Field notes · Time Tracking · 9 minute read
From biometric punch to payroll line: getting the integration right in SuccessFactors Time Tracking.
Most payroll disputes don't originate in payroll. They originate in time. Here is how to engineer a path from a fingerprint reader on a shop floor to a clean payslip line, and why getting it wrong creates the audit findings nobody wants.
Why 80% of payroll disputes start here
Look at the root cause distribution of any mature SAP payroll programme and a pattern shows up. Engineering errors in the payroll engine itself are rare. Payroll vendors spend decades refining the calculation logic. Configuration errors in pay element setup are more common but get caught in parallel runs. The dominant cause of post-go-live disputes, across the customers we've seen, is upstream: the time data that arrives at the engine is wrong.
That is not a payroll problem. It is an integration problem. A biometric reader produced a punch. Something happened between the punch and the payroll calculation. By the time an employee disputes a payslip line, the trail back to the original event is cold.
Three reasons this is structurally hard.
Devices are noisy
Fingerprint reads fail, badge readers double-scan, employees forget to clock out. The raw event stream is not the truth.
NoiseMapping is multi-step
A punch becomes a shift, a shift an attendance record, then a time account input, then a payroll line. Each step can drift.
DriftReconciliation happens late
Most integrations reconcile at month-end. By the time HR sees an anomaly, the employee has left for the day.
LateThe mapping problem: punches to schedules to time accounts
A clean integration into SuccessFactors Time Tracking makes four decisions on every event.
Whose punch is this?
Device-side identity must map to a SuccessFactors person. Badge IDs and templates are not SF identifiers, so the mapping has to be authoritative and current.
Against which schedule?
The work schedule assignment on that day tells us whether the punch is a shift, overtime, a rest day, or a make-up. A punch is meaningless without it.
Which time type?
Regular, overtime, sick, on-call and allowance-eligible hours each create a different time account input. The mapping is evaluated at ingest, not patched at month-end.
Which approval state?
Some events need manager approval, some do not. The integration routes accordingly and pauses the downstream flow when approval is missing.
The day we stopped reconciling punches in a spreadsheet on the 28th, we got our team back.
Resolving missing punches without manual intervention
Missing punches are the single largest source of dispute volume. An employee forgets to clock out. The device doesn't read their fingerprint cleanly. The badge is forgotten at home. The integration sees the in-punch but never sees the matching out-punch.
Three patterns we have seen fail.
Default to the scheduled time
Assume the missing punch happened at the scheduled start or end. Fails the first time someone works overtime, leaves early, or arrives late. Creates lines that look correct but are not.
FailsHold the record indefinitely
Wait for a human to resolve. Works for a small team. At 2,000 employees it builds a queue HR cannot clear.
FailsDrop the record
Ignore unmatched events. The employee loses pay, then disputes the payslip.
FailsWhat works: surface the missing punch the same day, route it to the employee's manager for resolution, and pause downstream processing for that record until resolution happens. The integration has to know how to wait. It has to know when waiting becomes a blocker. And it has to close the loop before the time evaluation window closes.
The Time Connector approach
Time Connector is the solution we ship for this problem. The architecture is intentionally narrow: three responsibilities, and nothing more.
Exhibit · Integration flow
Time Connector turns noisy device punches into clean payroll inputs.
Every device event becomes a canonical time account input, posted to SuccessFactors through standard OData.
Biometric terminals
Fingerprint, badge, face reads
Time Connector
Normalise and apply rules
SuccessFactors Time Tracking
Time accounts to payroll line
Biometric terminals
Fingerprint, badge, face reads
Time Connector
Normalise and apply rules
SuccessFactors Time Tracking
Time accounts to payroll line
What the channel resolves on every event
Source: Raptors Time Tracking practice, May 2026.
Normalise events at ingest
Every device event is translated to a canonical shape: person, timestamp, location, type, raw payload. The pipeline operates on the canonical form, not per-device dialects.
Apply per-site rules
Each site's schedule patterns, grace windows, overtime treatment and break policy are codified once. New rules are configuration changes, not code.
Post through standard OData
Records arrive in SuccessFactors Time Tracking as normally-shaped time account inputs. No middleware, no point-to-point custom layer. Time evaluation runs unchanged.
That last point is load-bearing. The payroll engine sees inputs that look the way SAP's standard expects them to look. Time evaluation runs unchanged. The downstream is clean.
Three customer outcomes
Customers anonymised; figures verified pre-publication.
KSA retailer · 4,200 employees · 180 stores
Payslip queries fell to 1.1%
Queries dropped from 6.4% of payslips in the first month after go-live to 1.1% in steady state. The HR team redeployed two FTEs from reconciliation to onboarding.
VerifiedConstruction group · 2,800 site workers · 14 projects
Cost allocation within 24 hours
Project cost allocation once lagged payroll by two weeks because attendance arrived late. It now runs within 24 hours of payroll close.
VerifiedF&B chain · 1,600 employees · 90 outlets
Month-end in half a day
Month-end reconciliation took five working days and four people. It now runs in half a day, by one person.
VerifiedWhere to go next
If your time-to-payroll path involves a spreadsheet, or a screen-scraping bot, or a middleware layer that nobody wants to touch, the conversation starts with two artefacts. Your current device inventory and a month of payroll query volume. We can map them against a Time Connector deployment in 30 minutes.
See the Time Connector product page for the device catalog and architecture detail, or talk to our team directly.
Next step