Nobody can quote an ERP project from an email. That is not sales tactics, it is arithmetic: the same number of employees can mean four processes or forty. Still, the price is not a mystery. It rests on a handful of drivers, and once you know them you can steer them.
Driver one: how many processes move into the system
Every process that enters the system needs to be understood, configured, tested and taught. Selling and invoicing is a small set. Add a warehouse and you add receiving, put-away, picking, packing, returns and stock corrections. Add production and you add bills of material, routings, work orders and scrap. The count of processes, not the count of people, drives the effort.
Driver two: how much old data has to come along
Products, partners, open orders and stock levels are the minimum. History is where it gets expensive. Three years of accounting history in a foreign format is a project of its own. In most cases the sensible answer is to migrate open items and balances, and to keep the old system readable for a while. That single decision often saves more than any negotiation on the day rate.
Driver three: how many foreign systems are connected
Each interface is a small product: a data model on both sides, error handling, retries, a log and a person who is told when something fails. One shop connection is manageable. A shop, a payment provider, a carrier and a fiscal device in three countries is a different order of magnitude.
Driver four: how special your business really is
Most companies are standard in eighty percent of what they do and genuinely unusual in the rest. The unusual part is where custom work happens, and it is also the part that earns your money. A good partner pushes you towards the standard everywhere else, precisely so that the budget stays where it matters.
What a typical shape looks like
A bounded module, for example a costing extension or an interface, is a matter of weeks. A complete system for a company with a warehouse and production is several months, delivered in slices that each go live on their own. Anyone who promises a full ERP in three weeks is either selling you a template or has not asked enough questions.
How to keep the price down without wrecking the project
- Reduce scope, not quality. Two processes done properly beat six done halfway.
- Decide fast. Waiting for a decision is the most expensive activity in any project.
- Appoint one person on your side who can say yes. A committee doubles the calendar.
- Do not migrate history you will never read.
- Test with your own people, on your own data, early.
What to ask for in a proposal
Ask for a scope split into slices, with a price per slice and a definition of what going live means for each. Ask what happens when a slice runs long, and who pays. Ask what the running cost after go live is: hosting, backups, updates, support hours. A proposal that only shows a total number hides the risk instead of managing it.
My own approach is simple: after the first conversation you get a range, and the range is honest, including the upper end. After we map the processes together you get a fixed number for the first slices. If the mapping shows that the project makes no sense, I say that too. A project that should not exist is the most expensive one of all.