
Most HR system implementations do not fail at go-live. They fail weeks earlier, in a workshop nobody scheduled, when a team decides to recreate the process it already has inside a new platform.
That pattern repeats because it is comfortable. The software is capable, the project plan is detailed, the budget is approved, and the organisation still ends up with expensive licences and the same manual workarounds it had before. The system was implemented. The work was never redesigned.
If you are leading an HRIS project in a UK business of 50 to 1,000 people in 2026, this checklist sets out the seven phases that decide the outcome, the compliance dates your configuration now has to handle, and the failures that are predictable enough to prevent.
What an HRIS implementation actually involves
An HRIS implementation is the work of moving your people processes, records and reporting onto a new system, then getting managers and employees to use it as the normal way of working. It covers work design, data governance, configuration, testing, payroll, adoption and support.
It is not a software installation with a training session attached. Treat it as a work design project that happens to end in a system, and the sequence below makes sense. Treat it as an IT project, and every awkward process you run today becomes a configuration request.
The seven phases at a glance
A phase is only complete when you can evidence the outcome in the middle column. Attendance at meetings is not evidence.
| Phase | What you should be able to evidence | Where it usually goes wrong |
|---|---|---|
| 1. Work design and discovery | A documented decision on which processes stay, change or stop | The current process is mapped and relabelled as requirements |
| 2. Selection and fit | A scored shortlist tested against your hardest real scenarios | The demo is judged on interface polish |
| 3. Data governance before migration | Named owners, agreed definitions, one authoritative source per field | Migration is treated as a copy and paste exercise |
| 4. Compliance configuration | Each rule configured, tested and evidenced against its live date | Compliance is assumed to arrive with the vendor template |
| 5. Testing and parallel payroll | Payroll matches for two cycles and edge cases are proven | Only the happy path is tested |
| 6. Adoption | Managers finish real tasks in the system without a workaround | Training is a launch webinar rather than a change in routine |
| 7. Go-live and review | Data quality, usage and service levels measured at day 90 | The project team disbands on day one |
The dates your configuration has to survive
The Employment Rights Act 2025 is being switched on in stages, and each stage lands in system configuration rather than in a policy document. As of November 2026, the position is this.
Already live, and therefore already a live configuration risk:
- Statutory Sick Pay with no waiting days and no Lower Earnings Limit, since 6 April 2026
- Day one paternity leave and unpaid parental leave, since 6 April 2026
- A six-month tribunal claim time limit for most claims, since 1 October 2026
- A duty to take all reasonable steps to prevent sexual harassment, and not to permit harassment by third parties, since 30 October 2026
Landing on 1 January 2027, which means inside any project that is still in flight:
- The unfair dismissal qualifying period falls to six months, with uncapped compensatory awards
- The duty to inform workers of their right to join a trade union
Two consequences follow. First, an implementation that runs across the new year has to be configured for the rules that are live today and tested for the rules that arrive mid-project. Second, the tests that matter prove the rule behaves correctly on the date it applies, not merely that a field exists.
If your requirements are not yet written, read our guide to the Employment Rights Bill and HRIS readiness first. The compliance scope is wider than most requirement documents assume.
Phase 1: Redesign the work before you configure anything
Discovery is where most of the value is created or lost. The temptation is to interview the team, document the current process and hand it to a vendor as requirements. That approach guarantees you buy the system you already have.
Take each core process (joiners, movers, leavers, absence, pay changes, performance, reporting) and answer four questions:
- Who is trying to achieve what, and which decision does the output support?
- Which steps exist only because the current system cannot do something?
- Which approvals add control, and which add delay?
- What would this process look like if it took half the time and produced data we trusted?
Work design comes first for a practical reason. Configuration choices are cheap before they are built and expensive afterwards. Deciding that a three-stage approval becomes a single check is a workshop conversation. Deciding it after go-live is a change request, a retest and a difficult conversation with the managers who just learned the old version.
Name a business owner for each process, and bring finance, payroll and IT into the room. Implementations stall because a decision is waiting for someone who was never invited.
We run this sequence in our own work redesign projects, and the three pre-implementation steps that matter most are covered in 3 must-do steps before any HRIS implementation.
Phase 2: Selection that survives the demo
Selection fails when the shortlist is built on feature lists and the decision is made on demo performance.
Build the shortlist around the work you redesigned in phase one. Score each vendor against the four or five scenarios that cause you the most pain today: a mid-month mover with a pay change, a manager who needs headcount by cost centre, a leaver who has to come out of three systems on the right day.
Then test what demos hide:
- What does this look like for a line manager on a phone, on a poor connection?
- Which fields are standard, which are configurable, and which need a paid change request?
- Is configuration done by your team or by the vendor, and at what rate?
- How does the vendor handle UK statutory change, and when did they last ship one?
- Who supports the system after go-live, at what cost, and in which time zone?
Record the scores before the demo rather than after. A structured scorecard is the only reliable defence against a persuasive presentation.
Our HRIS selection work follows this structure, and how to choose an HRIS at 200 to 1,000 employees sets out the shortlist logic in detail. If you are weighing platforms directly, how to choose the right HR system covers the decision criteria in plain terms.
Phase 3: Data governance before migration

Migration is the phase most often under-resourced, and the one that decides whether anyone trusts the new system by month three. Moving messy data into a better platform gives you the same problems at a higher licence cost.
The principle is simple: do not move data you cannot define, own or explain.
Before extraction, settle five things:
- Which fields are essential, and who owns each one?
- What is the authoritative source for people, jobs, cost centres and organisational structure?
- Which definition of headcount, attrition and absence will the reports use?
- What happens to historical records that no longer serve a business purpose?
- Who signs off the reconciled population before it loads?
Then reconcile in the open. Pull headcount by department, location and contract type from the old system, payroll and finance, and explain every difference before you migrate. A difference you cannot explain becomes a reporting argument you cannot win.
If you need a focused way to do this, treat it as a data sprint: one population, one set of definitions, one reconciliation, finished in weeks. Our people data work covers the ongoing version, and keeping HR data accurate after an HRIS implementation covers what to do once the system is live.
Data quality is not a migration task with an end date. It is a design decision about who owns which field and how errors get corrected, which needs to exist before the first load and to still exist a year later.
Phase 4: Configure for compliance you can evidence

Compliance configuration attracts the most anxiety and the least testing. The aim is not a compliant-looking screen. It is a rule you can demonstrate behaves correctly on a specific date, to an auditor or a tribunal.
Work through the rules that are live now:
- Statutory Sick Pay calculated with no waiting days and no Lower Earnings Limit, including the interaction with company sick pay schemes and part-year earnings
- Day one entitlement for paternity leave and unpaid parental leave, with notice handling that matches the statutory process
- Absence and case records that support a defence of taking all reasonable steps to prevent sexual harassment, including confidential reporting and an audit trail
- Retention and access rules that let you meet a six-month tribunal claim window without holding records you cannot justify
Then the rules landing on 1 January 2027:
- Service-based rules that produce a six-month qualifying period, and dismissal workflows that reflect uncapped awards
- Onboarding workflows that issue the trade union right to join notification automatically, with evidence it was sent
For every rule, write three lines: the statutory requirement, the system setting that delivers it, and the test that proves it. That document is what turns configuration into assurance, and it is the artefact most implementations never produce.
Phase 5: Testing, parallel payroll and the edge cases
Testing is where a good project discovers its real problems while they are still cheap.
Test in this order:
- Configuration tests. Each statutory rule from phase four, run against a dated scenario. A new starter who is absent on day one. A leaver whose final pay crosses a pay period. A mover whose cost centre changes mid-month.
- Permission tests. Managers see their team and nothing else. HR business partners see their population. Payroll sees what payroll needs. Weak permissions are a data protection issue, not a usability preference.
- Integration tests. Employee data, absence, benefits and payroll files exchanged in both directions, with a failure tested as well as a success.
- Mobile tests. Holiday requests, payslip access and approvals on a phone, performed by the people who will actually do them rather than by the project team.
- Parallel payroll. At least two full cycles alongside the existing process, with every variance explained before you sign off.
Use real people and real records for user acceptance testing, with the names changed if policy requires it. Synthetic test data hides the duplicates and half-filled fields that caused your reporting arguments in the first place.
Phase 6: Adoption is a work change, not a training event
Most adoption plans are communication plans. They announce the system, run webinars and assume use follows. It does not.
People adopt a system when it is easier than the workaround, when someone checks that they did it, and when the first experience is not a mess. Design for those three conditions.
- Cut the number of steps a manager needs for the five most common tasks, and test whether a first-time manager can finish them without help
- Replace the old route rather than running alongside it. If the spreadsheet survives go-live, so does the workaround
- Give the first two weeks a visible owner who answers questions the same day, in a channel people already use
- Brief team leaders before their teams, so the first question an employee asks is answered by their manager
- Keep a short monthly check on who is using which task, and follow up where usage stalls
Training supports adoption but does not create it. What creates it is a process that is genuinely shorter than the one it replaces.
Phase 7: Go-live, then measure at day 90
Go-live is a date, not an outcome. What tells you whether the implementation worked is what you can measure three months later.
Run the first fortnight with a named response owner, a single place to raise issues, and a daily look at anything blocking pay or starters. Then move to a structured review.
At day 90, report on four things:
- Data quality: completeness and accuracy on the fields that feed reporting and pay
- Usage: which tasks are being done in the system, by whom, and where the workarounds reappeared
- Service: how quickly queries are resolved and how many are repeat questions rather than system faults
- Value: the time returned to HR and managers, compared with the case you made before you bought
That fourth point matters most, because it is the one that decides whether the next phase of work is funded. If you cannot show what the system returned, the answer will be that HR technology does not pay back.
For the reporting side, our HR technology optimisation work covers the review and the fixes, and if the platform is HiBob, implementing HiBob sets out what we change first.
The three failures that are predictable
These show up in almost every review we do, and each one is a design failure rather than a team failure.
Rebuilding the old process. The project documents the current workflow, the vendor configures it faithfully, and the organisation gets a faster version of work it should have stopped doing. The fix is the phase one decision about what stays, changes or stops, made before requirements are written.
Migrating without governance. Data is extracted, mapped and loaded by whoever is available, with no owner, no definitions and no reconciliation. The system then produces two headcount figures and the trust problem moves house. The fix is to name owners and agree definitions before extraction, not after the first board pack.
Treating go-live as completion. The project team disbands, hyper-care ends, and nobody measures whether the work got better. Six months later the spreadsheets are back and the case for further investment is gone. The fix is a day-90 review with numbers agreed before launch.
Frequently asked questions
How long does an HRIS implementation take?
For a UK business of 50 to 1,000 employees, plan for three to six months from kick-off to go-live, with parallel payroll cycles inside that window. Smaller populations with simple payroll can move faster, but compress discovery and you pay for it in rework after launch.
What should an HRIS implementation checklist include?
Work design and discovery, selection tested against real scenarios, data governance before migration, compliance configuration with evidence, testing including two parallel payroll cycles, an adoption plan that removes the old route, and a day-90 review with agreed measures.
What changes on 1 January 2027 that affects HR system configuration?
The unfair dismissal qualifying period drops to six months for dismissals from that date, compensatory awards are uncapped, and the duty to inform workers of their right to join a trade union comes into force. Any implementation spanning the new year needs both configured and tested.
Do we need a new HRIS, or should we fix what we have?
Not always. If the platform supports your redesigned processes and the problem is configuration, data and ownership, fixing those is faster and cheaper. Replace the system when the audit shows a genuine capability gap, not when the interface feels dated.
Can we keep our existing payroll provider?
Often yes, if the integration is proven in testing rather than assumed from a vendor list. Test a full cycle in both directions, including a failure scenario, before you commit to go-live.
How do we keep data clean after go-live?
Name an owner for each critical data domain, publish the definitions your reports use, run a monthly completeness and accuracy check, and fix errors at source. Data quality is a routine, not a project.
Where to start this week
An implementation is expensive, slow and difficult to reverse. Before you commit to a platform, establish whether the work, the data and the decision rights it depends on are ready for it.
The free Work Design Readiness Assessment takes about ten minutes and scores your function across work clarity, decision rights, operational drag, data foundations and measurement. It gives you a defensible answer to the question every board asks before signing off a system budget: what will this change?
If you would rather talk it through, the HR system selection and implementation pages set out how we work, and the HR work design FAQs answer the questions that come up most often.
By Kesiena Ogefere, founder of Copieux Group. Copieux Group is system agnostic and works with the HR, payroll and people systems you already run.
