulation 10 daily rest. Latest available revised version. Available at Working Time Regulations 1998 regulation 4 and regulation 10 (accessed October 2026).

[2] Powell S G and Baker K R and Lawson B. Errors in Operational Spreadsheets. Journal of Organizational and End User Computing 21(3), pages 24 to 36. 2009. Peer reviewed audit of 50 spreadsheets and 270,722 formulas. Available at Errors in Operational Spreadsheets (accessed October 2026).

[3] Panko R R. Spreadsheet Errors. What We Know. What We Think We Can Do. Proceedings of the European Spreadsheet Risks Interest Group. 2000. Academic review summarising earlier field audits. Available at Spreadsheet Errors What We Know (accessed October 2026).

[4] Office for National Statistics. Sickness absence in the UK labour market 2023 and 2024. Released 4 June 2025. Elementary occupations 2.9 per cent, against a UK average of 2.0 per cent. Available at ONS sickness absence in the UK labour market (accessed October 2026).

[5] Living Wage Foundation. Sufficient and Predictable Hours. Policy briefing. October 2024. Polling of more than 2,000 UK workers by an advocacy organisation. The government factsheet on reforms of zero hours and similar contracts quotes the same figures. Available at Sufficient and Predictable Hours and Factsheet reforms of zero hours and similar contracts (accessed October 2026).

[6] UK Government. Plan to Make Work Pay and Employment Rights Act timeline update. Updated September 2026. Right to guaranteed hours and right to reasonable notice and short notice payments in 2027, with timings to be updated after consultation. Available at Employment Rights Act timeline update (accessed October 2026).

[7] British Cleaning Council. Research Report 2026. The Cleaning, Hygiene and Waste Industry. Published February 2026. 67 per cent of workers in cleaning activities work part time. More than 83 per cent of the 26,315 businesses in the cleaning activities sub industry are micro. Available at British Cleaning Council Research Report 2026 (accessed October 2026).

[8] Office for National Statistics. Management practices and the adoption of technology and artificial intelligence in UK firms 2023. Released 24 March 2025. Management and Expectations Survey of firms with 10 or more employees. Available at ONS management practices and technology adoption 2023 (accessed October 2026).

[9] UK Government. National Minimum Wage and National Living Wage rates. £12.71 an hour for workers aged 21 and over, from April 2026. Available at National Minimum Wage and National Living Wage rates (accessed October 2026).

[10] HMRC. Rates and thresholds for employers 2026 to 2027. Employer Class 1 rate of 15 per cent above a secondary threshold of £5,000 a year. Available at rates and thresholds for employers 2026 to 2027 (accessed October 2026).

[11] The Pensions Regulator. Making contributions to your pension scheme. Minimum employer contribution of 3 per cent of qualifying earnings between £6,240 and £50,270. Available at making contributions to your pension scheme (accessed October 2026).

[12] UK Government. Holiday entitlement. 5.6 weeks of paid holiday a year. Available at holiday entitlement rights (accessed October 2026).

About Operify AI#

We sell scheduling software. That is a commercial interest in this finding, and readers should weigh it. Two things constrain it. Every rate used here is published and named, so any reader can rerun the arithmetic. Section 5 states what the data does not show.

Operify AI is operated by Codedevza AI Ltd, company number 16485057. Correspondence to hello@operifyai.co.uk. Book a demo at operifyai.co.uk/demo.