An ERP project begins with the question of whether the real bottleneck is a software problem at all. Two starting points lead to the same decision. A company implements an ERP system for the first time, or an existing system has reached a version where support, security or functionality run out. Both cases call for the same calculation. Costs include, besides the licence, data migration, training and parallel running. Then there are the realistic duration and the consequences of postponing the step.
At a glance
- An ERP rollout rarely fails because of the software. Usually there is no process owner, the master data is poor, or the scope grows uncontrolled during the project.
- For an existing system, an upgrade is due when support for the running version ends, a new legal requirement arises or the system hits a scaling limit. Just because a new version has been released, nobody has to switch.
- The largest cost block in both cases is custom adaptation, data migration, training and ongoing operation. The licence is often smaller.
When a new ERP implementation is really necessary
Many mid-sized companies notice late that their IT, which has grown over the years, is holding them back: isolated tools that do not talk to each other, spreadsheets as a secret database, duplicate master data in sales, inventory and accounting. A suitable ERP system bundles these processes into one shared body of data. Four signs suggest that it is time.
- Employees maintain the same record in several places.
- The month-end close takes long because figures are gathered from several sources.
- Quotes, orders and invoices are not linked end to end.
- Growth, new sites and legal requirements can no longer be mapped cleanly with the existing tools.
Not every pain point justifies a new ERP. Sometimes it is enough to define a process properly or to connect two existing systems. An ERP project ties up money, staff and attention for months, and a clear bottleneck should justify that.
The phases of an ERP project
A reliable ERP project runs in traceable phases, and each one delivers a result that the next builds on.
| Phase | What happens |
|---|---|
| Current-state analysis and goals | Which processes run how today, where do they break, and what should the ERP measurably improve? Without prioritised goals, every later detail becomes a point of dispute |
| Requirements and selection | Functional and non-functional requirements are documented and weighted. The software is chosen on real scenarios from your own operations |
| Concept and adaptation | The target concept is drawn up. The most important decision falls here: which processes does the company adapt to the software, and where is a custom adaptation worth it? |
| Data migration | Master and transaction data are cleaned, mapped and transferred. This phase is almost always underestimated |
| Testing and training | Key users test against real business cases, errors are fixed, and staff are trained at their actual workplace |
| Go-live | Planned cleanly with a cut-over date, responsibilities and a fallback scenario |
| Aftercare | The first weeks after go-live with extra support, quick fixes and close guidance for users |
All at once or in stages?
With a full cut-over, the whole ERP goes into production on one date. That is fast and avoids long parallel running, but it concentrates all the risk on one day. If something goes wrong, the whole company feels it. A staged implementation rolls out modules or sites one after another. That lowers the risk per step and allows learning, but it lengthens the project and forces temporary double data entry and interfaces between old and new system.
There is no blanket recommendation. For manageable organisations with clear processes, the full cut-over can be the more pragmatic path. With many sites, complex workflows or little internal IT capacity, the staged variant is usually the better choice. What decides it is a view of your own risk tolerance and ability to absorb an outage.
Cost drivers and why projects fail
Licence or subscription prices are rarely the biggest item. The real cost drivers are custom adaptation, data migration, training and ongoing operation covering hosting, maintenance and further development. Anyone who looks only at the licence plans the budget wrongly. Custom adaptation is especially treacherous, because every special solution raises the cost of implementation and weighs on every later update.
Organisational causes are the most common, and four of them keep coming back.
- No process owner. Nobody decides bindingly how a process will run in future. Without ownership from the business side, every discussion fizzles out.
- Poor master data. Duplicates, gaps and inconsistent formats from the legacy system are taken over unchecked and burden the new ERP from day one.
- Uncontrolled scope growth. Every department adds one more requirement, and schedule and budget slip out of control.
- Too much custom adaptation. The software is bent to every historical habit. That is expensive, error-prone and makes later updates harder.
When an upgrade pays off
The second starting point concerns companies that already run an ERP system and face the question of whether and when to move to a newer version. Three triggers decide this, whatever system is in use.
- Support and security. Every ERP system has a support lifecycle. If the version in use leaves vendor maintenance, there are no more security updates. Running an installation without security fixes is usually the most compelling reason to change. Deadlines and paid extensions are stated by the vendor in its support terms, which you should check before any decision.
- New legal requirements. Changes in tax reporting, invoice formats or retention duties can force a version that the old one can no longer deliver.
- Scaling limits. Growing data volumes, new sites or additional requirements can bring a system to its limit regardless of its support status.
If the new version also brings functions that solve a real business problem, that strengthens the case. If not, support and security remain the main arguments. A stable, still maintained version without urgent functional need may deliberately keep running.
The biggest effort of an upgrade sits mostly in custom code. Interfaces, custom reports and adapted screens have to be checked for compatibility at every major version jump. A five-step approach has proven itself.
- Inventory. A complete survey of all customisations, interfaces and add-on modules.
- Test run on a copy. The migration is rehearsed first on a copy of the production database, never directly on the live system.
- Code adaptation. Custom-developed extensions are ported to the new version.
- Structured tests. Core processes such as quote, order, invoice, inventory and reporting are tested thoroughly against the migrated database.
- Planned cut-over. Only when everything runs reproducibly and without errors does the switch follow, in a planned maintenance window with a tested fallback plan.
A common assumption is that several version steps can be skipped without consequence. Technically, though, every migration builds on the changes of the earlier versions. The bigger the jump, the more changes hit the customised code at once, and the harder it is to narrow down the cause of an error. Very old installations therefore often benefit from a step-by-step migration through intermediate versions. Each step stays testable and reversible on its own. That costs more planning at first and considerably lowers the risk of expensive, hard-to-find errors in live operation.
Upgrade, replace or wait
| Situation | Recommendation |
|---|---|
| Version still under vendor support, no new requirements | Wait and note the end of support in the calendar |
| Support ends within the next year, the system otherwise covers the requirements | Plan the upgrade now |
| New legal requirement the system cannot map | Start the change regardless of support status |
| The system structurally hits limits (data volume, sites, processes) | Check whether an upgrade is enough or a system change is needed |
| Custom adaptations clearly exceed the standard, each change becomes more costly | Consider replacing it with a system whose standard scope fits better |
The link to financial accounting
In Germany, accounting is central, above all the interface to the tax adviser's software. A modern ERP system has to deliver data in a format the adviser's software can process, set up cleanly and matched to your own chart of accounts. If a system does not bridge this stably, permanent manual extra work results. Whether format and posting logic meet your adviser's expectations and the commercial and tax retention rules, you should agree with the adviser in every case. We deliberately give no blanket compliance promise.
Change management, realistic duration and outlook
An ERP changes how people work every day. Change management is therefore critical to success. Early involvement of key users, open communication about the reason and training on concrete workflows decide whether the system is accepted or worked around. The realistic duration of an implementation or an upgrade depends on scope, data situation and the extent of customisation. A focused project with consistent use of standards is productive much sooner than one that maps every special requirement individually. What is credible here is an estimate calibrated to your own processes.
In the medium term, the focus shifts from implementation to continuous development. A modern ERP is a platform that grows with the company and makes automation on your own data possible. Our ERP and CRM for mid-sized companies is part of Beyonetix OS (BEYOS), described under ERP and CRM. Which external services are used is set out on the Sovereign infrastructure page. If you want support with a migration or an adaptation, our services are listed under Services.