By Lee Clarke, Senior Business Director
– Last Updated: June 2026
Most organisations treat ERP risk as an implementation problem. The data tells a different story. By the time go-live pressure builds, the mistakes that matter most have usually already been made.
In every ERP conversation I have with technology leaders and transformation professionals, there’s a moment where the same uncomfortable truth surfaces. The most expensive mistakes in an ERP programme rarely happen during configuration, testing, or even go-live. They happen much earlier than that, during Discovery, Planning, and Vendor Selection, when everything still looks manageable and the real complexity hasn’t yet revealed itself.
This is the phase where business requirements are gathered, processes are analysed, technology options are evaluated, governance structures are established, and the programme roadmap is built. Get it right, and the programme has a genuine chance of delivering transformational value. Get it wrong, and organisations tend to spend the following 18 to 24 months firefighting delays, budget overruns, scope creep, and outcomes that fall well short of what was promised.
What’s changing the game right now is AI. Used well, it can fundamentally accelerate how organisations approach this critical early phase. But, and this is the part that matters most in my experience, AI alone isn’t enough. The organisations seeing the best results are the ones pairing AI-powered insight with experienced ERP and transformation professionals who know how to interpret the outputs, challenge the assumptions, and turn data into decisions that actually stick.
Discuss your plans with our experts
Traditionally, understanding how a business truly operates meant months of workshops, interviews, documentation reviews, and process mapping exercises. It was thorough, but it was slow, and by the time you had a clear picture, the urgency of the programme had already started to build.
AI can now analyse process documentation, transaction histories, operational workflows, support tickets, and business data to surface patterns and inefficiencies far more quickly than traditional methods would allow. Process bottlenecks, manual workarounds, duplicate activities, reporting gaps, integration challenges, automation opportunities; a lot of that can be identified early and with a level of consistency that manual discovery struggles to match.
The important distinction:
Knowing the difference between a process that needs redesigning and one that simply needs to be used properly is the kind of judgement that only comes from having been in the room when those decisions have gone both ways.
It’s been true for years, and it remains true now: vague, incomplete, or conflicting requirements at the start of a programme create problems that compound at every subsequent stage.
AI is increasingly being used to analyse workshop outputs, meeting transcripts, process documentation, and stakeholder feedback to identify common themes and requirements across departments far more efficiently than manual synthesis allows. That means organisations can capture requirements more completely, spot conflicting stakeholder needs earlier, highlight gaps before they become risks, and build more comprehensive requirement catalogues in less time.
Where specialist ERP interims remain essential is in validating those findings. Requirements that look comprehensive on paper can still be shaped by legacy thinking, where the instinct is to replicate what already exists rather than to define what the business actually needs going forward. Experienced professionals catch that, and redirect it.
Choosing the right ERP platform has always carried significant weight, and the consequences of getting it wrong tend to be felt for years. AI is genuinely helping here, by enabling more systematic evaluation of functional fit, industry-specific capability, implementation complexity, integration requirements, vendor performance history, and total cost of ownership.
That gives leadership teams the ability to assess options more quickly and with greater objectivity than a series of vendor demonstrations typically allows.
But vendor selection is rarely a purely data-driven exercise. Experienced ERP professionals bring something that analysis alone can’t: practical implementation experience with the platforms being evaluated, the ability to challenge vendor claims in the room, a realistic view of implementation partner capability, and an instinct for the risks that never show up in a demo environment. That combination of data and experience is where the best vendor decisions get made.
AI is increasingly being applied to project planning and risk modelling, drawing on data from previous ERP programmes to help predict delivery risks, resource constraints, timeline pressures, budget challenges, and dependency conflicts before implementation begins. For leadership teams trying to build realistic roadmaps, that level of early visibility is genuinely valuable.
Paired with ERP programme leaders who have delivered multiple transformations, organisations can build plans that are grounded in reality rather than optimism, and avoid many of the pitfalls that tend to emerge later when course correction is far more costly.
None of this is an argument against AI. The acceleration it brings to Discovery and Planning is real, and the organisations ignoring it are putting themselves at a disadvantage. But ERP implementations are fundamentally business transformation programmes, and they involve people, politics, competing priorities, and decisions that carry significant organisational weight.
AI can process information faster. It can identify patterns more consistently. It can support better-informed decision-making. What it cannot do is align stakeholders who aren’t aligned, challenge assumptions that nobody wants challenged, navigate the organisational dynamics that sit underneath every programme, or build the kind of consensus that transformation requires.
That is still where specialist ERP and transformation professionals earn their place. The best ones understand both the technology and the realities of delivery. More importantly, they know how to use AI-driven insight to sharpen their own judgement rather than replace it.
Discovery and Planning has always been the most important phase of an ERP implementation. AI is making it faster, smarter, and more data-driven than it’s ever been. But the organisations that get it right aren’t the ones with the best tools. They’re the ones with the right combination of technology and expertise guiding the process from the beginning.
When ERP programmes fail, it’s rarely because the technology wasn’t capable. It’s because the decisions made at the start were built on incomplete information, unrealistic assumptions, or insufficient experience.
If you’re approaching an ERP programme and want to talk through how to structure the Discovery and Planning phase effectively, or what good looks like in terms of the interim and specialist talent that can make the difference, feel free to connect or message me directly.
Talk to us about how to get your ERP change right
Lee Clarke – Senior Business Director
Technology & Business Transformation
