Claims that "AI cuts ERP costs by such-and-such a percent" say little on their own; the rates vary from company to company and from project to project. What is meaningful is not the percentage but the mechanism: which tasks in an ERP project does AI make cheaper, and which does it, on the contrary, make valuable? This article examines the impact of AI on ERP implementation cost through this question, stage by stage.
The Hidden Effort Map of a Classic ERP Project
In a traditional ERP project, the major effort items are well known: testing, data migration and mapping, documentation, training, customisation code. What they have in common is that they are mechanical tasks that come after the decisions have been made. Discovery and analysis, by contrast, are a small slice of total effort — and this smallness is an economic choice, not a technical one: in the man-day model, discovery is the hardest stage to invoice. The customer thinks "nothing has been installed yet"; the vendor, too, wants to shorten discovery and move on to the billable implementation.
The result is the familiar picture: the most critical decisions — scope, process design, the limit of customisation — are made in the weeks when the least is known. The value curve is shaped accordingly: throughout months of implementation, the value produced is close to zero; everything is piled into the steep rise at the moment of go-live. In the industry, this curve is called the "hockey stick".
What AI Really Does: It Changes the Effort Map
According to BCG’s analysis of generative AI and ERP, the largest savings estimates always cluster in the same area: testing (an estimated time saving of 60-70%), the production of training and documentation, data cleansing and mapping, and standard configuration. McKinsey’s ERP research puts forward an even more ambitious estimate: AI agents could cut ERP implementation effort by at least half. Both of these figures are forecasts, not measured industry averages — but the mechanism they point to is sound: repeatable, rule-based, high-volume tasks are the ones best suited to automation; the late stages of ERP are largely of this kind.
Stage by stage, the picture looks like this:
| Stage | Impact of AI | What is happening? |
|---|---|---|
| Test scenarios | Getting cheaper | Generation and execution are being largely automated |
| Documentation and training content | Getting cheaper | Content production is being automated |
| Data cleansing and mapping | Getting cheaper | Mapping suggestions are automatic, approval stays with a human |
| Standard configuration | Getting cheaper | Fast set-up from templates |
| Discovery and process analysis | Gaining value | Becoming the stage that determines the quality of decisions |
| Scope and design decisions | Gaining value | Can now be made with the most information |
| Human approval / oversight | Gaining value | The safety condition of automation |
To put it in terms of the X-ray analogy: the machine has become cheaper — taking a film is now fast and inexpensive. But when everyone can take a film, being able to read the film correctly has become the real differentiating skill.
The "Hockey Stick" Is Flattening: Resource Shifting
The name BCG gives to this picture sums up the mechanism: resource shifting. The effort gained in the late stages does not vanish; it is transferred to the early stages. Discovery stops being a "formality that is hard to invoice" and can be re-priced as the highest-return investment in the project — because every hour invested in discovery now determines the quality of all the mechanical stages that have become cheaper. Value, too, starts accumulating from the first weeks instead of exploding at the moment of go-live: the hockey stick is flattening.
There is a measured example too: in Deloitte and UiPath’s SAP S/4HANA migration, more than 200 automations were deployed in core processes, more than half of the test scenarios were run automatically and a 93% "clean core" rate was achieved. A single case, from a vendor source — not enough to generalise from; but a sign of the direction.
The Condition for the Mechanism to Work: Measurement First, Then Automation
This mechanism has a hidden precondition: automation inherits the quality of the ground it is built on. The three root causes in the second article of the series return precisely here: test automation on dirty data tests the wrong thing fast; standard configuration on unmeasured processes sets up chaos fast; and unaudited automation scales the error. The savings in the stages that are getting cheaper materialise only if discovery and diagnosis have been done correctly. In the new model, money flows to the stage that used to be considered "free" — diagnosis.
The 5 Rules of the New Economics
- Mechanical work gets cheaper: testing, documentation, training content, data mapping — do not pay the full man-day rate for these items.
- Diagnosis gains value: discovery is no longer a formality but the highest-return investment in the project.
- Sequence is everything: measurement first, then automation; the reverse accelerates chaos.
- Early decision = cheap decision: scope and design decisions should be pulled to the moment when the most information is available.
- Human approval is not negotiable: accelerated analysis does not make decisions; the physician who reads the film must stay inside the model.
Conclusion: The Right Question Has Changed
The question in the first article of the series — "is your consultant selling hours or outcomes?" — has become clearer with this mechanism. The new question at the proposal table is this: "How much of my money goes to mechanical work, and how much to diagnosis?" If the rate for mechanical work is falling while the value of diagnosis is rising, the weight of the budget should shift in the same direction. The final article in the series will examine what this approach looks like in practice: how an ERP journey that says diagnosis first progresses, step by step.
This article is the third in the growing «From Man-Days to Value» series.
1. Is Consulting That Sells Man-Days Coming to an End?
2. Let’s X-Ray the Promises of AI
3. What Has AI Made Cheaper in ERP, and What Has It Made Valuable? (this article)
4. Diagnosis First: What Does the New Path in ERP Look Like in Practice?
5. The Unmeasured Layer of Digital Transformation: Leadership
6. Why Do Reports
Stay on the Shelf in Digital Transformation?
7. (coming soon)
The final article in the series shared a free tool with which you can see your own ERP risk in 10 questions: Diagnosis First: What Does the New Path in ERP Look Like in Practice? To make sure you do not miss new analyses, subscribe to the MindDX-Digital Excellence newsletter.
