CEO Analysis: The Truth Behind ERP Cost Overruns

According to Godlan's 2025 research on discrete manufacturing, ERP projects average 215% cost overruns: meaning a project originally budgeted at $10 million actually costs $31.5 million. This isn't a story of individual "screw-ups"—it's the statistical average across the entire industry, repeatedly validated in 2020–2025 data. For any CEO or board member, the real question is no longer "Will we overrun?" but "To what extent will we be held hostage by overruns?"—and who will pay for that 215% gap.

Cost Overrun Experiences: Hershey, Waste Management, Revlon

Hershey: Spent $112 Million, Lost $100 Million in Orders

In (1999), Hershey invested approximately $112 million across SAP, CRM, and supply chain systems, hoping to go live before Y2K. From the board's perspective, this was an investment "to lay the foundation for the next decade": unified systems, improved visibility, reduced supply chain risk. The result:

During the Halloween and Thanksgiving peak season, Hershey couldn't fulfill approximately $100 million in orders—inventory in warehouses, orders in the system, but processes so broken they couldn't deliver.
Beyond direct (revenue loss), the company's stock price came under significant pressure during the incident, and management faced intense internal scrutiny.

From a pure technical perspective, Hershey's problems can be simplified as: insufficient testing, wrong go-live timing, inadequate employee training. But from an executive's perspective, the real tension lies in:

Who is responsible for the "must go live before Y2K" time assumption?
Who decided to accept compressed testing and training to maintain the original timeline?
When the project team raised risk warnings internally, were they heard?

The first management question that emerges here is: When time pressure conflicts with risk assessment, who has the authority to hit pause? Without a risk owner with real veto power, "on-time go-live" often gets rationalized in board presentations, while underlying testing and training gaps become revenue and brand costs paid weeks later.

Waste Management: From "Out-of-the-Box" to $500 Million Lawsuit

Waste Management's SAP project is a classic case of cost overruns and risk misjudgment:

The company alleged that SAP sold them a solution "tailored for waste management, validated in customer environments," only to discover the product had "never run in a production environment."
The project itself cost over $100 million in implementation and related costs, plus a $350 million claim for "lost commercial benefits" due to unrealized expected returns, bringing the total lawsuit amount to $500 million.

In this story, there are several key decision points:

What assumptions did the board approve? "Proven product"? "Industry-standard solution"?
Did the company have a second opinion from both technical and business perspectives to verify claims like "out-of-the-box"?
After the project spiraled out of control, how did management weigh "continue investing vs. cut losses"?

This case exposes not just vendor sales narrative issues, but gaps in board governance structure: Who's guarding the gate? When a project with annual investments in the hundreds of millions has its risk due diligence primarily led by interested SIs and vendors, while the company lacks truly independent technical and industry advisors, cost overruns are just a matter of time.

Revlon: $64 Million in Unfulfilled Orders Becomes Shareholder Class Action

Revlon's SAP project brought "cost overruns" to the board's legal liability table for the first time in the form of investor class action lawsuits:

In 2018, after Revlon went live with SAP at its North Carolina factory, system issues prevented order and inventory processing, leading to approximately $64 million in net sales that couldn't be fulfilled.
In SEC 10-Q reports and earnings calls, the company admitted "material weaknesses in internal controls" during implementation, and spent an additional approximately $54 million in extra costs in 2018 to restore service levels.
Multiple law firms subsequently filed class action lawsuits against Revlon on behalf of investors, questioning its actions in project planning, oversight, and disclosure.

For CEOs and CFOs, the question is no longer just "project failure," but:

When did ERP projects shift from "operational risk" to "securities law and governance risk"?
When internal warnings appeared, did leadership communicate promptly to the board and market, rather than pushing risks to the next quarter?

Revlon's case essentially hung a sign at every public company's door: ERP cost overruns and failures are no longer just a CAPEX nightmare, but events that penetrate stock prices and legal liability.

The Five Components of Cost Overruns: Where Does the Money Actually Go?

Godlan's data tells you "215% average overrun" but doesn't clarify where this money actually flows. From extensive case studies and consulting practice, it can be clearly broken into five structural parts: delays, customization, training and change management, hidden costs, and opportunity costs. Behind each lies a management decision. (godlan)

1. Delays: Every Month Burns Money, But Few Calculate It Clearly

According to 2025 manufacturing ERP research, approximately 64% of projects experience budget overruns, most accompanied by 6–12 months or more of delays. Each month of delay doesn't just burn consultant fees: ((Manufacturing ERP Cost))

Full-time project team → Monthly internal labor costs can be $200,000–500,000.
External SI consultant daily rates $1,500–2,500, 10–20 person teams, one month is $300,000–1 million.
Change freezes, other projects paused → Delayed operational improvements and innovation projects.

Take a project originally planned for 18 months at $10 million:

Reality often becomes 30 months, over $30 million.
The extra 12 months, if monthly burn rate is $400,000–800,000, quickly stacks up an additional $4.8–9.6 million.

But in most board presentations, delays are described as "to ensure quality, proceed cautiously," rather than an ongoing, underestimated cash flow risk.

The real management question here is: Who is responsible for putting the cash cost of "each month of delay" in black and white on the decision table?

2. Customization: From "Necessary Advantage" to "Habitual Dependency"

Statistics from Software Path and multiple consulting firms show that 90% of ERP projects do some level of customization. Light customization itself isn't a problem, but three scenarios quickly devour budgets:

Treating "we've always done it this way" as a process advantage that must be preserved.
Letting each BU request "dedicated processes," turning ERP into "a collection of multiple systems."
Rushing to develop custom features when requirements are still unstable, leading to repeated rewrites.

Each custom feature, from specs to development, testing, and training, costs $50,000–250,000 individually. Large integrations or complex logic can reach $500,000–1 million. For a manufacturing group with 10–20 production sites, it's easy to accumulate 50–150 custom items:

50 × $100,000 = $5 million
150 × $200,000 = $30 million

Most CEOs hear when approving budgets: "We'll only do limited customization, mainly configuration." But a year later, reality shows:

Any small process change triggers a chain of custom code.
Every version upgrade requires retesting and patching these customizations.
The originally estimated $10 million implementation cost adds another $1–3 million annually in maintenance.

What's trickier: 30–40% of these customizations are actually just to make the system look like the old system—stemming from user anxiety about change, not competitive advantage needs. This turns "customization" from a strategic choice into expensive psychological comfort.

The management question here is: Who has the authority to say "no"? Who can clearly define: what kind of process differences are true strategic differentiation, worth protecting with customization?

3. Training and Change Management: First Cut from Budget, Most Expensive Pit on Site

Research from McKinsey, BCG, and others on large transformation projects repeatedly points out: Change management and capability building are the key differentiators between success and failure. But in actual ERP budgets, training and change management often only account for 3–5%, far below the 15–20% recommended in successful projects.

In Hershey's case, one key mistake was:

Under compressed timeline pressure to meet Y2K, training and system testing were reduced.
Result: System technically went live, but operators unfamiliar with processes, errors amplified into supply chain paralysis.

Revlon admitted in its 10-Q that the SAP project exposed "material weaknesses in internal controls," one of which was insufficient controls related to process design, responsibilities, and training.

From a financial reporting perspective, these costs are usually buried in "Other operating costs," rarely broken out independently in ERP project summaries.

But from the organization's real experience, this is the most painful part:

Frontline supervisors caught between new systems and old processes.
IT departments become "customer service centers," handling endless operational and process issues.
Departments that were supposed to be freed by ERP are instead slowed by the new system.

The real question here is: Who is responsible for "how much time and support people need to digest this change"?

If this responsibility has no clear owner, training budgets will always be compressed in early presentations, then come back to collect at higher hidden costs after go-live.

4. Hidden Costs: The Half Not Written in ERP Budgets

Many companies think they've done "complete TCO (Total Cost of Ownership) calculations," but 2025 manufacturing surveys show: 78% of enterprises underestimate ERP 3–5 year operational costs by 30–50%. These hidden costs are usually scattered across various budget line items, appearing unrelated to ERP:

Additional cloud or hardware resources (reporting, integration middleware, security modules).
Additional third-party tools purchased to compensate for what ERP does poorly.
New internal IT positions and teams to "maintain" this system.
Peripheral systems and processes modified to accommodate ERP.

For a mid-to-large discrete manufacturing group, Godlan's data indicates that nearly 40% of projects have "actual five-year total costs" more than double original expectations. The problem is that these costs are never viewed as part of the same decision package, leading to:

At project initiation, it looks like a $15 million project.
Five years later, total spending scattered across department budgets may approach $30–40 million.

The key question here is: Which version do the board and investors see? The presentation that only includes license + SI fees, or the true TCO picture including hidden operational costs?

5. Opportunity Costs: The Options Squeezed Out by ERP Projects

The final part, often clearest to CFOs but least discussed in project presentations: opportunity costs.

Godlan's research indicates that nearly 4/5 of manufacturing enterprises freeze or delay other critical investments during ERP projects, including capacity expansion, automation, product line expansion. This means:

During ERP projects' extended 12–24 months, companies actually bind "investment budgets and senior management attention" to a bet with a 25–30% success probability.
The same capital, the same mid-to-senior managers, if used for production line automation, lean improvements, or new market expansion, might yield more predictable returns.

But in practice, ERP is often packaged as "the foundation transformation must have," so all opposition gets simplified to "you don't understand long-term thinking."

This pushes CEOs and boards into an awkward position:

Not doing it seems like "abandoning modernization."
Doing it means accepting that many other important projects will be forced to yield over 2–3 years.

The question that really needs to be clarified here is: Under the same capital and management attention budget, is ERP really the option with the highest risk-adjusted return right now?

Why SI Incentives Almost Guarantee Overruns

If we view ERP projects as a value chain: software vendors, system integrators (SIs), external consultants, internal IT, business units—each party's revenue structure is somewhat decoupled from "cost control."

SI Revenue Model: More Hours = More Revenue

Most large SIs' business models still center on "person-days × rates":

Consultant daily rates $1,500–2,500, 10–30 person teams, monthly bills are $300,000–1.5 million.
A 6-month project delay is pressure for clients, but an additional $1.8–9 million in revenue for SIs.

Interviews by Godlan and multiple consultants indicate that approximately 60–70% of ERP project implementation costs ultimately go to external partners. Under this structure:

Every scope expansion, every new customization, every "for stability" extension directly increases SI short-term revenue.
Conversely, approaches that truly help clients shorten timelines, reduce customization, minimize on-site rework directly compress SI bills.

From an Incentive Design perspective, this is almost a textbook mismatch:

What clients want is "go live within controllable budget and achieve benefits."
SI's actual incentive is "keep the project in high hours, high complexity state as long as possible."

Of course, this doesn't mean all SIs "maliciously delay." More common is a frog-in-boiling-water dynamic:

Initial estimates biased optimistic (to pass approval), budgets compressed.
During go-live, encountering real complexity, scope gradually increases.
Every change can be reasonably packaged as "for your long-term benefit."
Finally, while the project overruns, the narrative becomes "original requirements were unreasonable."

From a CEO's perspective, what really needs to be asked is: Under this contract structure, is there any party that truly maximizes benefits from "helping you succeed within budget"? If not, that 215% average overrun is just a mathematical outcome.

Budget vs. Reality: The Two Columns That Don't Appear in Board Presentations

Most ERP projects, at initiation, present boards with a very "clean" table:

License & Subscription: X
Implementation Services: Y
Contingency (usually 10–15%): Z

Godlan's data shows that discrete manufacturing's actual five-year costs average 2–2.5x the original initiation budget. If we lay this gap on a CFO's desk, we'd see a more realistic comparison table (using original budget of $15 million as example):

Initial approved budget (what was told to the board):
- License & Core Implementation: $15 million
Five-year actual spending (accounting items scattered across different cost centers):
- Initial implementation: $15–20 million
- Additional customization and changes: $5–10 million
- Additional third-party tools and integration: $3–8 million
- Internal labor and operational costs: $5–10 million
- Operational disruption and efficiency losses: Difficult to fully monetize, but manifested as "hundreds of millions in revenue" for Hershey and Revlon.

When these numbers are reduced to a question, it's devastating for any executive:

If you had known that this project's true cost five years later would be $30–40 million, with only a 25–30% success rate, would you still have made the decision the same way?

A Necessary Structural Reminder

Records of the Grand Historian has a saying:

"Those who bear responsibility carry the weight; those who hold office cherish their salary."

—Meaning, those who shoulder the work and those who receive the salary are often not the same people.

Applied to ERP, the specific management implication is:

Those truly bearing daily operational risks (factory managers, supply chain managers) are not the ones who voted yes in the boardroom.
When SIs, vendors, and internal project offices gain short-term benefits from projects, those facing Execution Debt and Organizational Drag for the next 5–10 years are the people left to run the business.

This structural misalignment won't disappear by switching to cloud ERP or adding AI modules.

In the next subsection and subsequent power structure sections, we'll further see: cost overruns aren't a single decision error, but an inevitable result stacked by an entire system of incentives, responsibilities, and narratives.

And right now, for any CEO looking at an ERP investment presentation, there's really only one tricky question:

In this game, is there any role that truly gets the most reward for "helping you succeed within budget"?

If the answer is no, then that 215% average overrun is no longer just a statistic—it's the default script.