What actually happens between signing and go-live, what each phase should produce, and where the time goes.
An ERP implementation runs through discovery, design, configuration and build, data migration, testing, training, cutover and hypercare. The phases overlap in practice, but each produces something specific. Projects slip when a phase is declared complete without its deliverable existing — most often design sign-off without a documented design.
The phases are not the interesting part; the exit criteria are. A project that moves from design to build because the calendar says so, rather than because the design is agreed and written down, will rediscover the design during build at several times the cost. Almost every overrun traces back to a phase that was exited early.
Current processes, systems, data and pain points documented. Produces a scope and a fit-gap, not a wish list.
How the system will be configured, what will be customised and what the business will change. Signed off in writing before build starts.
Configuration, extensions, integrations and reports, delivered in increments the business can see.
Extract, cleanse, map, load, reconcile — repeatedly. Begins in the first weeks and runs in parallel throughout.
Integration and user acceptance testing against real data, training on the configured system, then a rehearsed cutover and hypercare.
A mid-market ERP implementation budgeted at six months breaks down, in practice, quite differently from how it is usually planned.
Discovery and design take around six weeks and are almost entirely decision-making rather than software. Configuration and development take around ten weeks and are the part everyone pictures when they imagine the project. Data migration takes around eight weeks running in parallel, and it is the activity that most often slips. Testing takes around six weeks including at least two full cycles. Training and cutover preparation take around four. Hypercare runs four to six weeks after go-live.
The two observations that matter: migration and testing together exceed configuration, and both depend heavily on the client's own effort rather than the implementer's. Plans that allocate most of the calendar to configuration and treat migration as a late task are the ones that overrun. If you want to know whether a plan is realistic, look at what it assumes about data cleansing and who is doing it.
| Term | What it means |
|---|---|
| Discovery | Current-state mapping and fit-gap, producing scope rather than a sales proposal. |
| Conference room pilot | Walking real business scenarios through the configured system before build completes. |
| Cutover | The transition from old system to new, rehearsed before it is performed. |
| Hypercare | Intensive support immediately after go-live, with the build team still engaged. |
| Benefits realisation | Checking afterwards whether the business case actually happened. Routinely skipped. |
If it is not written down, it will be remembered differently by each party during build.
The most common single cause of a delayed go-live. Cleansing is the client's work and takes months.
Sample data passes. Real data contains the twenty years of oddities that break things.
Go-live is when problems surface. A team that disbands on the day leaves them with nobody.
| Metric | What it tells you | What distorts it |
|---|---|---|
| Open design decisions | Decisions waiting on the business. The most common cause of schedule slip. | |
| Test cases passed on first execution | Build quality. Low rates early mean the design was not understood. | |
| Migration rehearsal count | Two or three full rehearsals before cutover is normal; fewer is optimistic. | |
| Defects found after go-live | Testing effectiveness, and the thing hypercare is sized against. |
We build and support the systems this page describes. Start with a free working proof of concept — no cost, no commitment.