Short answer: an ERP rollout in a small or mid-sized company takes months, not years, if you go module by module. Most overruns aren't caused by the software. Three things cause them: requirements that keep growing along the way, messy data and too much custom development. Keep those three under control, and an ERP implementation isn't the nightmare its reputation suggests.
The reputation isn't unfounded. Statistics on large ERP projects have been sobering for decades. But the lessons from multi-year, multi-million enterprise projects only partly apply to a company of 30 to 200 people, where processes are simpler and the rollout can be split into modules.
Where companies stand
According to Eurostat, 46.45% of EU enterprises used ERP software in 2025. The gap by size is huge: 41.08% of small companies against 88.71% of large ones. Most smaller companies still work between spreadsheets, a separate invoicing program and a few specialist tools, and sooner or later reach the point where that no longer keeps up with growth.
What the numbers show
The most-quoted source is Panorama Consulting's annual ERP report. Always read its figures with the year attached, because the sample changes from year to year.
2017 (342 respondents):
- 74% of projects exceeded their original budget, and 59% their schedule;
- the average implementation took 16.9 months;
- Panorama's clients spent on average 3.6% of annual revenue on the project;
- the main causes of budget overruns: unanticipated technical or organisational issues (23%), additional technology requirements (22%) and expanded scope (20%);
- the main causes of schedule overruns: expanded scope (19%) and data issues (16%).
2026: Panorama's latest report says "more than a quarter" of organisations exceeded their budgets, with additional technology needs as the leading cause. That's far better than earlier years, though the samples aren't fully comparable.
The long view: in 2010 Computerworld cited data from 1,600 organisations: 51.4% went over budget, and more than half achieved no more than 30% of the benefits they expected.
One important caveat: these samples mostly cover mid-sized and large US companies. In an SME the numbers are smaller, but the causes of overruns are the same.
How long does it take in an SME?
An implementation typically has five phases:
- Process assessment: how the company works today, what hurts, what the system has to do. A few weeks in an SME.
- Design and demo: mapping the processes onto the system, with key users involved.
- Configuration and data migration: master data, opening stock, open items.
- Testing and training: with real cases, led by key users.
- Go-live and follow-up: the first weeks bring the most questions and fixes.
If all of that runs at once for every module, even an SME can easily spend six months to a year. If you go module by module (for example invoicing and inventory first, then purchasing, then CRM), the first module can go live within weeks or a few months, and the company uses it while the next one is being prepared.
What does it cost?
The licence or subscription is only part of the cost. A realistic budget has six items:
- software: a subscription per user or module, or a licence;
- implementation: the vendor's or consultant's fees from assessment to go-live;
- data migration: cleaning and loading old data, which is almost always underestimated;
- integrations: web shop, bank, tax authority reporting, specialist tools;
- customisation: everything beyond standard functionality;
- internal time: the time of key users and managers, which most budgets leave out entirely.
The last item is the biggest hidden cost. During the rollout months your best people have to test, decide and train on top of their day jobs. If they don't have time, the project slips, or the system won't reflect how the company really works.
What makes it fail?
1. Growing scope
Most ERP projects don't fail on one big mistake but on a hundred small "while we're at it" requests. Each is reasonable; together they double the project. In Panorama's data, expanded scope accounts for 20% of budget overruns and 19% of schedule overruns.
2. Messy data
Old systems contain duplicate partners, wrong item numbers, outdated prices and fields nobody understands. Load them without cleaning and the new system calculates wrongly from day one, and users lose trust in it.
3. Too much custom development
Every customisation makes the rollout more expensive and maintenance and upgrades harder. A Hungarian ERP consultant suggested in a newspaper piece that custom development should cover no more than 10 to 15% of requirements. That's a rule of thumb, not research, but it's a good starting point: if you need much more, you probably chose the wrong system.
4. Poor selection
The same piece names selection as the main cause of failure. A common mistake is choosing on feature lists instead of your own processes. Every demo looks good. The right question is how the system handles your most complicated everyday case.
5. No owner
If "IT" runs the project and the business side only joins at the end, the system will work technically but not the way the company works. You need a business-side owner who can make decisions, and key users from every area.
How to choose a system
Selection is the cheapest phase of the project, and it decides the most. Questions worth asking every vendor:
- Show me my most complicated everyday case, not your demo script. For example, a partial delivery with a return and a foreign-currency invoice.
- What is standard and what is customisation? Get it in writing, requirement by requirement.
- How does it connect to the web shop, the bank, tax authority reporting and your specialist tools?
- What happens to customisations when the system is upgraded?
- Who leads the rollout, and how many companies of a similar size have they done it for?
- How do you take your data with you if you ever want to switch?
Compare two or three vendors on the same cases and the differences show quickly. Feature lists rarely decide; cases do.
All at once or module by module?
In Panorama's 2019 data, 33% of companies rolled out ERP by functional area (module by module), 30% all at once ("big bang"), and the rest by business unit, by location or in a hybrid way. Panorama describes big bang as high-risk, high-reward, and phased rollouts as safer but more expensive.
For SMEs we recommend going module by module, for three reasons:
- earlier value: the first module already works before the others are finished;
- lower risk: a problem affects one area, not the whole company;
- the team learns: experience from the first module speeds up the next.
The price is that for a while the old and new systems run side by side, with a temporary link between them.
What we've seen
Two earlier projects of our team show what this means in practice. Both were built on the modular ZEBEQ platform, which we still use most often today.
- HR processes in one system: at the local unit of an international fleet management company, leave, onboarding, performance reviews, recruitment and shift planning ran in spreadsheets. In seven months they moved into one system used by 70 people, HR and employees alike.
- Replacing SAP Emarsys: at an electronics retail chain operating in two countries, a large, complex SAP Emarsys setup was replaced in five months by an in-house marketing system we developed on a no-code platform. It runs campaigns and customer communication automatically, with extended functionality and at lower cost.
Neither was a big bang. Both started with a clearly bounded area that could go live before the next step came.
The first month after go-live
Go-live day isn't the end of the project; it's the start of the most critical period. The first weeks show where the configuration doesn't fit real work, and that's when users have the least patience.
Three things help. A visible, fast channel for questions and problems, with someone who answers the same day. Key users in every area whom colleagues can turn to first. And a short weekly review: what improved, what's still open, what's next. If the first month goes well, the system is accepted. If it goes badly, it stays "the new system that doesn't work" for years.
How to stay in control
- Write down the scope per module, and put every new request on a list rather than straight into the project. The list comes after go-live.
- Clean the data before migration, and decide what gets migrated at all.
- Limit custom development. If a need can be met with standard functionality, change the process, not the system.
- Appoint a business-side owner and key users, and free up their time.
- Define acceptance criteria up front: when is a module done?
- Put internal time into the budget.
- Don't close the project on go-live day. Plan follow-up for the first month, and only then start the next module.
If you're about to roll out an ERP and want someone to take you through it module by module, from process assessment to go-live, have a look at our business management (ERP) and IT consulting services.
FAQ
How long does an ERP implementation take in an SME?
With a module-by-module rollout, the first module can go live within weeks or a few months; the whole system typically takes a few months to a year. In Panorama's 2017 report, which covered mostly larger companies, the average was 16.9 months.
How much does an ERP implementation cost?
It depends on company size, the number of modules and how much customisation you need. Among Panorama's 2017 clients it averaged 3.6% of annual revenue, but those were mostly larger companies. The most underestimated items are data migration and internal staff time, so always budget for them.
What most often causes delays?
Growing scope, data issues and unexpected organisational or technical problems. In Panorama's 2017 data, expanded scope caused 19% of schedule overruns.
All at once or module by module?
For SMEs, usually module by module: earlier value, lower risk, and the team learns along the way. A big bang can be faster, but one problem affects the whole company.
How much customisation is acceptable?
As little as possible. One practitioner rule of thumb is no more than 10 to 15% of requirements. If you need far more, reconsider your choice of system.
Sources
- Panorama Consulting – 2017 ERP Report
- Panorama Consulting – Latest study of ERP implementation outcomes (2026)
- TechTarget – ERP vendors get high marks in Panorama report (2019)
- Panorama Consulting – ERP Implementation Strategies
- Computerworld – Widespread discontent persists with ERP projects
- Eurostat – E-business integration (Statistics Explained)
- HVG – The dark secrets of failed ERP projects (in Hungarian)
