
Most software rollouts that disappoint do not fail because the technology is bad. They fail because nobody checked, in a structured way, that the system works the way real people work. A missing approval step, a report that does not match the ledger, a screen that freezes with real data volumes: these are ordinary problems that proper testing finds before go-live instead of after.
This checklist is written for teams rolling out an ERP or another business system. It follows the order we use in our QA and testing work: plan, test real scenarios, check data, performance and security, run user acceptance, and prepare the go-live and the first weeks after it. Use it as a working tool in meetings.
Key takeaways
- Define scope, owners and success criteria before testing begins.
- Test end-to-end business scenarios, including partial deliveries, several currencies and missing permissions.
- Reconcile migrated data and reports with the ledger.
- Check performance with realistic volumes, plus security, roles and restores.
- Hold user acceptance sessions with written sign-off, then plan go-live and early support.
1. Define scope, owners and success criteria
Testing needs a target. Before anyone opens a screen, agree on what is in scope and what is not. List the modules and processes that must work on day one, such as sales invoicing, stock movements, purchasing, payroll and month-end reporting, and note what can wait for a later phase.
Assign a named owner for each area, usually the department head or a power user. These owners write and approve the test scenarios and sign off at the end. Agree on success criteria in plain language, for example: every invoice type can be created, approved, printed in Arabic and posted correctly; the month-end close takes three days instead of ten.
Finally, decide how defects will be tracked. A shared list with severity, owner and status is enough. The point is that nothing is lost in chat messages and that everyone sees the same picture.
2. Test real scenarios end to end
Feature-by-feature clicking is not enough. Write scenarios that follow a transaction through the whole business, the way a customer or a supplier would experience it.
Examples worth testing: a customer order that becomes a delivery, an invoice, a payment and an aged-debt report; a purchase request that is approved, received into stock, invoiced by the supplier and paid; an employee hired, paid, given leave and terminated; a return that reverses stock and revenue correctly.
Include the awkward cases that happen in real life: partial deliveries, discounts, several currencies, cancelled documents, backdated entries and users who lack permission. Most defects hide in these corners. Run each scenario with the people who will use it, not only with the implementation team, because they notice what is unnatural or slow.
3. Verify data, migration and reports
Numbers are the soul of a business system, so verify them carefully. If you migrate data from an old system, compare totals before and after: customer counts, supplier balances, stock quantities and values, open invoices and the trial balance. Differences must be explained, not ignored.
Check master data quality. Duplicate customers, inconsistent item codes and missing tax details create problems that no software can fix. Clean them before loading, and decide who owns each list afterwards.
Then reconcile reports with the ledger. A sales report, a stock valuation and an aged-debt list should agree with the accounting totals for the same period. Test Arabic printouts and exports too: alignment, number formats, dates and currency symbols. If your finance team trusts the reports in testing, they will trust them in production.
4. Test performance and weak connectivity
A system that is fast with ten test records can crawl with ten years of real data. Test with realistic volumes: thousands of customers, items and transactions. Open the heaviest reports, run the month-end process and check how long searches take.
Then test the conditions your users actually face. What happens on a slow link? Can a branch keep working when the connection drops, and does it synchronize safely afterwards? What if the power fails halfway through posting a document? Our article on offline-first software explains the design patterns behind these behaviors.
Record response times for the key actions and agree on acceptable limits with the owners. Performance is a feature, and unclear expectations lead to disappointment even when nothing is technically broken.
5. Check security, access and backups
Business systems hold money, payroll and customer data, so security testing is not optional. Start with roles and permissions: confirm that each role can do only what it should. A cashier should not edit prices, a clerk should not approve their own purchase and an auditor should be able to read but not change.
Review authentication: strong passwords, locked accounts after repeated failures and a clear process for removing leavers. Check that sensitive data is protected in transit and at rest, and that actions are written to an audit log.
Finally, test the backup and restore process, not just the backup. Restore yesterday's data on a separate machine and check that it opens correctly. A backup that has never been restored is only a hope. Good hosting with monitored servers and regular backups makes this routine.
6. Run user acceptance testing
User acceptance testing, often called UAT, is where the business confirms that the system is ready. It is not another round of technical checks. It is the people who will live with the system trying their own work in it.
Give each owner a short script based on the scenarios you wrote, a test environment with realistic data and a deadline. Ask them to record what they tried, what happened and whether it is acceptable. Hold a short daily meeting to triage defects: fix now, fix before go-live, or accept and schedule later.
Sign-off should be explicit. Each owner confirms in writing that their area is ready, listing any known issues they accept. This protects everyone and prevents arguments after launch. It also builds ownership, because people who tested the system tend to support it.
7. Prepare the go-live and the first weeks
A calm launch is prepared in advance. Choose a go-live date outside peak periods such as month-end or stock-taking. Write a cut-over plan with times, owners and checks. Prepare a rollback plan that you have rehearsed, so a serious problem does not become a crisis.
Train users before launch, in short sessions focused on their daily tasks, in Arabic or English as they prefer. Prepare quick reference sheets and make the support contact visible on every screen or desk.
For the first weeks, staff extra support, review issues daily and keep a list of improvements for the next phase. Compare your success criteria with real numbers after thirty days. If the process was followed, most surprises will be small. If you need a modernization or integration step afterwards, see our application modernization service.
Get a second pair of eyes
Independent testing finds what the build team no longer sees. If you are about to roll out an ERP, a new platform or an upgrade, the MOMK QA team can review your plan, write scenarios with you, test performance and security and support your acceptance sessions. Read our guide to choosing a system in the ERP buyer's guide, learn about ERP applications, or contact us to arrange a consultation.
Frequently asked questions
How long should testing take for an ERP rollout?
It depends on scope, but a good rule is to reserve a meaningful share of the project timeline, often a quarter or more, for testing, fixing and user acceptance. Rushing it usually costs more after go-live.
Who should perform user acceptance testing?
The people who will use the system every day, such as finance, stock, sales and HR staff, led by an owner for each area. The implementation team supports but does not replace them.
What is the most common reason for go-live problems?
Poor data quality and untested real-life scenarios. Cleaning master data, reconciling reports with the ledger and testing awkward cases prevent most surprises.
Can MOMK test a system we did not build?
Yes. Our QA and testing service covers functional, performance and security testing for systems built by us or by others.
What is regression testing?
It means re-running earlier tests after a change to confirm that existing features still work. It prevents a new fix from breaking something that worked yesterday.
Related services, solutions and reading
- QA & Testing: Functional, performance and security testing that finds defects before your users do.
- ERP Applications: Finance, HR, inventory and customers on one centralized platform, shaped to each client.
- Application Modernization: Move legacy systems to cloud-native, scalable and secure architectures with minimal downtime.
- RGB Business Management: A complete ERP for businesses of every size: finance, HR, sales, procurement and inventory in one place, with business intelligence for better decisions.
- GOLD Accounting: Accounting software that automates financial tasks and gives a complete, accurate and compliant view of your company’s finances.
- How to choose an ERP system in Sudan: a buyer’s guide: Five questions to ask before you sign: from Arabic and multi-currency support to power cuts, training and local support.
- Modernizing legacy applications without stopping the business: A phased path from ageing systems to cloud-ready, secure and maintainable software.
- Latest industry news for Sudan and MENA: verified headlines on government, cloud, AI and security.
- Talk to the MOMK team: get a short consultation and a tailored plan.
Want help with this?
Our team can review your situation and propose a plan, in person in Khartoum or online.
Contact us


