← SORT

ERPNext implementation and data migration

Setting up an open-source ERP, moving existing records into it, and building the parts it doesn’t provide. Including the part most people skip: deciding whether the move is worth making at all.

When an ERP is the right answer

An ERP earns its place when the same information is being kept in more than one system and somebody has to reconcile them by hand. Invoices in one application and the work they relate to in another. A spreadsheet that is the real record, emailed around, with three versions in circulation.

It is the wrong answer when the thing you actually need is one report, one screen, or one question answered. Those are cheaper to build directly than to get out of a system designed to do everything.

What a migration actually involves

Very little of it is the import. The work is in the shape of the data: deciding what each record means, what is a duplicate, what was never filled in, and what has to be true before a row can move at all. Old systems accumulate assumptions that only exist in the head of whoever used them.

Then it is reconciled against the original — not spot-checked. Counts, totals and balances have to agree with the system being replaced before anyone is asked to trust the new one.

Customisation that survives an upgrade

A migration that works once is a trap. If the custom parts are made by hand inside the live database, the next upgrade quietly breaks them and you are paying somebody to redo the work.

So anything custom is installed as a proper part of the system. It upgrades with everything else, it comes back by itself on a rebuilt server, and reinstalling it re-applies the definitions without writing over the records. The test is that your data stays usable without anyone re-running the migration.

When not to switch it on

I once finished an accounting migration, reconciled it against the original, and then advised against going live with it. What it would have replaced was cheap enough that the change did not pay for itself.

Finishing a build is not a reason to recommend using it. That judgement is part of the work, and it is better made after the migration proves the data can move than guessed at before.

What it costs to run

Open-source means no licence per user. It does not mean free: there is hosting, backups, upgrades and somebody who understands the system. I’ll put a number against those before you decide, because they are the costs that decide whether this is cheaper than what you have.

Common questions

Is ERPNext free?

There is no licence fee per user, which is usually the largest line in a commercial ERP. It is not free to run: there is hosting, backups, upgrades and somebody who understands the system. Those costs decide whether the move is cheaper than what you have, so they belong in the decision rather than after it.

Will customisations survive an upgrade?

They do if they are built as a proper part of the system rather than made by hand inside the live database. Done that way they upgrade with everything else, come back by themselves on a rebuilt server, and reinstall without writing over your records. Done the other way, the next upgrade quietly breaks them.

How do I know the migrated data is correct?

Because it is reconciled against the system being replaced, not spot-checked. Counts, totals and balances have to agree with the original before anyone is asked to trust the new system. Until they do, the migration isn’t finished.

Can ERPNext replace our existing accounting software?

Sometimes, and sometimes it shouldn’t. If the accounts and the work they relate to sit in different systems and somebody reconciles them by hand, bringing them together pays. If your current accounting software is cheap and works, replacing it may cost more than it saves — I’ve finished a migration and then advised against switching it on for exactly that reason.

What if the move turns out not to be worth making?

Then I say so, and that is part of the work rather than a failure of it. The judgement is better made after a migration has proved the data can move than guessed at beforehand. Finishing a build is not a reason to recommend using it.

Tell me what your team does by hand, what your current software can’t do, and what you want to change.

Email me about the work Book a call Twenty minutes, on a call