Odoo: what it really costs, and where it breaks

Odoo appears in more of the quotes that land on my desk for review than any other system. It also produces more stalled projects than any other, and almost always for the same reason: the company bought a licence believing it had bought something else.

Two products, one name

The single most expensive misunderstanding in this market is that Odoo is one thing. It is two.

	Community	Enterprise
Licence	LGPLv3, open source	Commercial, closed
Cost	Nothing	Per user, per month
Source code	Yours to modify	Readable, not redistributable

Companies hear “Odoo is open source and free” and budget against the first half of that sentence. The second half is the entire subject of the cost section below.

The product itself is made by Odoo S.A. of Belgium, founded by Fabien Pinckaers. It first shipped in 2005 as TinyERP, became OpenERP in 2009, and took the name Odoo in 2014 when it grew past resource planning into websites, e-commerce and point of sale. The current release is Odoo 19, and a major version ships roughly every year. That last detail is not housekeeping — it becomes a recurring invoice, and I return to it.

Who it fits

From the projects I have delivered, Odoo sits in a defined band: companies from about five users to roughly three hundred, with several operations that touch each other, needing one system across sales, inventory, accounting and manufacturing instead of four applications joined by brittle integrations.

It genuinely suits distributors and traders whose inventory and accounting have drifted apart, small and mid-sized factories with straightforward work orders, services firms tying project to cost to invoice, and multi-branch operations inside a single country.

It does not suit advanced production planning, large distribution warehouses, or groups consolidating dozens of legal entities. Those cases are in the last section, stated plainly.

What it actually costs

The figures below were read from Odoo’s own pricing page on 5 September 2026, at US pricing.

Plan	Billed annually	Billed monthly
One App Free	$0 — one app, unlimited users	—
Standard	$16.90 per user/month	$21.10
Custom	$25.50 per user/month	$31.90

A promotional discount currently shows Standard at $13.50 and Custom at $20.40. Its condition is published in plain sight: twelve months, and only for the users on your first order. User twenty-one, added in year two, joins at full price, and so does all of year two. Build a three-year budget on the discounted figure and you will be out by close to a quarter.

The gap between Standard and Custom is not cosmetic. Custom alone carries hosting on Odoo.sh or your own servers, the Odoo Studio editor, multi-company, and access to the external API. Any company planning serious customisation, running more than one legal entity, or integrating with another system is on Custom. Put that in the budget on day one, not in month four.

Then comes the number that appears on no pricing page: implementation. Process analysis, configuration, data migration, custom reports, training and post-go-live support. Across the projects I have seen in the UAE, Saudi Arabia and Egypt, first-year implementation runs two to five times the annual licence, and higher with complex manufacturing or dirty data. The licence is the smallest figure in the file, and it is the only one most buyers negotiate.

What it does well

Coverage against price. Nothing else in this bracket covers accounting, manufacturing, e-commerce, point of sale and HR with this much real integration rather than integration by brochure.

A near-open customisation ceiling. The code is readable, the architecture is modular, and Studio lets a non-programmer change screens, fields and workflows. When a vendor tells you Odoo “cannot do that”, they usually mean they would rather not.

An interface people actually use. This is measured in no comparison matrix and decides more projects than everything that is. A storeman opens Odoo without a week of training.

Arabic is native, right-to-left included, in the core product rather than as a partner’s layer on top.

A deep partner market in Dubai, Riyadh and Cairo, which means you can replace your implementer — a feature nobody values until the day they need it.

Where it breaks

This section is why the page exists.

The Community trap. The free edition has no full accounting. It has invoicing. No bank reconciliation, no asset management, no complete accounting engine. A company picks Community because it costs nothing, builds on it for six months, and then the accountant discovers the missing half is behind the licence wall. By then you have paid for a full implementation to arrive at the decision you should have started from. The same applies to Studio, e-signature, marketing automation, quality, maintenance and the mobile app.

Pricing by headcount, not by use. Forty warehouse staff opening a screen twice a day cost what forty accountants cost. In labour-heavy companies this inverts the whole cost model, and the usual outcome is worse than the invoice: the company buys ten licences and shares passwords, losing the audit trail it bought the system for.

The annual upgrade bill. Custom code written against version 16 does not run on 19 without migration work. The more you customise, the larger this recurring bill, and almost nobody budgets it. Companies that ignore it freeze on an old release without security updates, then pay for a full project four years later to catch up.

Breadth at the cost of depth. Odoo does everything, and that is precisely the problem. Its manufacturing is not SAP’s, its warehouse is not a dedicated WMS, and its multi-entity consolidation is not NetSuite’s. For ninety per cent of companies this never matters. If you are in the other ten, you find out in month eight of the implementation, not in the demo.

Partner quality varies widely. The tiers — Ready, Silver, Gold — measure sales volume, not delivery quality. I have watched Gold partners deliver worse than a new partner with one serious engineer. Ask for a reference in your own industry, and speak to them without the vendor in the room.

E-invoicing is covered in Saudi Arabia and Egypt, and unmentioned for the UAE. The detail matters more than the verdict here. Odoo publishes its own connector to the Saudi platform: technical name l10n_sa_edi, described in the Odoo 19 documentation as enabling ZATCA phase 2 API integration. It is not behind the Enterprise wall — the file sits in Odoo’s community repository on branch 19.0 with LGPL-3 declared in its manifest. Egypt has l10n_eg_edi_eta for the tax authority. The UAE localisation page in the same documentation does not mention e-invoicing at all: it covers the VAT201 return, the corporate tax report and reverse charge, and nothing about the mandate. (All read on 6 September 2026.)

So ask about the UAE specifically, not about “e-invoicing” in general: who builds the connector, who maintains it, and who carries the risk when the specification changes. Make the answer a clause in the contract. And note the second trap: a module being free in Community does not make running it free — you still need the digital signing certificate, registration with the authority, and somebody’s hours to build and test the integration.

Learning it

Open a trial on Odoo Online and switch on sales and inventory only. Then run one complete document cycle by hand — purchase order, receipt, sales order, delivery, invoice, payment — and watch the journal entries each step produces. Anyone who understands that chain understands Odoo. Odoo’s own eLearning platform is free and covers each app functionally; the developer documentation comes after, not before.

The functional path always precedes the technical one. People who learn the code before the document cycle build elegant solutions to problems that do not exist.

A fuller version of this path — ten stations, each with a concrete output and an acceptance test — is published in Arabic as the Odoo learning path.

The certification

The official exam is the Odoo 19 Functional Certification, and the terms are on Odoo’s site: $250 (currently offered at $220), 120 questions, 90 minutes, pass at 70%. Marking is the part that fails candidates: a correct answer scores one point, a wrong answer scores minus half a point, and a blank scores zero.

Random guessing therefore has a negative expected return. Leave what you do not know empty — advice that contradicts every other exam you have sat. The certificate is issued to a person, not a company, so it travels with you.

It opens the interview; it does not close the hire. What closes it is showing an interviewer a system you configured yourself and walking them through a document cycle on screen.

When to choose something else

Be honest with yourself. Odoo is the wrong answer if your manufacturing needs finite capacity scheduling at work-centre level, if you run a high-volume distribution warehouse with wave picking, if you are a multi-entity multi-currency group consolidating monthly, if your first-year budget including implementation is under $10,000, or if you have fewer than five users and simple operations.

Outside those five cases, Odoo earns a place on the shortlist — provided you walk in carrying the implementation number, not the licence number.