7 public sector ERP failures in the U.S. that cost billions
Public sector ERP failures cost U.S. taxpayers more than $3.5 billion in the last two decades, and that only counts the projects big enough to make headlines.
The U.S. Air Force burned through $1 billion on a logistics ERP that produced almost no working software.
New York City’s payroll modernization ballooned from $63 million to $760 million and ended in federal fraud charges.
California’s state controller terminated its $373 million payroll project after seven years and $254 million spent.
And the list continues.
This post walks through the 8 most instructive public sector ERP disasters in U.S. history.
For each one, I’ve noted what went in and what came out as a result.
If you’re planning or currently inside a public sector ERP implementation, this is the cheat sheet of what to avoid.
Why do public sector ERPs fail so much more than the private sector?
Public sector ERP failures happen at roughly double the rate of private sector ones.
Gartner data shows 50% of public sector implementations fail versus 25% overall.
And when journalist Chris Kanaracus reviewed the top ERP failures of the last decade, over half came from the public sector despite the public sector representing only 20% of the market.
Three structural reasons this keeps happening:
1. Procurement rewards the lowest bid.
Government RFPs weight cost heavily, so integrators lowball proposals to win, then discover the scope was never realistic.
Corners get cut, quality slides, and total cost of ownership goes up.
2. Fixed-price contracts kill change management.
These contracts sound safe on paper. In practice, they push integrators to rush past the exact activities that make ERP work: process redesign, user training, phased testing.
3. Commercial ERP doesn’t fit government processes.
Fund accounting, procurement rules, union pay contracts, multi-agency reporting, none of it maps cleanly onto software built for a manufacturing company. So governments customize heavily, and every customization becomes a permanent maintenance liability.
Every failure below is some combination of these three problems, plus one or two project-specific catastrophes.
1. U.S. Air Force ECSS
The Expeditionary Combat Support System (ECSS) is the largest single public sector ERP failure in U.S. history.
The Air Force started the project in 2005 to replace 240 legacy systems with one Oracle-based platform covering logistics, supply chain, and financial reporting for 250,000 users.
Seven years and $1.03 billion later, the Air Force canceled it.
The system had produced no significant military capability.
An internal review estimated finishing even a quarter of the original scope would cost another $1.1 billion and push delivery to 2020.
Senators Carl Levin and John McCain called ECSS “one of the most egregious examples of mismanagement in recent memory” in a formal letter to Secretary of Defense Leon Panetta. Of the $1 billion spent, the Air Force said it received usable hardware and software worth less than $150 million.
What went wrong?
The project was restructured three times in its final three years.
The scope was too big to manage. Over 250 supply chain systems and 250,000 users in a single go-live. Oracle and prime contractor CSC ended up in a public disagreement about what capability the Air Force actually received. Underneath all of it was a total absence of business process reengineering.
The Air Force tried to force existing processes onto commercial software without redesigning either.
2. California 21st Century Project
In 2005, California’s State Controller’s Office contracted SAP to modernize the state payroll system covering 240,000 employees across 160 departments and 21 bargaining units.
The entire Budget was around $373 million.
Eight years later, the state had spent $254 million and never gotten it to work.
Controller John Chiang piloted the system with just 1,300 employees in his own office (supposedly the easiest test case) and found errors in one out of every three tasks the system performed.
Not a single pay cycle ran without material payroll errors.
Chiang terminated the SAP contract in February 2013 and sued the vendor.
SAP maintained the software worked as designed and it was California’s implementation choices that broke the project.
The state sought to recover up to $135 million from SAP; the litigation took years to settle.
A California Senate report later found the state controller’s office had failed to properly monitor the project.
Independent audits identified the same governance issues, over and over, across multiple oversight cycles, and no one acted on them.
3. Cook County, IL Oracle Fusion
Cook County, Illinois (home to Chicago and the second-most populous county in the U.S.) launched an Oracle Fusion ERP in 2016 to automate HR, finance, and procurement across county government.
The target of completion was 2018.
The county spent more than $75 million before pulling the plug on this attempt.
A separate $36.5 million contract with Tyler Technologies to modernize the county’s court system ran 18 months late and was branded by county leaders as “one of the worst IT projects” in Cook County history.
The root causes had almost nothing to do with Oracle’s software. The county went in with no formal business case, no Phase Zero blueprint, no target operating model. Then it over-customized to preserve legacy processes rather than adopt Oracle’s out-of-the-box workflows. Leadership turned over during the project, so decisions kept getting revisited. And change management was largely ignored.
Cook County eventually completed a different, phased Oracle E-Business Suite implementation with IBM as integrator, but only after treating the first attempt as sunk cost.
4. LAUSD Payroll
In January 2007, the Los Angeles Unified School District (the second-largest school district in the U.S.) went live with a $95 million SAP payroll system implemented by Deloitte.
The system covered roughly 120,000 monthly paychecks.
The rollout was a disaster from day one.
In the first year, the system generated $53 million in overpayments and $7 million in underpayments.
Some 40,000 teachers experienced pay errors. Some got no paycheck at all. Teacher Jordana Alzate, who had just had a premature baby, went nine months before receiving a correct check.
LAUSD ended up spending an additional $37 million in repairs and delays just to get the system stable, which didn’t happen until January 2008.
Deloitte eventually paid LAUSD $8.25 million in settlement and forgave $10 million in unpaid invoices.
Teachers’ union president A.J. Duffy called the settlement “outrageous,” arguing Deloitte should have paid $20 to $30 million minimum.
LAUSD’s other major implementation, the My Integrated Student Information System (MiSiS), turned into its own multi-year disaster.
Deleted student records, mixed-up class schedules, no report cards on time.
The school board approved an additional $33 million in bond funds just for MiSiS repairs, then tentatively approved another $12.1 million on top of that.
In 2026, a former LAUSD IT project manager was charged with felony money laundering for allegedly steering $22 million in MiSiS contracts to a single vendor in exchange for $3 million in kickbacks.
5. Marin County MERIT
Marin County, California, started implementing SAP ERP in 2005 with Deloitte as systems integrator.
The MERIT (Marin Enterprise Resource Integrated Technology) system went live in July 2006.
The county said it never worked properly.
By 2009, a Marin County grand jury put the total project cost at $28.6 million, for a system that couldn’t produce basic financial reports.
Pension officials had to rely on an outdated 2007 actuarial report because the system couldn’t extract current data.
The county sued Deloitte in 2010 for $30 million, then escalated by filing a federal RICO complaint alleging Deloitte and SAP “engaged in a pattern of racketeering activity designed to defraud the County of more than $20 million.”
Deloitte had allegedly staffed the project “with dozens of neophyte consultants, many of whom lacked even a basic understanding of SAP.” The former county project manager left MERIT to join SAP as a sales executive, a move the county cited as evidence of a bait-and-switch scheme.
The 2013 settlement netted Marin County just $3.9 million, against $5 million in legal fees.
The county then started completely from scratch, signing an $8.2 million contract with Tyler Technologies in 2015 for Munis ERP.
6. U.S. Navy ERP Pilots
Starting in 1998, the U.S. Navy launched four separate ERP pilot projects meant to modernize supply chain, acquisitions, and financial management.
By 2005, the Navy had spent about $1 billion on the pilots.
The problem was, the four pilots weren’t interoperable, even though they overlapped in function.
Different design choices, different implementation approaches, no unified architecture.
The Government Accountability Office reported that the $1 billion was largely wasted.
The Navy eventually scrapped three of the four pilots and worked with SAP to deploy a single consolidated ERP at an estimated cost of another $800 million.
According to more recent GAO reporting, the Navy is now spending $1.4 billion between 2022 and 2024 just to maintain its ERP.
7. U.S. DoD financial systems
The Department of Defense has failed seven consecutive financial audits, most recently in 2024, despite spending over $903 billion in 2024 alone.
The GAO has kept DoD business systems modernization on its High-Risk List since 1995 (three decades and counting).
DoD financial systems are a case study in what happens when ERP fragmentation compounds over decades.
The Army, Navy, and Air Force each built their own ERP systems without achieving interoperability.
Now those siloed systems can’t produce audit-quality data.
The Pentagon can’t accurately account for 61% of its $3.5 trillion in assets, according to the House Oversight Committee.
The Pentagon’s own goal for a clean audit has slipped from 2028 to 2030.
The GAO says there is little evidence it can meet even that revised deadline.
And DoD confirmed $10.8 billion in fraud between 2017 and 2024 (a number GAO says is “only a small fraction” of the actual fraud exposure given how weak the financial controls are).
What do all 8 failures have in common?
If you go back through every one of these projects, the same five patterns show up:
1. Wrong implementation partner
Marin County didn’t interview the specific Deloitte consultants staffed on MERIT.
LAUSD assumed Deloitte’s SAP practice had specialized payroll experience.
Cook County trusted its integrator to fill in for missing internal capability.
In every case, the actual people doing the work weren’t the people who sold the work.
2. No business case or target operating model
Cook County’s Oracle Fusion failure had no Phase Zero blueprint.
ECSS never mapped what “success” would look like beyond generic goals.
Without a business case, there’s no test for whether a decision during the project is right or wrong, everything becomes negotiable.
3. Over-customization to preserve legacy processes
Every one of these projects tried to make the software match the way the agency already worked, instead of adopting the software’s standard workflows.
That’s how you turn a 3-year implementation into a 7-year one and a maintenance nightmare afterward.
4. Fixed-price contracts that punish good decisions
Fixed-cost contracts are the root cause of many public sector ERP failures.
They incentivize integrators to cut corners on the exact activities like change management, business process management, and testing that determine whether the system works.
5. No change management or user training
LAUSD teachers were expected to use a brand-new payroll system with almost no training.
LAUSD MiSiS launched all-at-once with no phased pilot. Cook County ignored the people side of the transformation.
And when users can’t use the system, the system fails, regardless of what the technical audit says.
None of these are software problems.
They’re procurement, governance, and management problems that manifest as software problems.
Final Words
If you look at the pattern behind these failures, it’s the same mistakes, same vendors, same integrators, same eight-figure and nine-figure write-offs.
The good news is the failure patterns are well-understood at this point, which means they’re also avoidable.
Independent selection advice, a real business case before RFP, phased rollout with real user testing, and change management funded as a first-class workstream none of that is exotic.
It’s just not what the procurement process rewards by default.
If you’re planning or currently inside a public sector ERP implementation and want a vendor-neutral second opinion on vendor fit, contract structure, or a recovery path if things are already going sideways, book a call with KreativeCoreTech. We do this work for cities, counties, school districts, and public agencies every day.