Off-the-shelf ERP or custom software? A framework for an expensive decision

We are usually asked this at the worst possible moment: when a company has already bought a packaged system, spent eighteen months rolling it out, and discovered that the factory's core process does not fit inside it. Or the reverse — when a custom system has been commissioned that reproduces the current process exactly, including every inefficiency in it.

There is no general answer. There are five questions, though, and when they are answered honestly the decision tends to make itself.

1. Is this process a competitive advantage, or does it just have to work?

In most companies, accounting, payroll and stock control create no advantage at all; they simply have to be correct. For those, a packaged product is almost always the better choice: cheaper, faster, and carrying the accumulated experience of hundreds of other companies. But if the way your factory cuts material and minimises offcuts is what produces your margin, handing it to a generic module means converting your advantage into the industry average.

2. How much are you willing to change your process?

Every packaged system is a set of assumptions about how you work. A successful rollout means accepting a large share of those assumptions. Companies that try to configure a packaged system into an exact copy of their present habits usually end up with the worst of both: they pay for the customisation, they do not gain the flexibility of bespoke software, and every upgrade breaks what they changed.

3. How stable is your process?

If the way you work has been roughly the same for three years, a packaged system carries little risk. If you are growing quickly, changing production lines or entering a new market, software you cannot change quickly becomes a ceiling on growth soon enough. The real question is not what we need today, it is how long it takes to change this system once what we need has changed.

4. Have you compared the real cost, or only the price?

• Licences and annual maintenance, per user, over a five-year horizon
• Implementation, data migration and training — usually more than the licence itself
• Customisation, and how much of it has to be redone at every upgrade
• For custom software: maintenance, later development, and dependence on the team that built it
• The hidden cost of a poor fit: the work people keep doing in spreadsheets because the system does not cover it

That last line is the most important one, and it is almost never in the comparison. If half the organisation still keeps its own spreadsheet after the rollout, you have paid for a system and you still do not have one.

5. Who owns the data, and how do you get out?

Whatever you decide, a day may come when you want to leave. Settle it before the contract is signed: in what format and at what cost the data can be exported, whether a documented API exists, and how maintainable the system is if the relationship with the supplier ends. That clause is the cheapest insurance you can buy.

The worst position is a system that is neither yours nor possible to leave.

The middle path that usually works

In practice, most of the factories we have worked with have arrived at a mix: a packaged core for finance and inventory, plus a custom layer over the part that is genuinely specialised — production planning, quality control, or the link to a network of dealers and agents. This path has one condition for success: the border between the two parts, and how data crosses it, has to be clear from day one.