Business Transformation, Technology
When a new system goes live and things start to go wrong, people tend to focus on the launch itself.
In reality, the warning signs are usually there much earlier.
A finance process does not quite reconcile. A workflow works fine in a test environment but breaks when users start using it properly. An integration holds up under normal volumes but struggles when demand picks up. A report looks right at first glance, but the numbers are not reliable.
These are the sorts of issues that can cause real problems once a system is live. Teams lose time fixing things that should have been picked up earlier, users lose confidence, and the business can end up spending far more than expected to put things right.
That is why testing matters so much. It is not just the last job to get through before go-live. It is where you find out whether the new system will actually support the business day to day.
AI is starting to make a real difference here, particularly on large programmes where there is a huge amount to test and limited time to do it.
Anyone who has worked on a major system change will know how quickly the list of things to test grows.
There are business processes, approvals, reporting, integrations, data, exceptions and all the scenarios that were not included in the original plan but happen every day in the real world.
Creating and maintaining test scripts has traditionally taken a lot of time. AI can now help teams work through requirements, process documents, user stories and previous issues to suggest test scenarios and highlight areas that may have been missed.
That does not remove the need for proper testing expertise, but it can help teams spend their time where it is needed most.
If a process affects payroll, customer orders, supplier payments or month-end reporting, it needs more attention than something with limited business impact. AI can help identify those higher-risk areas sooner.
The biggest benefit of good testing is simple. It is much easier to fix an issue before go-live than after it.
AI can help teams spot unusual behaviour, inconsistent configuration, failed transactions and recurring problems while testing is still underway. It can also help make sense of a large volume of defects, rather than leaving teams to manually work through every issue and decide what needs attention first.
Integrations are a good example. Most businesses do not run one system in isolation. Their core platform needs to connect with finance tools, payroll, CRM, reporting, banks, suppliers and other systems.
That is often where problems hide.
A technical issue may look minor on paper, but the real question is what it means for the business. Will invoices go out? Will stock move? Will payments be processed? Will the right information reach the people making decisions?
This is where experienced people still matter. AI can flag a pattern or a potential issue, but someone with practical programme and business experience needs to decide what it means and what to do next.
User Acceptance Testing is usually where the real picture starts to emerge.
The system may be working as it was designed, but that does not always mean it works in the way people need it to. Users bring the day-to-day reality into the process. They know the workarounds, the exceptions and the pressure points that are easy to miss in a project plan.
AI can support this by helping create more realistic test cases, identifying gaps in coverage and reviewing common themes in user feedback.
But it cannot replace people who know the business.
The best testing teams combine technical specialists with people who understand the processes being changed. You need strong functional knowledge, clear ownership and people who are confident enough to challenge assumptions when something does not look right.
From a recruitment point of view, this is often where organisations are under-resourced. They may have a solid project team, but lack the experienced test lead, functional specialist or interim programme support needed to bring the right level of challenge and structure during the final stages.
Testing is not just about clearing a list of defects. It is about making an informed decision on whether the business is ready.
AI can give a clearer view of what is happening across the programme by looking at test completion, outstanding defects, process coverage, integration stability and feedback from users.
That is useful, but go-live decisions still need judgement.
Leaders need to know whether finance can close the month, whether customers can be served properly, whether suppliers can be paid and whether users are ready to work in the new system.
Those are business questions, not just technical ones.
AI can help teams test more thoroughly and spot risks earlier. It can make a difficult process faster and more manageable. But the strongest results come from combining it with people who know what good looks like and understand the impact of getting it wrong.
A successful go-live is not about proving that the system works in theory.
It is about knowing the business can rely on it when it matters.
Lee Clarke – Senior Business Director
Technology & Business Transformation
