ERP Implementation

The phases of an ERP implementation, what really drives cost and duration, the pitfalls that derail projects and how to choose and manage an implementation partner.

Abstract illustration of connected modules representing an ERP system

Key points

  • ERP projects fail on scope, data and people far more often than on software.
  • A sound project moves through discovery, design, build, data migration, testing, training, go-live and stabilisation.
  • Adopting standard processes and limiting customisation is the most effective way to control cost and risk.
  • The business, not the IT department or the partner, must own the project.

Replacing an ERP system is among the most disruptive projects a company undertakes. It touches every transaction and every department, and it cannot be done quietly in the background. Public failures, such as Birmingham City Council’s Oracle programme, whose problems were examined in detail by external auditors, show how expensive it becomes when governance, design decisions and testing go wrong. Most projects that succeed do so for unglamorous reasons: clear scope, clean data, enough testing and an organisation prepared for the change.

The phases of an implementation

PhaseWhat happensKey outputs
1. Preparation and discoveryDefine objectives, scope, budget, team and governance; document current processes and pain pointsBusiness case, project charter, requirements
2. SelectionShortlist, scripted demonstrations, references, commercial negotiationChosen software and partner, signed contracts
3. DesignWorkshops to map future processes to the system; identify gapsSolution design, list of configurations, integrations and any extensions
4. Build and configureConfigure the system, develop integrations and reportsWorking prototype for review
5. Data migrationExtract, cleanse, map and load master and open transactional data; repeated trial loadsReconciled data in the new system
6. TestingUnit, integration and user acceptance testing with real scenarios; performance and security testsSigned-off test results, resolved defects
7. Training and cutoverRole-based training; detailed cutover plan and rehearsalGo/no-go decision
8. Go-live and hypercareSwitch over; intensive support for the first weeks and the first month-endStable operations, handover to support
9. OptimisationAddress deferred items, measure benefits, extend functionalityBenefits tracking, roadmap

Implementation approaches

  • Big bang. Everything goes live at once. It shortens the period of running two systems and concentrates the risk in one event.
  • Phased. By module, site or business unit. It lowers risk at each step and prolongs the programme, with temporary interfaces between old and new.
  • Greenfield or migration. A new design from scratch, or a technical conversion of the existing system that preserves configuration and history. The choice is central to programmes such as SAP S/4HANA migration.

What drives cost and duration

  • Scope. Number of modules, legal entities, countries, sites and users.
  • Customisation. Every modification must be designed, built, tested, documented and maintained through each upgrade.
  • Integrations. Connections to e-commerce, warehouse, banking, payroll and other systems.
  • Data. The state of master data and how much history is migrated.
  • Internal capacity. Whether key people are released from their day jobs.
  • Decision speed. Slow or reversed decisions are a leading cause of overrun.

Budgets should cover software, partner fees, internal backfill, integration, training, a contingency of at least 15% to 20%, and post-go-live support. A small, standard cloud deployment may take three to six months. Mid-sized, multi-site projects commonly run nine to eighteen months.

Common pitfalls

Recreating the old system

Users ask the new system to work like the old one, and each request seems reasonable. The sum is a customised system that is expensive to upgrade. Start from the standard process and require a business case for every deviation. In cloud ERP this “fit to standard” approach is not optional; see cloud ERP versus on-premise.

Underestimating data migration

Duplicate customers, obsolete items, inconsistent units of measure and missing tax details surface late and hurt badly. Begin cleansing data before the project formally starts, assign owners for each data set and plan several trial migrations with reconciliation.

Thin testing

Testing is the first activity to be squeezed when the timetable slips. User acceptance testing should run complete business scenarios end to end, such as order to cash and procure to pay, performed by the people who will use the system, including month-end and exception cases.

Neglecting change management

People need to know why the change is happening, what will be different in their job and how to do it. Communicate early, involve respected users as champions and train close to go-live using the company’s own data.

Weak governance

A project needs an executive sponsor with authority, a steering group that meets regularly and makes decisions, a full-time project manager on the client side and honest reporting. Status reports that are green until the month before go-live are a warning sign in themselves.

Going live at the wrong time

Avoid peak trading periods and year-end. Define go-live criteria in advance and be prepared to postpone if they are not met; a delayed go-live costs less than a failed one.

Choosing an implementation partner

  • Experience in your industry and with organisations of your size.
  • The actual consultants proposed, not only the firm’s credentials; ask to meet them.
  • Reference customers you can speak to privately, including one where things were difficult.
  • A methodology that includes data, testing and change, not just configuration.
  • Clear commercial terms: fixed price or time and materials, how change requests are priced, what is assumed about your team’s contribution.
  • Support arrangements after go-live.

Whatever the contract says, risk cannot be fully outsourced. The partner brings product knowledge; knowledge of the business and ownership of decisions must come from you.

Measuring success

Set a small number of measurable targets in the business case: days to close the month, stock accuracy, order processing time, days sales outstanding, systems retired. Review them six and twelve months after go-live. Without measurement, benefits are asserted, not delivered.

Frequently asked questions

How long does an ERP implementation take?

Typically three to six months for small standard cloud projects, nine to eighteen months for mid-sized organisations and longer for multinational programmes.

Why do ERP implementations fail?

The recurring causes are unclear objectives, excessive customisation, poor data, inadequate testing, insufficient internal resources and weak executive sponsorship.

Should we use the vendor or a partner to implement?

Either can work. Smaller projects are usually delivered by partners. Judge the specific team and its experience in your sector.

How much history should we migrate?

Usually master data, open transactions and opening balances, with older history kept accessible in an archive or reporting database. Migrating everything adds cost and risk for limited benefit.

The bottom line

An ERP implementation is a business change programme with a software component. Keep the scope tight, accept standard processes wherever possible, invest early in data and testing, and keep the decisions in the hands of the business. For background, see what ERP is.

Sources and further reading

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top