Home · Solutions · HR & people
Solution · HR & peopleEvery hour, absence and bonus is checked before payroll runs, not after payday
Payroll inputs consolidated and checked before the run
Time, absence, overtime, bonus and contract data are gathered from every source, checked against the rules, and disputed items go to managers in Teams before the payroll file leaves the building.
Executive summary
Payroll errors are found by the employee on payday, not by the team that built the file.
We build the layer payroll has never had: one place where a month of inputs is assembled, checked and evidenced before anything is calculated.
Consolidation stops being an assembly job: clean records pass through untouched and payroll works on exceptions.
payroll system or provider upload; restricted payroll library on SharePoint; audit pack under a Purview retention label
Business problem
Payroll operations
Payroll is the last stop in a chain it does not control. Hours come from terminals at fourteen depots, absences and contract changes from the HR system, overtime approvals from schedulers, allowances and bonuses from depot managers with their own workbooks. Each source has its own owner and its own idea of a complete month.
Most of the work is not calculation, it is chasing. The depot that has not sent its bonus file is chased by email, then by phone, then through its regional manager. A contract annex signed on the twenty-eighth arrives after the period closes; an overtime line turns up with no approval. The deadline does not move: Polish labour law requires monthly pay within the first ten days of the next month.
At scale the pattern repeats rather than resolves. Every new depot, agency intake and seasonal peak adds an input owner and another template, and checking turns into sampling because there is no time to verify every record against its approval. What is not checked is found by the employee.
The control question is separate and usually unasked. The consolidation lives in a workbook that travels by email and is opened by whoever is covering. Nothing records which version produced the file sent to payroll, who changed a rate, or why an allowance was paid twice.
How it works today
This is what we usually find before automation, whatever the systems are.
- PersonPayroll exports the period's clockings and pastes them into the consolidation workbook, one sheet per depot
- PersonA second export brings absences and contract changes, matched on personnel number with the mismatches repaired by hand
- WaitingDepot managers are asked by email for overtime and bonus files; several arrive after the cut-off
- PersonBonus files come in four templates, so amounts are retyped and rounded by whoever is typing
- Risk of errorChecking is done on a sample: outliers are noticed, a missing approval or a night shift counted twice is not
- SystemThe consolidated file is uploaded to the payroll provider, usually as the last task of a long day
- Risk of errorPayroll output is compared with the input file on totals only, so individual errors survive to the payslip
- WaitingCorrections wait for the next run, with contributions and tax recalculated by hand
Why the current process costs more than it appears
Behind every exception is an hour nobody logged.
- Chasing is unbudgeted work that nobody records: a chain of emails, a phone call, a follow-up two days later, none of it visible in a process cost.
- Sampling produces comfort rather than coverage. Outliers are caught, quiet errors are not, and they surface in the most expensive place available: the payslip.
- A correction is never one correction. It reopens contributions and tax, creates a second entry in the following run, and consumes a conversation with the person and their manager.
- Late inputs push payroll into overtime exactly where accuracy matters most, which is how tired people generate next month's corrections.
- The workbook has an owner and a deputy. Its rules live in formulas rather than in a document, so handovers are slow and nobody can state which checks actually ran.
Cost of inaction
The month has a shape and it does not change: the same six days are consumed, the same depots send the same files late, and the same share of records enters the payroll file without meeting an approval. Each new depot adds an input owner, absorbed the only way the consolidation can: more hours inside the same window.
Two risks grow quietly behind that. A workbook whose logic exists only as formulas cannot be handed over, audited or reproduced. And pay data that travels as attachments between fourteen depots has an access list nobody can describe, which stops being theoretical the first time an auditor or a subject access request asks who saw what.
A plausible organisation with realistic proportions. The figures are there to be recalculated on your data; they are not a client result.
A European logistics operator, 3,200 employees across fourteen depots and a central warehouse, mostly shift workers on hourly and mixed contracts. An HR system holds contracts and absences, a separate system holds clockings, and payroll is run by an external provider. Microsoft 365 E3.
3,200 payroll records a month, one per employee and period: clockings from the terminals, around 400 absence events, roughly 900 overtime and allowance lines, 60 to 90 contract changes and fourteen depot bonus workbooks.
Two specialists consolidate into one workbook with a sheet per depot: exported, pasted, matched on personnel number, repaired by hand. Missing files are chased by email and whatever checking fits the time left is done on a sample.
About 2.5 minutes of consolidation and checking per record, compressed into the six working days between period close and the file deadline. Roughly one record in eight needs a question answered outside payroll.
Robots collect each source on a schedule, assemble one normalised record per employee and period, and run the rule set on every record. Discrepancies reach the owning manager as a card in Microsoft Teams with a deadline, and payroll output is reconciled back to the inputs after the run.
In the modelled case, manual consolidation falls to the exception share, the discrepancy conversation spreads across the whole period instead of the final two days, and corrections that begin as a missing or unapproved input largely stop. All of it is built from the assumptions of the illustrative model, not measured at a client.
Proposed solution
We build the layer payroll has never had: one place where a month of inputs is assembled, checked and evidenced before anything is calculated. Robots read each source on a schedule, take a dated snapshot into a restricted payroll library, and assemble one record per employee and period holding hours, absence days, overtime, allowances, bonus lines and the contract terms for each day.
The rules are the actual product. Booked hours are compared with the contract in force on that date, overtime with its approval, absences with the work calendar and the leave balance, bonus lines with their limits and plan. Duplicates are removed, joiners and leavers pro-rated, records for people who have left the HR system stopped rather than paid. Each rule has an owner in HR or finance, a definition and a version.
Discrepancies become a conversation with a deadline instead of an email thread. The depot manager receives one card per case in Microsoft Teams: the employee, the period, the value in dispute, the rule that fired and what is needed. Answers write back into the record, and anything unanswered at the cut-off escalates and shows on the readiness summary.
UiPath Orchestrator queues, time triggers, folder-level RBAC and audit logging; UiPath Integration Service connectors for Microsoft OneDrive & SharePoint (files, lists, Excel Online) and Microsoft Teams; UiPath Action Center tasks completed inside Microsoft Teams; Microsoft Purview sensitivity labels, data loss prevention and retention labels
The collectors, the normalised payroll record, the rule set with owners and versions, the discrepancy dialogue with its scopes and deadlines, the readiness summary, the consolidated file, the reconciliation of output against inputs, the audit pack and the access model
The HR and time and attendance systems through the vendor's API where one exists, otherwise a scheduled secure export; upload to the provider's portal where no interface is offered
How the automated process works
- AutomationOn a fixed schedule, robots collect the period's clockings, absences, contract changes, overtime and allowance lines into a dated snapshot in the payroll library
- AutomationOne record per employee and period is assembled from the snapshots, carrying the contract terms valid on each day, not the contract valid today
- SystemThe rules run on every record: contract against booked hours, overtime against approval, absence against calendar and balance, duplicates, component limits, joiner and leaver pro-rating
- PersonEach discrepancy reaches the manager who owns it as a card in Microsoft Teams: one employee, the rule that fired, the value in question, a deadline
- AutomationAnswers write back into the record; anything unanswered at the cut-off escalates to the HR business partner and shows on the readiness summary
- SystemThe consolidated file is generated in the provider's format, checked against control totals per depot, and released by the payroll lead
- AutomationAfter the run, payroll output is reconciled line by line against the inputs and the audit pack of sources, rules, decisions and timestamps is filed under a retention label
Human-in-the-loop model
Automation handles
- Collection of clockings, absences, contract changes, overtime and bonus files on a fixed schedule
- Assembly of one record per employee and period, with the contract terms valid on each day
- The rule set on every record: limits, approvals, duplicates, calendar and contract consistency, pro-rating
- File generation, the reconciliation of payroll output against inputs, and the audit pack
People decide
- Every discrepancy the rules raise, answered in Teams by the manager who owns that data
- Overrides outside the rules, approved by a payroll lead with a reason stored against the record
- What happens to inputs still missing at the cut-off: pay as booked, hold, or carry forward
- The rules themselves, their limits and thresholds, owned by HR and finance and versioned
Before and after
Systems and integrations
Every entry can be checked in vendor documentation. The evidence class is stated next to each one.
Inputs
- time and attendance system (clockings, shift plans)
- HR system (absences, contracts, personal data)
- depot bonus and allowance workbooks on SharePoint
- approved overtime records
Automation layer
- UiPath Orchestrator
- UiPath Robots
- UiPath Integration Service
- UiPath Action Center
Target systems
- payroll system or provider upload
- restricted payroll library on SharePoint
- audit pack under a Purview retention label
Human touchpoints: discrepancy cards in Microsoft Teams; Action Center exception queue; readiness summary in the payroll channel
Technologies used
collect each source on schedule, queue every record, run the rules, retry, log and audit
Areads depot workbooks, writes the consolidated file, posts discrepancy cards and the readiness summary
Amanagers answer discrepancies and payroll clears exceptions without leaving Teams
Adepot bonus and allowance templates in, the consolidated payroll file out
Arestricted payroll library holding snapshots, the consolidated file and the audit pack
Asensitivity labels, data loss prevention, retention and the tenant audit record for payroll data
Asource of hours, absences and contract terms, read by vendor API or scheduled export
CIllustrative economic model
The arithmetic is open, so it can be argued with.
Nothing below was measured at a client; the figures are an illustrative shape for a payroll of this size. The 2.5 minutes per record is a conservative blended average across records that pass untouched and records needing matching, a question and a re-check. €27 is a fully loaded hourly cost for payroll administration in Central Europe. We model capacity released, not headcount removed.
Run the numbers on your data
An illustrative estimate from your own inputs. It models released capacity; it is not a promise of savings.
Business benefits
- Consolidation stops being an assembly job: clean records pass through untouched and payroll works on exceptions
- Discrepancies reach the person who can answer them on the day they appear, not two days before the file is due
- Errors are caught before the payment run, so the payslip is not the employee's first sight of the problem
- Retroactive corrections fall, and with them the recalculations of contributions and tax behind each one
- New depots, agency intakes and seasonal peaks are absorbed by adding a source and a rule, not overtime in payroll
- Employees receive a payslip they can reconcile against their own calendar, the cheapest way to reduce queries
The management view
- Readiness becomes a fact: the payroll lead sees which depots are complete and which discrepancies are open
- The pay rules become an asset the company owns, applied identically to everyone, instead of formulas inside one workbook
- Payroll stops depending on two people who know where the exceptions hide, so holidays and resignations stop being risks
- Every run leaves an audit pack that answers a labour inspection or a works-council question without reconstruction
Board-level KPIs
Security and governance
Where the data sits and who can see it.
- Robot identities are separate from people: each collector runs under its own account, read rights limited to the fields its rules need, secrets held in the credential store configured for your tenant, and no personal login reused.
- The automation runs in its own Orchestrator folder, visible only to the payroll team and the automation administrators, so pay data never enters a shared queue.
- The payroll library is a separate site with one named security group, a Microsoft Purview sensitivity label restricting what a recipient may do with a file, a data loss prevention policy that stops payroll files leaving by mail or chat, and a retention label on the audit pack. That site is in your Microsoft 365 tenant, and the collectors writing to it run on a UiPath Automation Cloud tenant in the EU region.
- Managers see only their own people: a card carries one employee's line and the rule that fired, never a depot pay list, and absence cases show the fact and approval status, never a medical reason.
- Every step is recorded: source read, rule fired, who answered, what changed, who released the file. Release is a separate role from answering discrepancies, and the Orchestrator and Purview audit logs answer the auditor.
Why now
The payment date is set by law, not by capacity: monthly pay must be paid within the first ten days of the following month, so every late input compresses the same window, at a modelled €3,600 a month of consolidation effort.
Workforce structure is moving the wrong way for manual consolidation: shift patterns, mixed contracts, agency intakes and depot allowances multiply the input owners, and each new site arrives with its own template.
The controls are configuration rather than a build: connectors for SharePoint, Excel Online and Teams, Action Center tasks answered in Teams, and Purview labels, data loss prevention and retention.
Relevant executive roles
Pay accuracy is the most visible promise HR makes, and it depends on what fits into six days of checking
Personnel cost becomes traceable to approved inputs, and the correction tail that distorts each month shrinks
Depot managers answer one card about one employee instead of assembling an unverifiable workbook
Pay data stops circulating as attachments and moves into one labelled library with an access record
Common questions and objections
Then it delivers a scheduled export to a controlled location and the robot collects it there. The rules, checks and Teams dialogue are identical, and the interface can be upgraded later.
A scoped read account with a complete audit trail is a narrower exposure than a workbook emailed between fourteen depots. Access shrinks to one named group, every read is recorded, and Purview labels and data loss prevention stop the file leaving the tenant.
A card holds one employee, one rule and one question with a deadline, and it escalates by itself. The difference from email is that the readiness summary shows which depots are complete, so a delay is visible while it is fixable.
When this is not the right solution
- A single site with a few hundred salaried people and one source of hours, where a firm cut-off is cheaper
- Pay rules that are still unsettled, with a collective agreement under renegotiation or allowances agreed case by case
- Clocking data that is not trusted at source, from missing punches or shared badges, where the fix belongs at the terminal
A question for the next management meeting
Without opening anybody's spreadsheet, for how many of last month's payslips could we prove that every input behind them was complete and approved?
Implementation approach
Delivery runs in stages, so it can be stopped at any point.
We deliver
- A source-by-source review of what payroll receives today: format, owner, timing and lateness
- Collectors for each system, with dated snapshots into a restricted payroll library
- The normalised payroll record and the rule set, written down, versioned and owned in HR and finance
- The discrepancy dialogue in Microsoft Teams: scopes, deadlines, escalation and the readiness summary
- File generation in your provider's format, the reconciliation of output back to inputs, and the audit pack
- A parallel run on closed periods, then a controlled first live run under payroll supervision
We need from you
- Three closed pay periods: source files, the file sent to payroll, the corrections after it
- The pay rules as actually applied: limits, allowances, rounding, approvals and the owner of each
- Read access for the robot accounts in the HR and time systems, and a named input owner per depot
- Who may see payroll data, which we implement as groups, labels and folder permissions
Stages
Discovery
Sources, formats, owners, deadlines and the real exception categories
Rules
The rule set written and agreed with HR, finance and the works council
Build
Collectors, payroll record, checks, the Teams dialogue and file generation in your tenant
Parallel run
Closed periods rebuilt and compared with what payroll actually produced
Go-live
One live period with the payroll team supervising every exception, then hypercare
Optimisation
Tuning from recurring exceptions, extension to further depots or countries
Departmental. Effort follows the number of input sources and depots, how far the pay rules are already written down, and whether the systems offer an interface or only an export.
Payroll closes on the sixth working day; the depot's bonus file arrives on the seventh.
Send us one closed pay period: the source exports, the depot bonus files and the corrections that followed. We return the rule set those corrections imply and a first estimate of how many would have been caught before release.
Rebuild one pay period with usThe neighbouring process usually has the same problem
You approve the hours in one place and pay the agency invoice in another, and nothing checks that the two agree.
View solution HR & peopleJoiner, mover, leaver: access on day one, revoked same dayNew starters wait days for access; leavers keep theirs for weeks. Both are the same missing handover.
View solution Finance & accountingExpense reports audited automatically before reimbursementStop paying nine expense claims in ten on trust because nobody has time to check them against policy.
View solution HR & peopleShift cover filled in minutes, with the rules checkedA sick call should not cost a supervisor an hour of phone tag and a payroll correction.
View solution Case studyEmployee data mass updateHundreds of data changes done quickly, safely and error-free.
View case study Case studyDriver settlementNo more manual document collection and spreadsheets.
View case studyIndustries we deliver this in most oftenManufacturing & industryTransport & logisticsRetail & e‑commerceShared services