HRMS implementation: a 90-day plan

An HRMS implementation done in 90 days follows a fixed order: map the HR processes as they are actually run, configure the system to that map, integrate payroll and identity, migrate employee data and rehearse the migration, train each role on its tasks, go live before a payroll cut-off with hypercare, and measure adoption at 30, 60 and 90 days. The order matters more than the speed. Teams that configure before mapping, or migrate without rehearsing, spend the second quarter undoing the first. This is the plan we run inside our enterprise platform implementation service, adapted here for HR.

HR systems are unusual among enterprise platforms because every employee is a user, most use it rarely, and one process (payroll) is unforgiving. A CRM that is wrong loses a deal; an HRMS that is wrong pays someone the wrong amount and the whole company hears about it by lunch. The plan is built around that asymmetry.

What an HRMS implementation covers

A typical mid-size scope includes the employee master, organisation structure, leave and attendance, payroll or payroll integration, expenses, onboarding and exit, performance cycles, and employee self-service on web and mobile. In India it also means statutory reporting, provident fund, professional tax and tax declarations, which is where HR system implementation India projects differ from the generic playbook. Each module has a process owner in HR, a data source, and usually a spreadsheet it is replacing.

The 90-day HRMS implementation plan

Days 1–15: map the process as it is run

The HR policy document describes how leave works. The HR executive's inbox describes how it actually works: the manager who approves by WhatsApp, the plant that records attendance on paper, the contractor category that has no policy. Both must be captured, and the second is the one that determines configuration. Interview managers as well as HR, because managers are the users whose approvals block everything downstream. Write down every exception and decide, with the HR head, whether the new system will support it or retire it. Retired exceptions need a communication; supported ones need configuration.

Days 10–35: configure to the map

Configuration runs in parallel with the tail of mapping. Build the organisation structure first, because everything hangs off it, then leave and attendance policies, then approval chains, then self-service. Resist the vendor's template wherever it disagrees with the map. Test with ten real scenarios: a new joiner mid-month, a resignation with notice buy-out, a maternity leave, a shift worker with overtime, a transfer between entities. If the sandbox cannot handle those, the live system will not either.

Days 20–50: integrate payroll and identity

Payroll and identity are the long pole of any HR software rollout. Payroll integration means either configuring the HRMS payroll engine with your components, statutory settings and tax rules, or building a reliable exchange with an external payroll provider: inputs out, payslips and journals back. Run at least one full parallel payroll where the old and new systems both process the same month and every difference is explained. Identity means single sign-on and the joiner-mover-leaver flow: a new hire in the HRMS should create accounts, a leaver should remove them the same day. Integrating a new platform with finance, identity and messaging goes deeper on the patterns.

Days 30–60: migrate, rehearse, reconcile

The employee master will be dirtier than HR expects: inconsistent designations, missing dates of joining, leave balances that live in three places. Agree cleansing rules early, cleanse on a copy, and rehearse the full migration at least twice. Reconcile headcount by entity, total cost-to-company, leave balances and bank details, and have the HR head and finance sign the reconciliation. Bank details in particular deserve a sample check by a human, because a wrong account number is discovered on payday.

Days 45–65: train on tasks

Five audiences, five sessions. Employees learn to apply for leave, mark attendance, view payslips and submit tax declarations on the mobile app. Managers learn to approve and to see their team. HR operations learn the joiner and leaver flows. Payroll learns the monthly run. Finance learns expenses and journals. Each session is hands-on in the real system with realistic data, and each ends with the person completing the task alone. A recorded version of each session becomes the onboarding material for future hires.

Days 60–90: go live, hold the line, measure

Cut over immediately after a payroll cycle closes so the first cycle on the new system has a full month of clean inputs. Hypercare for the first month is daily triage and someone physically present where the users are; Hypercare: what the first month after go-live should look like describes it. At 30 days review adoption: what share of leave requests came through the system, how many managers approve in-system, which spreadsheets survive, and what the tickets say. Fix the top three issues before the 60-day review.

A worked example

A hospital network with several sites had attendance on biometric devices at some sites and registers at others, three leave policies that had drifted apart, and payroll run by an external provider from spreadsheets emailed on the 25th. The mapping phase surfaced the drift; the HR head chose one policy set with two documented exceptions. Integration connected the devices where they existed and gave the register sites a mobile check-in, and payroll inputs went to the provider as a file with a reconciliation report. Two parallel runs found differences that were all explained by the old spreadsheets, not the new system. Go-live followed a month-end, and the 30-day review showed managers approving in-system at most sites, with one site needing a second training visit. We later added a multilingual voice agent for the same network, which is a separate story, but the clean employee master made it easier.

Team and timeline

Our side: a solutions consultant who owns the process map and configuration, an integration engineer for payroll and identity, a data engineer for migration, and a trainer. Your side: an HR process owner with decision authority, a payroll owner, a finance contact for journals and expenses, an IT contact for identity, and a super-user per site. The 90-day plan fits the enterprise platform implementation service from $28,000 (from ₹18,40,000), with the exact figure driven by modules, sites and integrations. After hypercare, a Care Plan from $1,000 per month covers fixes and policy changes; details are on the pricing page. Where the HRMS is a custom build rather than a product, custom enterprise software applies instead.

Before you start: a checklist

• Confirm the payroll cut-off dates and plan go-live immediately after one
• Name the HR process owner, payroll owner and IT identity contact
• Collect every leave and attendance policy in force, including unofficial ones
• Locate every source of employee data and leave balances
• Agree cleansing rules and who signs the migration reconciliation
• Decide which exceptions will be supported and which retired
• Plan two parallel payroll runs before go-live
• Write the 30, 60 and 90 day adoption reviews into the plan with the sponsor's name

Questions clients ask

• Can we go live module by module? Yes: employee master, leave and attendance first, payroll next cycle, performance later. Payroll should never go live mid-month.
• Should we keep the external payroll provider? Often yes for the first year; integrate cleanly and reconsider once the employee master is trusted.
• What about employees without email? Mobile self-service with phone-number login and a kiosk at sites covers most cases; plan for it in configuration.
• How do we handle statutory reporting in India? Configure provident fund, ESI, professional tax and TDS in the payroll engine or provider, and verify on the parallel runs.
• Does AI have a place in an HRMS? Yes, for policy questions, drafting letters and anomaly checks on payroll inputs; AI in HR software covers it.

Related reading

Why enterprise software implementations fail on adoption explains the risks this plan is designed around, and Data migration for platform implementations covers the migration phase in detail. For identity integration, Microsoft's Entra ID documentation describes the joiner-mover-leaver provisioning patterns most HR systems connect to.

Ninety days is enough for an HRMS if you map before you configure, rehearse before you migrate, and keep measuring after you go live.