A legacy ERP fails one of two tests: the infrastructure under it can no longer be kept current and reachable, or the system no longer fits the work. Only the second test justifies replacing the ERP. When every sign is an infrastructure sign, hosting or upgrading the current system fixes it without a replacement project.
Most conversations about cloud ERP start with two complaints in the same meeting. IT says the server under the system runs out of support next year. Finance says the team re-keyed another batch of orders and rebuilt another report in a spreadsheet before anyone would trust it. Then somebody says the business needs to move to the cloud, and both complaints get filed under one answer.
They are different problems. One is about where the system runs. The other is about whether the system still fits how the business works. Across 28 buyer meetings Crestwood has mined from its own discovery and demo calls, administrative work that grows with volume came up in 22. None of the 14 pain patterns indexed across those meetings is about where the software is hosted.
A disclosure, because the argument depends on it. Crestwood sells and implements cloud ERP, and it also hosts existing ERP systems on Microsoft Azure for businesses that are keeping them. We are paid on either answer. Read the two tests below knowing that.
What makes an ERP legacy: two tests, not one
A legacy ERP is a system that fails at least one of two tests. The infrastructure test asks whether the server, database and access around the system can still be kept current and reachable. The work test asks whether the system still fits the work the business does today.
Age is not the test. A ten-year-old system on a supported server that still fits the work costs the business very little. A system the team works around every day costs it every day, however recent the install.
Most of the vendor pages ranking for this term treat old, on-premises and unfit as one verdict, and most of them end in the same recommendation: replace it with cloud ERP. That recommendation is right for some businesses and an expensive detour for others. The two tests are how to tell which one you are.
Test one: the infrastructure signs
Infrastructure signs are about where the system runs: the server, the database, remote access, upgrades and backups. They are real risks, and every one of them can be fixed without changing how a single person in the business does their job.
| Sign | What it looks like | What fixes it |
|---|---|---|
| Server or database past vendor support | Windows Server 2016 extended support ends in January 2027. SQL Server 2016 extended support ended in July 2026, and Microsoft sells Extended Security Updates for it in yearly terms to July 2029. | Upgrade the server and database, buy time with security updates, or move the system to a hosted server |
| Remote access that breaks | People outside the office connect through a VPN or remote desktop, sessions drop, and a team in another time zone cannot log in reliably | Host the system where remote users reach it directly |
| Slowdowns at the busy hour | The system slows at the same time every day, when the load peaks | New hardware or more hosted capacity |
| Upgrades skipped | Production is several releases behind because every upgrade needs the one person who understands the server | Make upgrades somebody’s job: an internal owner or a managed host |
| Backups nobody has restored | Backups run every night and nobody has tested a restore from one | Managed backup with a tested recovery |
Look at the right-hand column. Every fix leaves the screens, the reports and the daily work exactly as they are. If the business is otherwise content with how the work runs, these are smaller projects than a replacement, and nobody has to learn a new system.
Infrastructure signs are also rarely what moves a business. Across nine migration conversations in Crestwood’s buyer corpus, not one buyer named a published support deadline as their own reason to change systems. Where a deadline came up at all, the partner raised it. A support date is a good reason to act on the server. On its own it says little about the ERP.
Test two: the work signs
Work signs are about whether the system still fits how the business runs. The strongest one is work that grows with volume: when more orders, invoices or entities mean more people keying data, the system has been outgrown, wherever it is hosted.
| Work sign | What it looks like | Meetings, of 28 |
|---|---|---|
| Administrative work that scales with volume | More orders or invoices mean more hours of keying. One buyer: “the time spent in data entry is to the point of insane.” | 22 |
| Systems that do not talk to each other | Data is exported from one system, batched and keyed or imported into another | 16 |
| Nobody can see the operation | Leaders wait for somebody to build a report before they know where things stand | 16 |
| Inventory nobody trusts | What the system says is on hand and what is on the shelf disagree | 16 |
| Knowledge trapped in one person | The vacation test fails: if one person is out mid-process, nobody else can pick it up | 16 |
| A spreadsheet doing a system’s job | Budgets, payroll splits or costing live in Excel and get keyed in again | 15 |
| Reporting that cannot answer the question | Each new report is a custom build or a spreadsheet cleanup before anyone trusts it | 14 |
Two things stand out. Every sign in that table describes a person’s day. Buyers do not say “inefficient process.” They say who keys what, and how long it takes, and the tone is tiredness rather than ambition. And none of the 14 patterns in the index is about hosting. Where the software runs does not appear as a pain in its own right.
The fastest version of the work test is a volume question. If the business took on a third more orders next year, would it need more people to process them? If the answer is yes because of re-keying, reconciling or rebuilding reports, the system is limiting growth. Moving that same system to a hosted server leaves the answer unchanged.
Read the two tests together
The two tests produce four outcomes, and only two of them involve replacing the ERP. Where the system runs and whether it fits are separate decisions, and a business can make one without the other.
- Neither test fails: optimize. Keep the system and get more from it. The question is how much of it the team actually uses.
- Only the infrastructure test fails: host or upgrade. Fix where the system runs. Hosting an existing ERP on Microsoft Azure or a supported server leaves the work as it is.
- Only the work test fails: replace the ERP. Then decide where the new system runs as a separate choice. Cloud is the common answer, and it is a choice, not a consequence.
- Both tests fail: replace and re-host. Cloud ERP fixes both at once, which is why it is the usual answer here. This is the case the vendor pages describe, and it is one case of four.
The expensive mistakes sit in outcomes 2 and 3. A business with only infrastructure signs that buys a replacement pays for a project it did not need. A business with work signs that only moves its server has paid to host the problem.
Score the work test with the Modernization Decision Framework
The Modernization Decision Framework is the five-dimension self-score Crestwood uses to decide whether a system is building capacity or limiting it. Score each dimension 1, 3 or 5, add the five scores, and the total places the business in one of three ranges. The worksheet version, with an executive discussion guide and a partner evaluation checklist, is in the Modernization Decision Toolkit (PDF).
| Dimension | The question | Score it 1 when |
|---|---|---|
| Workflow integrity | Are processes running through the system, or around it? | Spreadsheets and workarounds have become the process |
| Financial visibility | Can the numbers be trusted and defended without manual work? | Month-end takes several people, several files and several days before anyone trusts the numbers |
| Operational intelligence | Does the team have current data where they work? | Field teams call the office for status, and reports are days old |
| System connectivity | Do systems share data automatically? | Orders are exported to spreadsheets and entered again |
| Growth readiness | Will the system carry the next three years? | People have been added to manage what the system cannot do |
- 5 to 12, Modernize. The system is constraining the business. The open question is when and how, not whether.
- 13 to 19, Assess. The problems are real. The question is which ones are process and which ones are system, because that decides the investment.
- 20 to 25, Optimize. The system is sound. The opportunity is using more of what is already paid for.
The framework scores the work test only. A business can score 22 and still have a server out of support next year. That is outcome 2, and the fix is the server.
What the split looked like at two companies that moved
Both companies below had signs on both tests. What they show is that the replacement decision and the deployment decision were made on different evidence.
xByte Technologies: replaced for the work, chose on-premises
xByte Technologies, an IT hardware reseller in Bradenton, Florida, was running Everest Software on an end-of-life Windows Server, and the ERP slowed down every afternoon. That is the infrastructure test failing. Invoicing also consumed hours of accounting time every day. That is the work test failing.
The company replaced the ERP with Acumatica and then made the deployment call on its own merits. Its chief executive, Ryan Brown: “The key thing for us with Acumatica is that not only was it web-based, but it wasn’t purely cloud. We liked that we could control it ourselves, so we bought the on-premises model.” Over five years, sales doubled from $12 million to about $24 million, and three hours a day came off daily invoicing. Read how xByte Technologies replaced an ERP on an end-of-life server.
Firewire Surfboards: the server sign arrived with a work sign
Firewire Surfboards was still on the basic business software it launched with. Its Australian arm crashed repeatedly logging in, because the system was hosted on-premises in the U.S. office. That is an infrastructure sign. Finance also lacked visibility into the European distribution entity and into weekly production finances. That is a work sign.
Moving the server closer to the Australian team might have fixed the logins. It would not have given finance the European numbers, and that gap is a work sign no hosting decision reaches. The company moved to Acumatica, and its case study reports visibility across all five operating entities and growth of more than 30 percent year over year in 2016. Read Firewire Surfboards’ move to one system across four countries.
Why Crestwood asks the work question first
Crestwood sells and implements two ERP platforms, Acumatica and Microsoft Dynamics 365 Business Central. Because the four outcomes above pay us differently, the order of the questions matters more than any answer we would prefer. The work question comes first because it decides whether a replacement is justified, and because it is the one a server upgrade cannot fix. If the system in question is Dynamics GP and the work test has failed, the data and cutover decisions that follow are in how to plan a low-risk Dynamics GP migration.
If the score lands in Assess or Modernize and the team cannot agree which problems are process and which are system, that is the question the Business Optimization Assessment answers. A Crestwood advisor works through five areas with the people who run finance and operations: workflow, reporting, system utilization, connected systems, and finance and controls. Within 24 hours the team receives a written, prioritized summary, at no cost, and it is theirs to act on with Crestwood or without us.
The business may not have a people problem. It may have an operating model problem.
We sell, implement, support, and optimize it. An in-house team of industry experts. One hand to shake.
Frequently asked questions
What is a legacy ERP?
A legacy ERP is a system that fails one of two tests: the infrastructure under it can no longer be kept current and reachable, or the system no longer fits the work the business does. Age alone does not make a system legacy. A supported, well-hosted system that still fits the work is not legacy in any sense that costs money.
What are the signs a business needs a new ERP?
The strongest sign is work that grows with volume: more orders or invoices mean more people keying, reconciling or rebuilding reports. The others are systems that do not talk to each other, inventory nobody trusts, spreadsheets doing the system’s job, reporting that needs a cleanup before anyone trusts it, and processes only one person understands.
Can an existing ERP move to the cloud without being replaced?
Often, yes. Many on-premises ERP systems can run on a hosted server in a cloud such as Microsoft Azure with no change to how the software works. That fixes server, access and backup problems. It does not fix re-keying, reporting or visibility problems, because the software and the way people use it stay the same.
Is on-premises ERP the same as legacy ERP?
No. On-premises describes where a system runs. Legacy describes whether it still works for the business. xByte Technologies replaced an ERP that was running on an end-of-life server, then chose to run its new web-based system on-premises because it wanted to control it.
When does Windows Server 2016 support end?
Microsoft lists the extended support end date for Windows Server 2016 as January 2027. SQL Server 2016 extended support ended in July 2026, with paid Extended Security Updates available in yearly terms to July 2029. An ERP on either needs an infrastructure plan, whatever is decided about the ERP itself.
How do you score whether an ERP is holding the business back?
Use the five dimensions of the Modernization Decision Framework: workflow integrity, financial visibility, operational intelligence, system connectivity and growth readiness. Score each 1, 3 or 5. A total of 5 to 12 means modernize, 13 to 19 means assess which problems are process and which are system, and 20 to 25 means optimize.
Does moving to cloud ERP fix manual work?
Only if the move changes how the work is done. Moving the same ERP to a cloud server leaves every manual step in place. A replacement fixes manual work when the new system is set up around how orders, invoices and reports actually flow, which is a design decision, not a hosting one.
