The answer
Largely. Collecting timesheets, checking hours against schedules, adding commissions and entering starters and leavers are mostly automatable today. In our model about 70% of payroll preparation time can go to software. What stays human is answering pay queries and the final review before money moves, because payroll mistakes reach people's bank accounts.
The tasks, and how much of each is automatable today
Six tasks make up payroll preparation. Collecting timesheets and checking hours are about 85% automatable. Entering starters, leavers and pay changes is about 80% automatable. Adding commissions and bonuses is about 70% automatable once rules are written down. Answering pay queries is partly automatable. The final review stays almost entirely human.
| Task | Min / week | Automatable |
|---|---|---|
| Collect timesheets and approvalsDigital timesheets with automated reminders and manager approval links. Late or disputed timesheets still need a person. | 90 | |
| Check hours against schedules and overtime rulesRules catch missing days, overtime and breaks. Unusual shifts still need context from the manager. | 60 | |
| Add commissions, bonuses and deductionsCalculated from CRM deals once the commission plan is written as rules. Disputed deals and one-off bonuses need a decision. | 45 | |
| Enter starters, leavers and pay changesPulled from the HR record when a change is approved. Final-pay calculations for leavers still need checking. | 30 | |
| Answer pay queries and fix errorsAI can answer 'when is payday' and send payslips. Questions about why pay is different need a person. | 45 | |
| Review and approve the payroll runChecks compare totals with last run. Someone accountable must approve before money moves. | 20 | |
| Whole process | 290 |
Hours saved per week
For a business with 25 hourly and salaried staff, paid every two weeks, payroll preparation takes about 5 hours a week on average by hand. Automating the tasks above returns about 3.5 hours a week, roughly 170 hours a year. The time left is the final review and the queries that follow each run.
Modelled on: A 25-person business with a mix of hourly and salaried staff, some on commission, paid every two weeks, prepared by a bookkeeper with input from HR.
- Figures are a weekly average across the two-week pay cycle.
- About half the team submits timesheets that need approving.
- Five or six staff earn commission calculated from CRM deals.
- One or two starters, leavers or pay changes happen most cycles.
Tools that do the job
Payroll software runs the payroll; the preparation happens around it. Timesheets can come from a scheduling tool or Google Sheets. Commission data comes from HubSpot or another CRM. n8n or Make pull everything together, check it against rules, and send exceptions to Slack before the bookkeeper loads the final figures.
- Google Sheets
- Simple timesheet and commission tables when no scheduling tool is in place.
- HubSpot
- Source of closed deals and amounts for commission calculations.
- n8n
- Collects timesheets, applies rules, calculates commissions and builds the exceptions list.
- Slack
- Sends timesheet reminders, manager approval requests and the exceptions list.
- Xero
- Payroll and accounting in one place in the regions where Xero offers payroll.
- QuickBooks
- Alternative with payroll in supported regions; receives the final figures.
A simple workflow
The workflow runs a few days before each pay date. It collects approved timesheets, compares hours with schedules and overtime rules, calculates commissions from closed deals, applies starters, leavers and pay changes from HR records, and sends one exceptions list for a person to clear before the payroll is approved.
- TriggerRun four working days before each pay date.
- CollectGather approved timesheets; remind anyone missing and their manager.
- CheckCompare hours with schedules and overtime rules. Calculate commission from deals closed in the period.
- ApplyAdd starters, leavers and pay changes approved in the HR record since the last run.
- ReviewSend one exceptions list and a comparison with last run's totals to the approver.
Keep these human
- Final approval of the payroll run. Errors reach people's bank accounts and are slow to reverse.
- Pay queries about why someone's pay changed. These are trust conversations.
Where it goes wrong
- Unwritten commission plans. If the plan lives in someone's head, automation gets it wrong. Write the rules down and test them on last month's deals.
- Late changes. A pay rise approved by email after the run is prepared gets missed. Make HR records the only route for changes.
- No comparison check. A run that differs a lot from the last one should stop for review. Compare totals every time.
What most guides miss
Most payroll guides compare payroll software. But most errors start before anything reaches the payroll system: a missing timesheet, an unapproved overtime shift, a pay rise agreed by email. Automating timesheet collection, approval and rule checks upstream removes more of the corrections EY counts each pay period than switching payroll provider does.
The evidence
- One in five payrolls contains errors, each costing $291 on average, and the average organisation makes 15 corrections per pay period.EY (via Business Wire), EY survey: Payroll errors average $291 each, impacting the economy (2022)
- Processing data has a technical automation potential of 69%, applying expertise 18%.McKinsey Global Institute, A future that works: Automation, employment, and productivity (2017)
“There is surely nothing quite so useless as doing with great efficiency what should not be done at all.”