← All articles

Blog

8 Odoo Customization Mistakes From Documented ERP Failures

Public ERP failures at Lidl, Birmingham City Council and Hershey show how customization goes wrong. Eight lessons to apply before you customize Odoo.

5 min read
  • narrative
  • odoo

Odoo customization goes wrong in predictable ways: customizing to preserve old habits, building critical features without testing them, and ignoring the upgrade bill. A caveat first. Detailed public post-mortems of Odoo projects are rare, because companies rarely publish their failures. So this article draws on well-documented failures of other ERP platforms (SAP and Oracle) and pairs each with what Odoo’s own documentation says. The platforms differ. The mistakes are the same.

The short answer

Customize only where the process is a real competitive advantage. Keep standard behavior everywhere else. Treat every custom module as code you will maintain, test and port through every Odoo upgrade for as long as you run the system.

The cases in brief

CaseWhat happenedSource
Lidl (Germany)Abandoned an SAP-based inventory system in 2018 after reported spending of about 500 million eurosConsultancy.uk
Birmingham City Council (UK)Oracle ERP forecast at about 144 million pounds against an initial estimate of 19 million; being reimplementedThe Register
Hershey (US)1999 ERP go-live disrupted Halloween orders; figures vary by sourceCIO

Figures in public reporting differ between outlets and change over time, so treat them as indicative.

1. Bending the software to protect an old process

Press reports on Lidl’s project say the retailer valued inventory using purchase prices, while the standard SAP retail software used retail prices. Lidl reportedly preferred to adapt the software rather than change its method. The project was halted after roughly seven years (Consultancy.uk, RetailDetail). The exact causes are reported differently by different outlets, but the pattern is familiar.

Odoo lesson: Before you write a line of code, ask whether the process or the software should change. A written “keep standard unless it costs us money or customers” rule settles most arguments.

2. Planning “out of the box” and then customizing the critical parts

Birmingham City Council planned to use Oracle out of the box but built several customizations. One of them, a custom banking reconciliation system, did not work properly, and the council has since bought a third-party product for it (The Register).

Odoo lesson: Money handling (reconciliation, tax, payroll, inventory valuation) is the worst place for untested custom code. If standard Odoo or a well-maintained add-on covers 80 percent, take it and adapt the process around the gap.

3. Rebuilding every legacy customization

A 2023 Register report, citing an unnamed insider, says the previous SAP system had been heavily customised and the council was struggling to recreate those functions in Oracle (The Register, 2023). Old customizations tend to encode old workarounds, not current needs.

Odoo lesson: When migrating to Odoo, audit every legacy customization and sort it into three bins: needed, replaced by standard, or retired. Ask the users, not just the spec.

4. Ignoring the upgrade path

Odoo’s documentation is blunt: a database with custom modules “cannot be upgraded until a version of your custom modules is available for the target version of Odoo.” It also says that for major versions an upgrade is mandatory every two years, and that you can stay on a version indefinitely but it advises against it (Odoo upgrade documentation).

Odoo lesson: Every custom module adds to a recurring upgrade cost. Budget for it when you approve the customization, not when the upgrade arrives. Fewer, smaller modules mean cheaper upgrades.

5. Skipping the test upgrade

Odoo strongly recommends requesting an upgraded test database first and spending time testing it before upgrading production. It also notes that test copies come without the filestore, so you must merge the production filestore for realistic testing (Odoo upgrade documentation).

Odoo lesson: Keep a test plan of your key workflows (quote to cash, procure to pay, month-end close) and run it on every upgraded copy. A custom module that installs is not the same as a custom module that works.

6. Going live at the worst moment

Hershey’s system went live in 1999, just as Halloween orders were arriving, and was reported to have disrupted deliveries, with figures such as $100 million in unfilled orders cited, though sources disagree on the numbers (CIO).

Odoo lesson: Schedule go-live outside your peak season and month-end, and keep a rollback plan. Customizations raise the odds of late surprises, so leave slack for them.

7. Never re-baselining the budget

Birmingham’s initial estimate was reported as 19 million pounds plus a 1 million contingency. The forecast to 2027/28 is now about 144.4 million pounds, with the original go-live target of December 2020 long past (The Register). Few Odoo projects approach that scale, but the dynamic scales down: small overruns that nobody re-approves.

Odoo lesson: Put customization in a separate, capped budget line. Every change request gets a cost, an owner and a yes or no from someone with authority.

8. Treating no-code changes as maintenance-free

A reply on the Odoo community forum explains that Studio does not generate source code. It mainly creates field, view, report and action records stored in the database (Odoo forum). That makes small changes quick, but it can make them harder to review and track than code in version control. Community sources disagree on how portable Studio output is, so check your version’s behavior before relying on it.

Odoo lesson: Keep a register of every customization, no-code or code, with who asked for it and why. Undocumented tweaks are what make upgrades slow.

A practical checklist before you customize

  • Can standard Odoo, configuration or a maintained add-on do it?
  • Would changing the process cost less than the code and its upgrades?
  • Who owns this change after go-live?
  • Is it covered by a written test you can re-run on each upgrade?
  • Is it in a capped budget, with the upgrade cost included?

Talk it through

If you are weighing a customization, or inheriting a heavily customized system, we are happy to have a no-charge conversation with Teknologia Solutions. No pitch, just a second opinion on what to keep standard.


Sources: Consultancy.uk on Lidl; RetailDetail on Lidl; The Register on Birmingham City Council; The Register (2023) on Birmingham; CIO on Hershey; Odoo upgrade documentation; Odoo community forum on Studio. Cases involve SAP and Oracle, not Odoo, and are cited as third-party examples. Figures vary between reports; verify against primary sources before relying on them.