Ask a leadership team what manual admin costs and you'll usually get a shrug, or a guess based on one person's salary. Neither supports a decision. Admin is spread across many roles, hidden inside salaries, and multiplied by errors and delays that never appear on a timesheet.
This is a method you can run with a spreadsheet and two weeks of observation. It follows the same structure we use in the Altitude Diagnostic, simplified so you can do it yourself.
What counts as manual admin
Manual admin cost is the total cost of repetitive, rules-based work done by people that a better process or system could remove, reduce or speed up. It includes the fully loaded cost of the time, the cost of the errors and rework it causes, and the commercial cost of the delays it creates, such as slower quotes or late invoices.
Typical examples:
- Re-keying enquiries from web forms, emails and phone messages into a CRM or job system
- Building quotes from spreadsheets and price lists
- Chasing approvals, signatures and missing information
- Sending payment reminders and reconciling payments
- Compiling weekly or monthly reports from several systems
- Updating rosters, schedules and bookings by hand
Step 1: Build a task inventory
Don't start with a survey asking people how much time they waste. Start with the work. Sit with each team for an hour, follow a few jobs from enquiry to invoice, and pull volumes from your systems wherever you can.
| Field | What to record | Example |
|---|---|---|
| Task | A specific, observable activity | Enter a web enquiry into the job system |
| Trigger | What starts it | New form submission |
| Frequency | How often, from system data where possible | 400 a month |
| Duration | Minutes per occurrence, from a timed sample | 6 minutes |
| People | How many people touch each occurrence | 1 |
| Role | Who does it, so you apply the right rate | Office coordinator |
| Error rate | Share of occurrences needing rework | 3% |
| Who waits | Customers or colleagues held up by it | Customer awaiting a call-back |
Time 10 to 20 occurrences of each task rather than asking for a figure from memory. A timed sample is quick, and it moves the number from estimated to measured.
The basic formula for each task is:
Annual hours = occurrences a year × minutes each × people involved ÷ 60
Step 2: Work out a fully loaded hourly cost
Salary divided by 38 hours a week understates what an hour of someone's time costs. Use this instead:
Fully loaded hourly cost = (salary + super + other on-costs) ÷ productive hours a year
Super and on-costs
The minimum employer super contribution (the SG rate) is 12%, according to the ATO. From 1 July 2026, under Payday Super, it is calculated on each eligible employee's qualifying earnings and paid with each pay. The rate itself hasn't changed.
Other on-costs depend on your business: payroll tax where it applies, workers' compensation insurance, equipment, software and a share of overheads. Use your finance team's figure if it has one.
Productive hours
Under the National Employment Standards, full-time employees get 4 weeks of paid annual leave and 10 days of paid sick and carer's leave a year, according to the Fair Work Ombudsman. Public holidays vary by state. A full-time year of 38-hour weeks is 1,976 paid hours before any of that comes out.
Worked rate
| Line | Calculation | Result |
|---|---|---|
| Salary | Assumption | $75,000 |
| Super | $75,000 × 12% | $9,000 |
| Other on-costs | 10% of salary (assumption) | $7,500 |
| Total annual cost | $75,000 + $9,000 + $7,500 | $91,500 |
| Paid hours | 52 × 38 | 1,976 |
| Annual leave | 4 weeks × 38 | −152 |
| Public holidays | 10 days × 7.6 (assumption) | −76 |
| Personal leave taken | 5 days × 7.6 (assumption) | −38 |
| Productive hours | 1,976 − 152 − 76 − 38 | 1,710 |
| Fully loaded hourly cost | $91,500 ÷ 1,710 | $53.51 |
That is about 41% more than the $37.96 you get by dividing the salary by 1,976 hours. Use each role's own rate: an hour of a senior estimator costs more than an hour of a coordinator.
Step 3: Add the costs that never reach a timesheet
Errors and rework
Error cost = occurrences × error rate × (rework hours × hourly cost + direct cost per error)
Direct costs include credit notes, under-quoted jobs, wasted materials and repeat site visits. Your credit-note register and job notes are good places to measure them.
Delay cost
Some admin slows revenue down. A quote that takes three days to send gives a competitor three days to answer first.
Delay cost = opportunities × change in conversion rate × average value × gross margin
It can easily become the largest number in the model, and it is the least certain. Label it, show it on its own line, and never let it carry the whole case.
Opportunity cost
Released hours only create value if they go somewhere useful: faster call-backs, better follow-up, more account management. Report them as capacity created, in hours, and say what they will be used for. Don't add a dollar value for redeployed time on top of the time cost, or you'll count the same hours twice.
Step 4: Model three states and label your confidence
Three states
- Now: today's volumes, times and error rates.
- Do nothing as you grow: the same process at the volumes in your growth plan. If admin scales with volume, this is where the next hire appears.
- Improved: the process after the changes you're considering, at the same future volume, with realistic reductions rather than zero.
The do-nothing baseline is often what makes the case. A process whose cost rises with every new customer is a different decision from one that stays flat.
Confidence labels
| Label | Meaning | Example |
|---|---|---|
| Measured | Counted or timed from real data | Enquiry volume from the CRM |
| Modelled | Calculated from measured inputs and stated assumptions | Annual hours from timed samples and system volumes |
| Estimated | A reasoned figure from the people who do the work | Minutes per report from the person who compiles it |
| Inferred | Derived from indirect evidence | Win-rate change from faster quotes |
| Qualitative | Real but not sensibly quantifiable | Staff frustration with double handling |
A model that shows which numbers are soft earns more trust than one that presents every figure with the same certainty.
A worked example
An illustrative home services business with 45 staff and a small office team. Every figure is an assumption or derived from one, and all time is costed at the $53.51 rate above.
Now
| Task | Volume and time (assumptions) | Hours a year | Confidence |
|---|---|---|---|
| Re-keying enquiries | 4,800 a year × 6 minutes | 480 | Modelled |
| Building quotes | 1,440 a year × 45 minutes | 1,080 | Modelled |
| Payment reminders | 1,080 a year × 8 minutes | 144 | Estimated |
| Weekly management report | 48 a year × 4 hours | 192 | Estimated |
| Total | 1,896 |
- Time cost: 1,896 × $53.51 = $101,455 a year, about 1.1 full-time equivalents of productive time (1,896 ÷ 1,710).
- Enquiry errors: 3% of 4,800 = 144 a year × 20 minutes of rework = 48 hours × $53.51 = $2,568.
- Quote errors: 4% of 1,440 = 57.6 a year × ($200 of margin lost + 0.5 hours × $53.51) = $13,061.
- Total now: $101,455 + $2,568 + $13,061 = $117,084 a year.
- Delay cost (inferred, shown separately): quotes currently take three working days. If same-day quoting lifted the win rate from 30% to 33%, with a $2,500 average job and 40% gross margin, that is 1,440 × 3% × $2,500 × 40% = $43,200 of gross margin a year.
Do nothing, and improved
Assume volumes grow 25% over the next two years and admin time and errors grow with them. The improved state assumes enquiry capture is automated (re-keying time down 90%), a quoting tool cuts each quote from 45 to 15 minutes, payment reminders are automated (follow-up time down 75%) and the report builds itself (time down 90%). Error rates fall to 0.5% for enquiries and 1% for quotes. All are assumptions.
| Now | Do nothing (+25% volume) | Improved (+25% volume) | |
|---|---|---|---|
| Admin hours a year | 1,896 | 2,370 | 579 |
| Time cost | $101,455 | $126,819 | $30,982 |
| Error cost | $15,629 | $19,536 | $4,617 |
| Total a year | $117,084 | $146,355 | $35,599 |
The improved hours come from 48 + 360 + 36 + 19.2 = 463.2 hours at today's volume, × 1.25 = 579. The gap between doing nothing and improving is $110,756 a year and 1,791 hours, a little over one full-time equivalent. That is capacity your team can put into faster responses and follow-up, or growth you can absorb without adding admin headcount. Set it against the one-off and running cost of the changes, which is scoped separately.
Step 5: Prioritise by impact, feasibility and confidence
Priority score = impact (1 to 5) × feasibility (1 to 5) × confidence (1 to 5)
Impact covers cost, hours and customer effect. Feasibility covers data quality, systems involved, change required, and cost and time to deliver. Confidence runs from 5 for measured down to 1 for qualitative.
| Opportunity | Likely move | Impact | Feasibility | Confidence | Score |
|---|---|---|---|---|---|
| Automated enquiry capture | Integrate | 3 | 5 | 4 | 60 |
| Quoting tool | Build | 5 | 3 | 3 | 45 |
| Payment reminders | Improve | 2 | 5 | 3 | 30 |
| Automated weekly report | Automate | 2 | 4 | 3 | 24 |
The biggest saving isn't always first. Here, enquiry capture scores highest because it is cheap, low risk and well measured, and it creates the clean data the quoting tool will need later. Payment reminders are marked Improve because the accounting or job system you already pay for may handle them with configuration alone. Check that before building anything.
Two rules before you automate. Ask whether the task should exist at all, because removing it beats automating it. And fix a broken process before automating it, or you'll just make the same mistakes faster.
What to do with the number
Run your own figures through the admin cost calculator. If the total is material, the next step is deciding which moves to make and in what order. That is where business process automation, system integrations and an AI strategy and roadmap come in. The trades admin automation example shows job intake, quoting, follow-up and reviews worked through this way, and five automations every service business should have running covers the common starting points.
If you'd rather have it done properly across the whole business, the Altitude Diagnostic models every opportunity in these three states, labels each estimate's confidence, and reports capacity created rather than jobs removed. It may also recommend leaving some tasks exactly as they are.