ERP AI project failures are costing businesses millions as both small firms and large corporations struggle with AI in ERP systems, often seeing no real improvement. ERP and AI consultant Noel D’Costa says the root cause is often poor system selection, weak planning, and unclear goals before the work even begins.
Across industries, ERP AI project failures are burning through budgets. Both small firms and large corporations have run into the same problem: adding AI to ERP systems and ending up with little to show for it. ERP and AI consultant Noel D’Costa says it usually starts with the wrong system choice, poor planning, and goals that were never really clear in the first place.
There’s always pressure to modernise. In sales presentations, AI is presented as the simple answer with better forecasts, faster processes, improved decision-making. And for businesses already using SAP, Oracle, or Microsoft Dynamics, it sounds tempting to just add AI onto what they already have.
But that’s rarely how it works out.
These features usually demand significant customisation. Integration can be complicated, and costs often go up far beyond the original budget. It is not unusual for companies to spend heavily and still find that the promised improvements never arrive.
The gap between promises and results
ERP and AI can absolutely deliver value, but the work involved often surprises people.
“ERP works best when it’s targeting a specific problem, like demand forecasting or supplier risk,” says Noel. “AI can make that better, but only if the groundwork is solid first.”
Too many projects begin without clear targets or even a shared definition of success. Vendors describe features in ways that make them sound ready to go, but in reality, many of them require months of data preparation, process changes, and fine-tuning before they can deliver anything useful.
The costs and effort needed for that preparation are rarely mentioned during the sales stage. By the time companies realise what’s involved, the project is already in motion and the spending is climbing. Business leaders also say it’s common for the salespeople who closed the deal to vanish once the delivery stage begins.
ERP AI Challenges for Small Businesses
Large organisations often have their own ERP teams and technical experts who can adjust the system and work through problems. Smaller businesses usually do not. They rely on their vendor or a single consultant. If the solution doesn’t fit, fixing it can be difficult and expensive.
Noel recalls working with a small manufacturing firm that had invested in an AI-based demand planning tool. On paper, it made perfect sense. The vendor’s pitch was convincing. But once it was installed, the company realised their sales data was incomplete and inconsistent. The AI tool could not be trained properly, so it sat unused, while subscription fees kept coming in.
When Noel reviewed the setup, he found the ERP had never been configured to collect some important sales and inventory details in a consistent way. The AI wasn’t the problem. The data feeding it was.
He helped the company:
- Clean up the data they already had
- Redesign the way sales and inventory details were entered
- Put processes in place to keep information accurate over time
Only after that did they try the demand planning feature again, starting small and adjusting as needed. A few issues came up and were fixed quickly. Within a few months, the tool was being used regularly, and forecasts were noticeably better. It took longer and cost more than expected, but this time they were getting value back.
Vendors profit even when outcomes fail
ERP vendors and their partners are often paid no matter the outcome. Licensing, integration, and support fees still come in even if the system doesn’t perform as promised. For many companies, this creates a feeling of being tied into contracts that are hard to escape.
Ending those agreements early can be costly, so most carry on and try to get what value they can. Some have started to delay projects until they can see detailed, itemised pricing and clear delivery terms. Others bring in independent advisors to look over proposals before they commit.
According to Noel, this step can:
- Stop businesses from paying for features they don’t need
- Keep the focus on results that actually matter
- Give leaders better leverage in vendor negotiations
It may slow things down, but often that’s better than rushing into another project that drains money without delivering results.
Independent ERP Consulting as a Solution
Independent ERP consultants don’t work for a specific vendor. This means they can give advice without being influenced by sales targets. They assess whether a system or feature will genuinely solve the business problem and whether the organisation is ready to make it work. Sometimes the advice is to move forward. Other times, it’s to wait or change direction.
Noel D’Costa describes AI as just another ERP tool. It needs to be evaluated like any other investment, looking at the potential return, the total cost, and the effort required to make it work. Too many projects underestimate that effort.
Through ERPConsult.ai, Noel works with companies on ERP system selection, SAP ERP implementation planning, and integrating AI in a practical way without adding unnecessary cost. His personal site, NoelDCosta.com, offers free resources, guides, and case studies to help decision-makers prepare before they start.
Why ERP and AI Projects often fail
From feedback and case reviews, several issues appear repeatedly:
- Vague goals that are never measured
- Poor quality or incomplete historical data
- Teams not trained to work with AI outputs
- Heavy reliance on vendors for setup and optimisation
- Extra AI features added mid-project without checking the impact
These problems can be avoided with early planning and honest assessments of what the technology can actually do.
The Cost of Getting Implementations Wrong
ERP projects often run over budget and behind schedule. Adding AI makes that risk bigger. Some reports suggest more than 70 percent of ERP and AI projects fail to meet expectations, with budgets going more than 30 percent over. That can mean hundreds of thousands of dollars wasted for mid-sized businesses. Larger companies can lose far more.
Steps to Ensure ERP AI Projects are Successful
Before you move forward with an ERP or AI Project, Noel D’Costa advises business leaders to take a step back and prepare thoroughly. Based on his work with companies across five continents in SAP ERP implementation and consulting, he suggests these five steps that will greatly improve your chances of a successful outcome:
- Before engaging with any vendor, write down your key business issues in detail and discuss it with your team. Be really clear with what needs to be fixed and what outcomes you are expecting.
- Confirm that your historical data is accurate, complete, and available for the project. Spend some time to fix the data before you start any implementation.
- Ask your potential vendors for a small pilot or test exercise to see if their solution works, before committing to a full rollout.
- Request for detailed pricing from your vendor, that separates licenses, implementation, integration, and ongoing support. Clarify all the details and don’t hesitate to take your time.
- Work with an expert, like Noel, who is independent. These experts review the vendor’s proposal, scope, and success measures. Prior project experience can help to reduce your costs and increase your chances of success.
Taking these steps early can help companies can avoid cost increases, reduce implementation risks, and ensure their ERP investments lead to measurable improvements.
Looking Ahead
ERP and AI will keep developing. Vendors may improve their delivery over time, but for now, results are still mixed. Companies that take their time, question what they are told, and make sure they are ready before starting will stand a much better chance of success.
For more information on ERP selection and implementation, visit ERPConsult.ai. For resources and insights, see NoelDCosta.com.