Every ERP vendor will tell you that you needed one two years ago. Most of the time that is wrong. Spreadsheets and a decent accounting package carry a company much further than the sales conversation suggests, and moving too early buys complexity you cannot yet absorb.
Moving too late is expensive in a different way. Here is how to tell which you are.
Five signals that it is actually time
One: the same number exists in three places and nobody knows which is right. Not "the reports differ" — that is normal. This is when a decision gets delayed because two people brought different figures to the same meeting.
Two: someone's full week is reconciliation. Add up the hours spent copying between systems, chasing statuses and rebuilding the same report. When that reaches roughly one full-time person, the system is now cheaper.
Three: month-end takes longer than it did last year, at the same volume. That is the clearest early indicator of structural strain rather than a busy period.
Four: you cannot answer a customer without asking a colleague. Stock, delivery date, order status. If the answer lives in someone's head, growth is capped by that person's availability.
Five: a compliance or audit request takes days. Swiss law requires business records to stay complete and auditable for ten years (Code of Obligations, Art. 958f). Traceability that has to be reconstructed by hand each time is a risk that will eventually be priced.
One signal is normal. Three at once means the reconciliation cost has crossed over.
What are the cheaper things to try first?
Fix the numbering, fix the master data, and connect the two systems you already have. Most small companies get a year or more of runway from consistent article and customer codes, one owner for master data, and a single integration between the webshop or CRM and the accounting package.
That is weeks of work rather than a programme, and it is not wasted if you do eventually move: clean master data is the largest cost line in any migration, so you are pre-paying it either way.
What does an ERP actually cost a small company?
| Company size | Typical first-year total | Where the money goes |
|---|---|---|
| 10–30 people | CHF 25,000–80,000 | Licences, configuration, data cleanup |
| 30–100 people | CHF 80,000–250,000 | Above, plus process work and integrations |
| 100–300 people | CHF 250,000–700,000 | Above, plus parallel running and training |
Indicative Swiss market observations, varying widely with complexity and how clean your data is. Two things are consistently underestimated: data migration, and the internal time your own people spend — which is real cost even though no invoice arrives for it.
Choose the smallest thing that solves it
A small company does not need the platform a 2,000-person manufacturer needs. The Swiss market has options that fit properly at this scale, including finance-and-payroll-first products that handle local compliance natively and remove an entire integration workstream.
The trap is buying for the company you hope to be in five years. You will migrate again anyway when you get there, and in the meantime you pay for complexity you are not using. Pick for the next three years.
What breaks when you move too early
A system bought before the company needs it arrives without the discipline to run it. Nobody owns master data, so the new system inherits the same duplicates in a structure that is harder to correct. Nobody has agreed the process, so it gets configured around whoever was in the room. And because the pain it was meant to solve had not yet appeared, no one can tell whether it worked.
The visible cost is the licence. The real cost is that a company of 25 people now spends part of every week feeding a system, and the flexibility that made it fast is gone. That trade is worth making — but only once the reconciliation cost has genuinely crossed over.
What to do before you talk to a vendor
Write down the ten things about your operation that break a standard system. Not three hundred requirements — ten. Then have vendors run those ten as scripted scenarios with your own data, rather than showing you a demo they rehearsed.
That method, and why the requirements matrix fails, is set out in ERP selection for the Swiss mid-market. And once you are running the programme, the constraint stops being technology and becomes how fast your own organisation answers questions — the pattern in why ERP programmes stall.
If the honest answer is "not yet"
Say so, write down the three signals you will watch, and revisit in six months. A deferred decision with a trigger is a decision. Our ERP practice tells companies to wait fairly often, and that conversation costs nothing.
Related reading
- ERP selection for the Swiss mid-market
- Why ERP programmes stall in Swiss mid-market companies
- Digital transformation for an SME: the first 90 days
Working on this right now?
Tell us where you are in two questions. A consulting partner reviews every enquiry within 2 business days.