Planning
- Shifts and rotas
- Hour norms
- Assignments
- Cover
Executive summary
1. What the review found
Time recording differs from the other processes in one respect: its result is contested. The employee remembers what they worked, the manager remembers something else, and payroll is built on a third document. Any inaccuracy here turns not into an operational loss but into a conflict.
The review found that plan and actual exist as independent spreadsheets. The shift is planned in one place, actual time is captured in another, corrections go into a third, and the monthly totals are assembled by hand from all three. And the correction overwrites the data instead of creating a history.
From this follows the central defect: it is impossible to reconstruct how a final figure came about. Not because someone made a mistake, but because the process does not preserve the path from the original record to the total.
| Finding | Statement |
|---|---|
| Primary constraint | A correction overwrites data instead of creating history. The final figure exists; the path to it does not — so a dispute about hours is settled by authority rather than by a record. |
| Recommended first step | Create an original time record linking employee → shift → site → actual, with the data source, then move corrections into a workflow with a reason and a confirmation. |
| Expected effect | A verifiable history from the original record to the payroll export, exceptions surfaced the same day, and a period close that no longer needs manual reassembly. Quantitative effect is assessed after rollout. |
Areas reviewed
Actual time
- Start and finish
- Breaks
- Site attribution
- Overtime
Exceptions
- Absences
- Lateness and no-shows
- Corrections
- Overlapping shifts
Period close
- Hour approval
- Decision history
- Documents
- Payroll export
Starting position
2. Baseline capture of current work
The review starts by capturing one complete payroll period. For each operation we record who performs it, how often, which carrier holds the data, and whether the same data is moved again. Everything else in this document rests on this table.
| Operation | Who performs it | Frequency | Where the data lives | Re-keyed |
|---|---|---|---|---|
| Shift planning | Department manager | Weekly | Spreadsheet or a printout on the wall | Yes — the plan is duplicated in messages to staff |
| Communicating the rota | Manager | On every change | Messenger, phone call | Yes — the rota is kept current by hand |
| Recording the start of work | Employee | Daily | Logbook, timesheet, a mark with the manager | Yes — the mark is transferred into the summary sheet |
| Attributing work to a site | Employee or manager | Per shift worked | Memory, a separate note | Yes — the link is reconstructed during review |
| Recording breaks | Employee | Daily | Not recorded, or recorded approximately | Yes — break rules are applied by hand at payroll |
| Absence request | Employee | As needed | Verbally, a message, a paper form | Yes — the data moves into the rota and the timesheet |
| Absence approval | Manager | Per request | A reply in correspondence | Yes — the decision is not structurally linked to the request |
| Correcting hours | Manager | Regularly | An edit in the spreadsheet | Yes — the original value is overwritten with no history |
| Reconciliation before close | Manager and HR | Monthly | Comparing three sources | Yes — this is the main manual work of the period |
| Payroll data export | HR or bookkeeping | Monthly | A hand-assembled file | Yes — the totals are entered into the payroll system again |
What follows from thisIn ten operations out of ten the data is moved again. The eighth row differs from all the others: the correction does not merely duplicate data, it destroys the previous value. It is the only operation on the list that makes the process unverifiable after the fact.
Diagnostic panel
3. State of the operational system
Losses occur between plan and actual, between an exception and its handling, and at the period close where all of it is brought together by hand.
46/100average process maturityManual reconciliation risk high
10points of re-keyingCandidates for removal
6critical data handoversNeed an owner
2priority pilot flowsTime record + exceptions
Process maturity
Expert assessment on a 0–100 scale. Higher means a more resilient process.
Structure of operational risk
Distribution of recorded observations by type.
16observations
40% — scattered data32% — losses at handover28% — unclear ownership
The numbers on this demonstration panel are illustrative. In a paid review the assessments are derived from interviews, a sample of timesheets across a period, observation of the month-end close and an agreed 0–100 scale.
Process map
4. Current process and handover points
The chain describes the path of a single time record from plan to payroll. It is a process model, not a snapshot of any particular client system.
Shift plan
The shift is planned and communicated.
Assignment
The employee is attached to a shift and a site.
Actual
Start, breaks and finish are captured.
Exceptions
Lateness, no-shows and overlaps surface.
Absences
Requests travel an approval route.
Corrections
Changes carry a reason and a history.
Approval
The period’s hours are accepted by an owner.
Export
Approved data goes to payroll.
Risks at the handover points
| Handover | What can be lost | Consequence | Priority |
|---|---|---|---|
| Plan → employee | The current version of the rota, cover, a site change | The employee turns up in the wrong place or not at all | High |
| Actual → record | Precise time, breaks, site attribution | The hour exists but it is unknown what it belongs to | Critical |
| Exception → handling | The reason for lateness, a no-show, an overlap | The exception surfaces at payroll instead of on the day | Critical |
| Request → approval | The manager’s decision and its basis | The absence is approved verbally and never reaches the rota | High |
| Correction → history | The original value, the reason for the change, the author | The total cannot be verified after the fact | Critical |
| Period → payroll | The marker that the data is approved | The export is built from unapproved data | Critical |
Risk heatmap
| Area | Likelihood | Impact | Detection | Result |
|---|---|---|---|---|
| Actual time not linked to a shift and a site | High | High | When a dispute is examined | Critical |
| Hour correction without a reason or history | Medium | High | On audit | High |
| Absence surfaces after the shift has started | Medium | High | On the day | High |
| Export built from unapproved data | Medium | Critical | After payroll | Critical |
| Break rules applied by hand | High | Medium | At period close | High |
Root causes
5. Why this happens
Disputes about hours, a long month-end close and corrections after payroll share the same structural causes.
A. Plan and actual exist as independent spreadsheetsThere is no link between them at record level. Comparison happens once a month by a person, so a discrepancy survives to the end of the period and surfaces when correcting it is expensive.
B. The correction overwrites rather than addsAn edit in a spreadsheet destroys the previous value. The process loses the ability to answer "how did this figure come about" — and with it the ability to settle disputes with a record.
C. Rules are applied by handRounding, breaks, night and overtime hours are computed by a person from memory of the rules. Identical situations get different results depending on who did the sums.
D. There is no exception queue and no ownerLateness, a no-show and overlapping shifts create no task. They exist as a fact in the data that somebody ought to notice — and the duty to notice is assigned to nobody.
Decision principleIn time recording the value comes not from measurement precision but from verifiability. A system that records time to the second but allows edits without history is worse than a paper timesheet with a signature: it adds precision and removes evidential weight.
Responsibility zones
6. Who does what — and how much of it is moving data
Here the division of labour is unusually clear: some record the time, others check it, and a third group reassembles it for payroll. Below is where work is duplicated and where a role does not decide anything but moves data between spreadsheets.
| Role | What it does today | How much of that is moving data | Review observation |
|---|---|---|---|
| Employee | Turning up for a shift, marking time, requesting absence | Duplicated marking: logbook, a message to the manager, confirmation at reconciliation | The mark is made once in self-service; the role keeps the work itself |
| Department manager | Shift planning, cover, corrections, resolving disputes | A substantial share of the time is manual plan-versus-actual reconciliation and spreadsheet editing | Discrepancies arrive as a queue on the day; the role keeps the decision on the exception |
| HR | Collecting timesheets, absences, documents, preparing the period | The monthly assembly from three sources is the main work of the period | Totals are built from approved records; the role keeps documents and people decisions |
| Bookkeeping | Receiving data, payroll, corrections after payroll | Re-entering totals and clarifying discrepancies after the fact | The export arrives with an approval marker and a version |
| Company management | Resolving conflicts about hours and pay | Reconstructing what happened from correspondence and memory | The record history is available without reconstruction |
The organisational conclusion stays with the companyThe review shows which part of the work is moving data and which part is deciding. What happens with the time freed up — redistributing tasks, retraining people or changing the structure — is the owner’s call. One particularity matters here: the largest relief lands with department managers, and it goes back into managing people rather than into reporting.
Transparency works in both directionsTime recording is the only one of these processes where the employee is an interested party and not merely a performer. If they see their hours, balances and request statuses exactly as the manager sees them, the number of disputes falls before any rule is introduced. One-sided control, by contrast, produces resistance and workarounds.
Target operating model
7. States, boundaries and rules
The time record passes through explicit states. Each defines an owner, the required data and the condition for moving on. The key requirement: a transition does not destroy the previous state.
| State | Owner | Required information | Transition condition |
|---|---|---|---|
| Planned | Department manager | Employee, shift, site, planned hours | The rota is published and communicated |
| Captured | Employee | Start and finish, breaks, data source | Record created with its source stated |
| Checked | System | Comparison of plan and actual, application of the rules | Deviations placed in the exception queue |
| Exception handled | Manager | Reason, decision, author, time of decision | The decision is recorded as an event; the original value is preserved |
| Approved | Owner of the period | Accepted hours, handled exceptions | Period closed, data version fixed |
| Exported | HR or bookkeeping | Data version, approval marker | The payroll file is generated from the closed version |
Boundaries of automation
The system should determine
- Whether plan and actual diverge on a specific shift
- Whether break rules and maximum durations are met
- Whether one employee’s shifts overlap
- Who must handle an exception and how long it has been open
- Which data version went to payroll and when
The system should not decide
- Whether lateness is excusable
- Whether disputed hours are paid
- Whether disciplinary measures are taken
- Whether to approve absence when staffing is short
- Whether to change an employee’s hour norm
Rules, deadlines and escalation target
A rule without a deadline and an addressee does not work. Below is the minimum set that brings exceptions into handling on the day of the event rather than at the end of the period.
| Rule | Deadline | Escalates to |
|---|---|---|
| A time record cannot be saved without a shift and a site | At the moment of capture | Blocking rule, no escalation needed |
| No mark at the start of a shift | 30 minutes after the planned start | Department manager |
| Shift ended with no finish mark | By the end of the working day | Department manager |
| Overlapping shifts detected for one employee | At the moment of capture | Manager, then HR |
| Exception not handled | 2 working days | Manager, then HR |
| Absence request not decided | 2 working days | Manager, then the next level up |
| Period not approved by the cut-off date | The payroll cut-off | HR, then company management |
| Correction entered after the period was closed | At the moment of entry | Owner of the period, confirmation mandatory |
Priorities and sequence
8. What to do and in what order
The sequence is built so that each stage can be verified across one payroll period and does not require the next one to be finished.
| No. | Recommendation | Why now | Acceptance criterion |
|---|---|---|---|
| 1 | The original time record | Links employee, shift, site, actual and data source | Every hour belongs to a specific shift and site |
| 2 | Exception queue | Moves discrepancies from the end of the period to the day of the event | Lateness, no-shows and overlaps surface automatically and have an owner |
| 3 | Correction workflow | Replaces overwriting with history carrying a reason and an author | The original value is preserved and the change is traceable |
| 4 | Absence approval routes | Separates request, review and final decision | Every request has a route, a deadline and a recorded decision |
| 5 | Period close and a protected export | Prevents payroll being built from unapproved data | The export is generated only from a closed version carrying an approval marker |
| Later | Load-based planning and staffing forecasts | Useful once the actual data is reliable | Shift history is fit for planning across an agreed horizon |
Do not start with devicesTerminals, biometrics and geotags look like the answer but measure the entry point of a process that has no model. While corrections overwrite data, precision at the point of marking does not increase trust in the total. The record with history first, the means of capturing it second.
Rollout plan
These durations are indicative for a demonstration sample. A real estimate is formed after access to the shift and break rules, the agreements in force, the payroll system formats and the integration constraints.
| Stage | Duration | Outcome | Checkpoint |
|---|---|---|---|
| Period capture and record model | 2–4 weeks | States, roles, required fields, rules, exceptions | Management and HR approve the model and the rules |
| Time record and self-service | 4–6 weeks | Capture, link to shift and site, visibility for the employee | The pilot department runs with no parallel spreadsheet |
| Exceptions and corrections | 3–4 weeks | Automatic checks, queue, change history | Discrepancies are handled on the day of the event |
| Absences and approvals | 2–4 weeks | Requests, routes, decisions, reflection in the rota | Approval stops living in correspondence |
| Period close and export | 2–3 weeks | Hour approval, data version, payroll file | The period closes without manual reassembly |
Measuring the outcome
9. Metrics, reporting horizon and next step
This sample claims no numeric effect. In a real review the baseline is recorded first across two closed periods, then the metrics are compared after an agreed period of use.
| Metric | How it is calculated | How to capture the baseline | Direction |
|---|---|---|---|
| Share of shifts with anomalies | Shifts with a plan-versus-actual discrepancy / all shifts × 100 | Analysis of timesheets and rotas across two periods | Down |
| Time to handle an exception | From the discrepancy appearing to a recorded decision | Manual timing on a sample of disputed cases | Down |
| Manual corrections per period | Number of hour edits per hundred shifts | Counting spreadsheet edits across two periods | Down |
| Changes after the period closed | Number of corrections made after handover to payroll | Comparing versions of the files handed over | Down |
| Time to prepare the payroll export | Working hours from period end to a finished file | Measured on the last two closes | Down |
| Disputes about hours | Number of employee queries about hour calculation per period | Counting queries over a quarter | Down |
Horizon: from the shift to the annual report
A review does not end at the payroll export. Working-time data has to reach management and annual reporting without being entered again — otherwise staff cost by site and project remains an estimate.
| Level | What must arrive without manual assembly |
|---|---|
| Day | Marks, deviations from plan, open exceptions |
| Week | Rota attainment, overtime, undecided requests |
| Month | Approved hours, absences, payroll export with a version |
| Quarter | Working-hours volume by department and site, structure of overtime |
| Year | Staff cost by site and project on one data model |
If staff cost per site is only approximately known, the cause is usually not payroll but that the hour is not attributed to a site at the moment of capture.
Questions to settle before development
- What rounding, break and maximum-duration rules apply, and what are they based on?
- What requirements for recording and retaining working-time data apply in your jurisdiction and to your form of employment?
- Who may enter a correction and who confirms it?
- What location data may be collected, on what basis, and for how long may it be stored?
- Is involvement of an employee representative body required, and in what form?
- What format does the payroll system accept, and how are retrospective corrections handled?
Recommended next decisionCapture one closed period in one department: rotas, timesheets, correspondence about corrections, absence requests and the finished payroll file. Output: a data model, a rule set, an exception list, a role matrix and a pilot scope for one department.
This document demonstrates the structure and level of detail of the deliverable of the “Process review” service. It is based on the mechanics of a delivered, anonymised platform for working time recording, is not legal, employment-law, tax or data-protection advice, and contains no confidential client information. Maturity scores are illustrative. Requirements for working-time recording, data retention and the involvement of employee representative bodies differ by jurisdiction and must be checked separately before development starts.