ERP

When to Replace vs. Fix a Troubled ERP Implementation (2026 Guide for Governments)

Should your government replace or fix an ERP implementation in trouble? Use this evidence checklist, decision matrix and 10-year cost test to decide.

Chris Rodney 17 min read
On this page
  1. At a glance
  2. Replace or fix an ERP implementation: the three real options
  3. What evidence to gather before you decide to fix or replace an ERP
  4. A decision matrix for fixing vs. replacing an ERP
  5. How to compare the cost to complete vs. the cost to replace
  6. How governments decided: fix, new partner, or replace (2022 to 2026)
  7. What each path means for procurement and council approvals
  8. How to present the recommendation to council
  9. The risks of each path, side by side
  10. Frequently asked questions
  11. Making the call

Whether to replace or fix an ERP implementation comes down to where the failure lives. Fix it in place when the software fits your requirements and the problems sit in governance, data, or delivery. Bring in a new implementation partner on the same software when the platform fits but the partner does not. Replace the software only when the fit gaps are structural or the vendor’s future is in doubt.

Leave the money already spent out of the comparison. Weigh the cost and risk of each path from today forward over the same 7 to 10 year horizon, with staff time and dual running included. The rest of this guide covers what to gather, how to score it, how to cost it honestly, and how to take a recommendation to council.

At a glance

You have three paths: fix in place, re-implement the same software with a new partner, or replace the software. In our experience, most troubled projects end up choosing between the first two.

Gather evidence on fit gaps, defect trends, data quality, vendor viability and roadmap, staff capacity, and your contract position before anyone picks a side.

A few findings settle the question on their own, such as a vendor ending support for your product or core requirements the software cannot meet without heavy custom code.

Compare cost to complete against cost to replace over the same long horizon, with sunk costs excluded.

A new partner or new software usually means a new procurement or a cooperative contract, plus a council or board vote, so build that time into the plan.

Oregon, Washoe County School District, San Francisco Unified and Cuyahoga County each took a different path since 2022, and their public documents show how they justified it.

Replace or fix an ERP implementation: the three real options

A troubled implementation has three realistic exits, and each one fixes a different kind of failure. Picking the wrong exit means paying twice for the same problem.

Swipe to see all columns →

PathWhat changesWhat staysBest fit when
Fix in placeGovernance, scope, project team, data approach, testing disciplineSoftware and (usually) implementation partnerSoftware meets core requirements; defects trend down once managed; partner is capable but poorly directed
Re-implement with a new partnerImplementation partner, often the configuration and project planSoftware and subscriptionPlatform fits, but the partner lacks capacity, product depth, or public-sector experience
Replace the softwareSoftware, partner, contracts, often the whole designYour requirements, data, and lessons learnedStructural fit gaps, vendor end of support or weak roadmap, or a 10-year cost that favors switching

A fix in place still changes a lot. It usually involves a re-baseline: new scope, new dates, new commitments from the vendor, and stronger governance on your side. If you need a refresher on why projects reach this point, our breakdown of why government ERP projects fail covers the common root causes.

What evidence to gather before you decide to fix or replace an ERP

Gather six kinds of evidence before the decision: fit gaps, defect trends, data quality, vendor viability and roadmap, staff capacity, and contract position. Pull it from documents and system reports wherever you can. Interviews help, but everyone on a troubled project has a position to defend.

Swipe to see all columns →

Evidence areaWhat to collectWho owns itWhat it tells you
Fit gapsRequirements traceability matrix marked met, met by configuration, met by workaround, met by custom code, or not metProject manager with department leadsWhether the software can do the job, or whether the gaps are in configuration
Defect trendsOpen defects by severity and module, opened vs. closed per week over recent weeks, repeat defectsTest leadWhether problems are converging or multiplying
Data qualityMock conversion results, reconciliation of balances, vendor and employee master error ratesFinance and HR data ownersWhether a new system would inherit the same mess
Vendor viability and roadmapProduct support dates, release roadmap, customer base in your segment, financial filings for public vendorsIT director, procurementWhether the platform will still be supported over your horizon
Staff capacityHours your staff have spent per week, turnover on the project, backfill status, upcoming fiscal close and payroll calendarFinance director, HRWhether your team can absorb a restart
Contract positionTermination for convenience and for cause clauses, payments tied to acceptance, deliverables accepted, warranty, data ownership and exit termsCity attorney, procurementWhat leaving costs and how strong your bargaining position is

A few practical notes on each area:

For fit gaps, the “met by workaround” and “met by custom code” columns matter most. A long list of workarounds in payroll, grants, or utility billing is a warning sign; a long list of “not met” items in core accounting is closer to a knockout.

For defect trends, the direction matters more than the count. Opened-versus-closed by week tells you whether the project is converging.

For data quality, be honest about where the problem started. Dirty data in the legacy system travels to any new platform, so data problems alone rarely justify replacing software.

For vendor viability, read primary sources. SAP, for example, announced in 2020 that mainstream maintenance for core SAP Business Suite 7 applications runs to the end of 2027. A product with a published end date inside your planning horizon changes the math.

For contract position, find out which deliverables you have formally accepted. Accepted deliverables weaken a claim that the vendor failed to perform.

A decision matrix for fixing vs. replacing an ERP

A weighted scoring matrix turns the evidence into a recommendation that council can follow. Score each path from 1 (poor) to 5 (strong) on the same criteria, apply weights, and compare totals. The weights below are a KCT rule of thumb for local governments; adjust them to your priorities and record why.

Swipe to see all columns →

CriterionExample weightFix in placeSame software, new partnerReplace software
Software fits core requirements25%Score from fit-gap evidenceSame score as fixScore of best alternative from market scan
Risk to payroll, close, and billing continuity20%Usually scores highest (least disruption)MiddleUsually scores lowest (longest dual running)
10-year cost from today20%From cost modelFrom cost modelFrom cost model
Time to stable operations15%Shortest if defects convergeMediumLongest
Vendor viability and roadmap10%Current vendor’s datesCurrent vendor’s datesAlternative vendor’s dates
Staff capacity to execute10%Lowest new demandMediumHighest new demand

Before you total anything, apply knockout tests. These findings decide the question on their own, whatever the scores say:

1

The vendor has announced end of support for your product inside your planning horizon, with no migration path you would accept.

2

Core accounting, payroll, or billing requirements are “not met” and could only be met with custom code you cannot maintain.

3

The vendor’s product has no meaningful customer base among governments of your type and size.

4

Your contract gives you no practical way to change partners without restarting the license (rare, but check).

5

Your staff cannot support another restart before a fixed deadline such as a fiscal year start or a payroll tax change.

Knockouts 1 to 3 push toward replacement. Knockout 4 pushes away from a partner change. Knockout 5 pushes toward the smallest change that keeps operations running, which is usually a fix in place with a narrower scope.

How to compare the cost to complete vs. the cost to replace

Compare forward-looking costs only, over the same horizon, for every path. Money already spent is gone whichever path you choose, so it belongs in the history section of the council memo, not in the comparison.

The sunk cost trap is real in public budgets because past spending is visible and politically painful. Two equally wrong arguments show up in council chambers: “we’ve spent too much to walk away” and “we’ve spent too much, so cut our losses.” Neither one asks what each path will cost from here.

Use the same 7 to 10 year window that ERP selection advisory work uses for total cost of ownership, and build each path from these lines:

Swipe to see all columns →

Cost lineFix in placeSame software, new partnerReplace software
Remaining or new implementation servicesRemaining contract plus change ordersNew partner contract, less credit for reusable workFull new implementation
Software subscription or licenseExistingExisting (usually unchanged)New, plus any early exit fees on the old one
Internal staff time and backfillRemaining project hoursRework hours plus remaining hoursFull project hours again
Dual running of legacy and new systemsUntil go-liveLonger, by the restart delayLongest
Data conversion and cleanupRemaining mock loadsRe-validation of converted dataNew conversion cycle
Independent oversight and testingRecovery oversightOversight plus re-planningSelection plus implementation oversight
Termination and legal costsNoneSettlement or close-out costsSettlement plus exit costs
Operating costs years 2 to 10Support, upgrades, retained workaroundsSameNew vendor’s support model

Staff time is the line most often left out. Count project hours for finance, HR, payroll, IT, and department subject matter experts, and price them at loaded cost plus any temporary backfill. Dual running belongs on the list too: every extra month on the legacy system means extra maintenance fees, duplicate entry, and reconciliation work.

Washoe County School District in Nevada published the numbers that make this comparison possible. In a December 9, 2025 board presentation, staff reported $1,869,558 already paid on an $11,209,491 implementation contract, which left $9,339,933 unspent. The replacement implementer’s contract was $9,954,960, with a credit still to be set for work that would not need to be redone.

Read forward, the choice was roughly $9.3 million to continue with a partner the district had lost confidence in versus about $10 million, less that credit, with a new one. The presentation also showed a total budget view that included the amount already paid, which is the right way to report history to a board. Keep the decision itself on the forward-looking numbers.

If you want a neutral read on these numbers before the decision hardens, KCT’s ERP Recovery & Stabilization team delivers an assessment within two weeks with a measurable baseline of where the project stands. That baseline is often enough to show which path the evidence supports.

How governments decided: fix, new partner, or replace (2022 to 2026)

Public agencies have taken all three paths in the last few years, and their decision documents show the reasoning. These examples are useful because they are public records you can cite to your own council.

Fix in place: State of Oregon, Workday payroll

Oregon launched a Workday payroll and HR system in December 2022 for about 44,000 state employees. According to Oregon Capital Chronicle reporting republished by Salem Reporter in June 2024, about 4,500 employees were underpaid or overpaid in January 2023, and auditors could not complete their review because of missing system documentation.

The state kept working through fixes inside the system. The Department of Administrative Services told the outlet that “significant progress has been made” and planned to meet all audit recommendations by year end. The lesson for a fix-in-place decision: expect the audit to examine your controls and documentation, not just the defect count.

Same software, new partner: Washoe County School District

Washoe kept Oracle Cloud ERP and replaced its implementer. Its board presentation lists the reasons: the prior implementer changed its project team at the start, district concerns about performance and staffing went unresolved, and deficiencies continued even after the district suspended work under the contract.

Staff called it a “quick hook” that limited financial exposure. They also accepted a longer schedule, moving go-live from April 1, 2026 to July 1, 2027, and told the board a longer timeline raised the odds of success. The Oracle subscription agreements brought back to the board were identical in terms and cost to those approved a year earlier.

Replace the software: San Francisco Unified School District

SFUSD went live on EMPowerSF in 2022 and decided to replace it two years later. Along the way, GovTech reported in September 2022 that the contract had grown from $9.5 million to $13.7 million and that the board was set to vote on more than $2.8 million for a consulting firm to clear the payroll backlog and stabilize the system.

In its February 29, 2024 announcement, the district chose a system already widely used by California K-12 agencies, planned a parallel implementation, committed to using only funds set aside in 2022 for ERP work, and brought the contract to the board on March 12, 2024. Labor partners publicly supported the change; one union leader called dropping EMPowerSF a step toward rebuilding trust.

Stay on the platform with stronger governance: Cuyahoga County, Ohio

Cuyahoga County’s ERP rollout began in 2016 with cost overruns and delays. In December 2025, GovTech reported that council approved about $34 million to upgrade on the same Infor platform, with the current version reaching end of life in 2030, and that the county added an executive steering committee to keep delivery on time and on budget.

Birmingham City Council in the UK shows how long a same-software restart can take. After its failed 2022 Oracle go-live, the council chose an “out of the box” reimplementation, which went live in August 2026. The Register put the total project cost at around £144.4 million against an initial £20 million estimate, and the council’s auditor still expects qualified audit years ahead while historic data is cleaned up. A same-software restart still needs realistic readiness gates.

What each path means for procurement and council approvals

Every path except a narrow fix in place usually triggers a procurement action and a governing body vote. Plan the procurement timeline into the decision, because it can add months before any new work starts. Rules vary by state and by your own procurement code, so confirm the specifics with your attorney and purchasing office.

Swipe to see all columns →

PathTypical procurement actionTypical approval
Fix in placeContract amendment or change order; possibly a separate contract for independent recovery oversightCouncil or board if the amendment exceeds the manager’s signing authority
Same software, new partnerTermination or mutual close-out with the current partner; new solicitation or cooperative contract for services; subscription often unchangedCouncil or board for the termination settlement and the new services contract
Replace softwareClose-out of current contracts; new software and services procurement (RFP or cooperative contract)Council or board for each contract, plus a budget amendment if funding changes

Some practical points:

Cooperative contracts can save time. Programs such as NASPO ValuePoint pool purchasing across states. Before relying on one, check with other agencies, your state, and associations for existing cooperative contracts, and confirm your state and local procurement rules allow you to use them.

Check the funding source. Washoe’s ERP costs were capital funded, and SFUSD limited itself to funds already set aside. Confirm whether bond proceeds, grants, or capital funds carry restrictions on a re-scoped project.

Clean up encumbrances. Open purchase orders to the outgoing partner should be reviewed and liquidated or amended so they do not distort budget reports at year end.

Assume public records laws apply. Steering committee minutes, defect logs, and vendor correspondence may be disclosable, so write them as if council and the press will read them.

Time the change around your calendar. Avoid cutting over or changing partners near fiscal year end, ACFR preparation, W-2 season, or a bargaining unit contract change that touches payroll.

How to present the recommendation to council

Frame the recommendation around the evidence and the options you weighed, and keep past choices in a short history section. Councils approve recovery plans when they can see the alternatives were weighed fairly and the numbers are forward-looking.

A staff report that works usually follows this order:

1

State the recommendation and the action requested in the first paragraph.

2

Summarize the current state with facts: what is live, what is not, open defects by severity, and audit or payroll impacts.

3

Show the three options considered and the scoring matrix, including any knockout findings.

4

Present the 7 to 10 year cost comparison from today, with sunk costs shown separately as history.

5

List the risks of the recommended path and how each will be controlled.

6

Describe governance going forward: steering committee, decision rights, reporting cadence to council, and readiness gates before any go-live.

7

Name the procurement steps and dates for the next council actions.

Sample wording that keeps the focus forward: “Staff evaluated three options against the same criteria and the same ten-year horizon. Amounts spent to date are reported for transparency but are excluded from the comparison, because they are not recoverable under any option.”

Ask for a regular progress report to council as part of the motion, whichever path you choose. Cuyahoga’s added steering committee and Washoe’s public cost comparison are both examples of oversight structures elected officials can see.

The risks of each path, side by side

Each path carries a different risk profile, and the right choice is the one whose risks your agency can actually manage. Use this table to frame the risk section of the staff report.

Swipe to see all columns →

PathMain risksHow to reduce them
Fix in placeThe same partner repeats the same mistakes; staff fatigue; audit findings on controls and documentationRe-baseline with acceptance-based payments, independent oversight, defect exit criteria, documented controls
Same software, new partnerRework of prior configuration; knowledge lost in transition; dispute with outgoing partnerNegotiate a transition period, secure all configuration and documentation, require a discovery phase with a credit for reusable work
Replace softwareLongest dual running; staff burnout; repeating the same requirements and data mistakes on a new platformFix requirements and data first, use a fresh market scan, plan parallel payroll and close cycles, phase go-lives

One risk applies to every path: carrying the same root cause forward. If the original problem was unclear requirements, weak scope control, or dirty data, it will follow you. Our guide to keeping a public-sector ERP project on scope covers the controls worth putting in place on day one of the new plan.

In our experience, the evidence more often supports a fix in place or a partner change than a software replacement. In the City of Oxnard, California, a prior consulting firm had not gotten a multi-module ERP program started; KCT took over and led the implementation with split go-lives, one module at a time.

Frequently asked questions

How do you know if an ERP implementation should be replaced?

An ERP implementation should be replaced when the evidence shows structural problems: core requirements the software cannot meet without heavy custom code, a vendor ending support inside your planning horizon, or a 7 to 10 year forward cost that favors switching. Delivery, data, or governance problems alone usually call for a fix or a new partner.

Can you change ERP implementation partners mid-project?

Changing implementation partners mid-project is possible and often keeps the software subscription intact. Washoe County School District terminated its implementer and approved a new one in December 2025 while keeping its Oracle subscriptions. Expect a discovery phase, some rework, a longer timeline, and a council or board vote on the new contract.

What is the sunk cost fallacy in an ERP project?

The sunk cost fallacy in an ERP project is letting money already spent decide whether to continue. Spent funds are unrecoverable on every path, so they should not appear in the comparison. Compare only what each option will cost and deliver from today forward, over the same multi-year horizon.

How long does it take to re-implement an ERP system?

Re-implementing an ERP system usually takes as long as a new implementation, which for governments typically runs 12 to 24 months. A partner change on the same software can be shorter if configuration and data work are reusable, but procurement and discovery add time.

Who should make the decision to fix or replace an ERP?

The decision to fix or replace an ERP belongs to the governing body, based on a staff recommendation. The finance director, IT director, and city or county manager should own the evidence and the recommendation, with the attorney and procurement office reviewing contract options. An independent assessment helps when internal teams are too close to the project.

Making the call

In our experience, a troubled ERP project rarely needs the most dramatic option. The software is often capable, and the real work is governance, data cleanup, partner performance, and a plan council can trust. Save replacement for the cases where the fit-gap, vendor, and cost evidence point there.

Gather the evidence and run the knockout tests before anyone takes a position in public. Then build the forward-looking cost comparison, because that is the page council will read twice.

If you want an independent read on your project, KCT’s ERP Recovery & Stabilization service starts with a two-week assessment, takes over vendor governance where needed, and re-baselines scope, timeline, and commitments. Book a conversation with the KCT team to talk through where your implementation stands.

Ready when you are

Let's talk before it slips.