Implementation and migration
A Guide to Moving from Spreadsheets to a Central HR System
Moving from spreadsheets to an HR system is not a file upload but a change in data ownership and operating process. A safe migration identifies the authoritative record, removes unnecessary data, maps fields, validates a small pilot, runs a short parallel check, and gives the company one clear cutover date after which the old sheet becomes read-only.
The migration backbone
Prepare
Data inventory, ownership, and cleanup rules
Validate
Small pilot, control totals, and user acceptance
Cut over
One date, parallel checks, and a closed old sheet
1. Assign a migration owner and success measures
One migration owner should manage dates and decisions, but that person cannot certify every field. Assign business owners for employee basics, compensation, leave, documents, and organization data. A technically successful import is not accepted until those owners validate it.
- ScopeWhich workflows move on day one, and which move later?
- SuccessHow many records, control totals, and user scenarios must pass?
- DatesCleanup, pilot, acceptance, cutover, and old-sheet closure.
2. Inventory the scattered sources
Beyond the main employee sheet, data may live in email attachments, personal folders, leave trackers, compensation files, and hiring documents. Record each source by its contents, owner, last update, and sensitivity—not only its filename.
When sources conflict, 'the newest file' is not enough. Define the authoritative source field by field: an approved employment record for start date, a signed-off carry-forward calculation for leave, or the latest valid compensation decision for pay.
- SourceWhere is it and who can access it?
- AuthorityWhich record wins when values conflict?
- RetentionWhat must move, remain archived, or be securely removed?
3. Clean before importing
A new system does not fix bad data. It makes it more visible. Deduplicate unique identifiers such as email or employee number, standardize statuses, use one date format, and distinguish an unknown value from zero or 'not applicable'.
- DeduplicationOne person does not appear under multiple spellings or emails.
- Controlled valuesDepartment, position, work type, and status use an agreed vocabulary.
- Dates and moneyDate format, currency, gross/net basis, and monthly/yearly period are explicit.
- Missing dataRequired gaps have owners and dates, and estimates are not presented as facts.
- Unnecessary dataPersonal data without a purpose is not moved merely because it exists.
4. Map fields and run a representative pilot
Maintain a mapping table with source column, target field, type, conversion rule, and example. A column called 'salary' is too ambiguous: currency, gross/net basis, and payment period need their own meaning.
Do not begin with everyone. Choose a small pilot representing different cases: new and long-tenured employees, managers, people with leave balances or compensation history, and former employees. Fix failures in the source or mapping rule. One-off edits in the target are hard to reproduce.
5. Validate totals and user scenarios
Matching row counts are not enough. Compare active/former employee counts, department totals, leave-balance totals, and key compensation units. Then ask real users to complete daily workflows.
- Record controlsExpected, skipped, duplicate, and failed record counts.
- Field controlsIdentity, employment, manager, leave, and compensation on sample employees.
- Access controlsEmployees and managers see only the data they should.
- Workflow controlsRequest, approval, change, and export scenarios complete.
6. Cutover, parallel checks, and rollback
Permanent entry into old and new systems creates two sources of truth. Announce a cutover date: new work starts in the central system, while the old sheet becomes a read-only archive. You can compare the first leave or payroll-preparation cycle in parallel, but enter new changes in one place only.
Define which error level delays launch, where the last clean source is held, and how a repeatable import runs again. Keep temporary migration files in access-controlled storage and remove them through the agreed secure process after the work ends.
7. Employee communication and the first 30 days
A login link is not an adoption plan. Explain why the process changed, where each action now happens, how employees review their details, and where to get help. Show managers their approval queue and team view separately.
- First weekAccess issues, wrong managers, and critical field errors are reviewed daily.
- First cycleLeave and payroll-preparation outputs are compared with source records.
- Day 30Manual touches, support requests, and incomplete data are measured.
- Old sheetNew entry is closed, access is limited, and retention and deletion dates are set.
About migration
It depends more on data conflict and source sprawl than headcount. A clean employee list can be piloted in days. Leave, compensation history, and documents across many sources can require several weeks. Plan around business-owner validation, not the upload date.
No. Separate data needed for operations or required retention from data kept only by habit. In some workflows, an approved opening balance is more reliable than importing every historical movement. Business and data owners should make that decision together.
Do not delete them without validating migration and retention needs. Move the source to a read-only, access-limited archive, stop new entry at cutover, and remove it through the authorized process when its retention period ends.
Compare record counts and control totals such as status, department, and leave balances. Inspect sample employee records field by field. Then complete workflows with an employee, manager, and process owner and record acceptance.