Start from obligations, not features
Feature lists converge: every ERP has invoicing, inventory and reporting. What differs is how well the system fits obligations you cannot negotiate — e-invoicing, VAT, Zakat, payroll rules, and whatever your sector's regulator requires. Score those first; a system that is delightful and non-compliant is not a candidate.
Compliance depth, tested rather than claimed
"ZATCA compliant" is written on every brochure in the market. The differences appear one level down:
- Which document types has the vendor actually cleared and reported in production — invoices, credit notes, debit notes, standard and simplified?
- What happens when ZATCA is unavailable during trading hours?
- Does the system generate the VAT return from the same ledger it posts to, or from a separate report that can drift?
- Are corrections handled as credit and debit notes with an audit trail, or by editing history?
Arabic is a functional requirement
Arabic support ranges from a translated menu to a genuinely bidirectional system. The gap shows up in the places that matter: right-to-left screens that mirror properly, Arabic on printed invoices and thermal receipts without broken glyphs, Arabic customer and product names that survive search and export, and reports that stay readable in both directions.
Test it with your own data, not the demo dataset. Ask for a printed invoice with an Arabic customer name, an Arabic product description and a discount line — that one document exercises most of what usually breaks.
Deployment, residency and exit
Decide where data must live before you shortlist, because it removes options. Some sectors and contracts require production data inside the Kingdom; some organisations require on-premise or private cloud outright. Ask where backups live, too — data residency that stops at the primary database is not residency.
Then ask the exit question early, while you still have leverage: can you export your data — transactions, master data, signed invoice XML — in a usable format, on demand, without paying for a project? The answer tells you a great deal about the relationship you are entering.
Total cost is rarely the licence
Compare implementation, data migration, training, integration with the systems you are keeping, support response times that are contractual rather than aspirational, and the cost of change once you are live. A licence that looks cheap alongside a rigid implementation model is usually the expensive option over three years.
Run a real pilot
The most reliable evaluation is narrow and real: one branch, one month, your own products and customers, your busiest day, and a close of period at the end of it. Demonstrations are built to succeed. A pilot that survives a month-end close and produces a VAT return you can reconcile tells you what a demonstration cannot.
Common questions
Where should an ERP evaluation in Saudi Arabia start?
With obligations rather than features. Feature lists converge — every ERP has invoicing, inventory and reporting — so what differs is how well the system fits the things you cannot negotiate: e-invoicing, VAT, Zakat, payroll rules and whatever your sector's regulator requires. Score those first, because a system that is delightful and non-compliant is not a candidate.
Every vendor says they are ZATCA compliant. How do we test that?
The differences appear one level down, so ask four things. Which document types has the vendor actually cleared and reported in production — invoices, credit notes, debit notes, standard and simplified? What happens when ZATCA is unavailable during trading hours? Does the system generate the VAT return from the same ledger it posts to, or from a separate report that can drift? And are corrections handled as credit and debit notes with an audit trail, or by editing history?
How do we test whether a system's Arabic support is real?
Arabic support ranges from a translated menu to a genuinely bidirectional system, and the gap shows in right-to-left screens that mirror properly, Arabic on printed invoices and thermal receipts without broken glyphs, Arabic customer and product names that survive search and export, and reports readable in both directions. Test it with your own data rather than the demo dataset: ask for a printed invoice with an Arabic customer name, an Arabic product description and a discount line, because that one document exercises most of what usually breaks.
What should we ask about data residency and getting our data back?
Decide where data must live before you shortlist, because it removes options — some sectors and contracts require production data inside the Kingdom, and some organisations require on-premise or private cloud outright. Ask where backups live too, since residency that stops at the primary database is not residency. Then ask the exit question early, while you still have leverage: can you export your transactions, master data and signed invoice XML in a usable format, on demand, without paying for a project?
What is the most reliable way to evaluate an ERP before committing?
A narrow, real pilot: one branch, one month, your own products and customers, your busiest day, and a close of period at the end of it. Demonstrations are built to succeed, whereas a pilot that survives a month-end close and produces a VAT return you can reconcile tells you what a demonstration cannot. Compare total cost the same way — implementation, migration, training, integration and the cost of change once you are live, not the licence alone.
This guide is general information, not tax, accounting or legal advice. Rules differ from country to country and change over time; confirm the current position with your tax authority or a qualified adviser before acting on anything here.
Ready to run your operation on a single workspace?