01 the business
Restaurant group
8 venues, 300 hourly staff
Before each pay run, the payroll officer scans timesheets for missed clock-outs, unapproved overtime and wrong rates. Errors found after payday mean awkward corrections and unhappy staff. Penalty and overtime rules are easy to get wrong.
02 before
The problem
Before each pay run, someone scans timesheets for missed clock-outs, unapproved overtime and wrong pay rates. Errors found after payday mean awkward corrections and unhappy staff. Award, overtime and holiday rules make it easy to get wrong.
- Checks are manual and rushed.
- Missed clock-outs inflate hours.
- Penalty rates are applied inconsistently.
03 the build
What the AI does
- Step 1: Pulls timesheets, rosters and approved leave from Deputy, Tanda or your time clock.
- Step 2: Compares hours worked with the roster and flags missing breaks, long shifts and clock-out gaps.
- Step 3: Checks pay rates and penalty rates against the pay rules you have written down for each role.
- Step 4: Sends each manager a short list of entries to confirm, with the reason each one was flagged.
- Step 5: Prepares the approved timesheets for import into Xero Payroll, Gusto or Employment Hero.
paused: waiting for a person to approve
How it was put together
- Pulled timesheets, rosters and approved leave from the time and attendance system.
- Compared hours with rosters and flagged missing breaks, long shifts and clock-out gaps.
- Checked pay and penalty rates against the written rules for each role.
- Sent managers a short list of entries to confirm, then prepared approved timesheets for payroll.
04 human in the loop
Where a person stays in charge
Managers confirm or correct each flagged entry, and the payroll officer approves the final pay run. The checks never change pay on their own.
- Managers confirm or correct each flagged entry.
- The payroll officer approves the final pay run.
- The checks never change pay on their own.
05 rollout
From kickoff to live in 5 weeks
- Weeks 1–2Write down pay rules per role and connect systems.
- Weeks 3–4Check two past pay runs and compare with corrections made.
- Week 5Run before every pay run.
06 after
What changes
Pay runs go out with fewer errors and fewer corrections afterwards. Managers fix problems while the week is still fresh.
Assume 300 staff paid fortnightly, with 3% of timesheets wrong and each error taking 30 minutes to fix after payday.
300 × 3% = 9 errors a pay run. 9 × 30 minutes = 4.5 hours per run, or about 117 hours a year, plus the trust lost with staff.
What to measure
- Pay corrections after payday.
- Flagged entries per pay run.
- Payroll officer hours per run.
- Staff pay queries.
07 evidence
What the research says
Gartner research put the average cost of poor data quality at $12.9 million a year per organisation.
Gartner, Data quality: why it matters and how to achieve it (opens in a new tab) · 2020
In 1,921 fraud cases, expense reimbursement schemes made up 13% of cases with a median loss of $50,000, and frauds at organisations with under 100 staff had a median loss of $141,000.
08 lessons
Mistakes to avoid
- Unwritten pay rules. Write them down first.
- Auto-correcting pay. People must confirm.
- Checking too late. Run before managers approve.
09 questions
Common questions
- Which payroll systems does it support?
- Xero Payroll, Gusto and Employment Hero are common, with Deputy or Tanda for time.
- Does it understand award or union rules?
- It applies the rules you write down. We help turn agreements into clear rules, checked by your payroll adviser.