"Johnny Yang, we spent serious money implementing an ERP last year. And after the implementation? The books still don't reconcile, staff still secretly open Excel, the warehouse still writes everything by hand. Tell me, did I just throw that money in the river?"
I've heard this at least twenty or thirty times over the past few years. The person saying it always has roughly the same expression. Not anger. That confusion of "I did exactly what I was supposed to, so why is it still like this?"
Let me say something that may offend others in my industry: Most ERP implementation failures aren't because the system is bad, and not because the vendor cheated you. It's because the implementation itself was never treated as a business project, from start to finish. The system just gets pushed forward to take the blame.
Who I am, so you know why I get to say this
I'm Johnny Yang. Seven years as a serial founder. My teams have done 252+ implementation projects across 32+ industries: manufacturing, trading, food, hardware, e-commerce, services. Some went beautifully, live and running smoothly within three months. Some were ugly: everyone was thrilled before signing, and three months after go-live the whole team wanted to quit.
Because I've been through both, I can tell you this: the difference between success and failure is almost always decided before anyone touches the system. In this post I lay out the 5 most common traps. Check them against your own company and a lot of problems will suddenly become very clear.
1. Rushing into a system before mapping your processes
This is the number one cause of failure. Nothing else comes close.
Many owners buy ERP thinking "my processes are a mess, so I'll buy a system to straighten them out." That's exactly backwards. ERP won't make messy processes tidy. It locks your existing mess in place, in a way that's more expensive and harder to change.
Before, three people each handled an order by gut feel. After go-live, three people each handle it by gut feel inside the system, and block each other.
The right order: first map the whole flow of one order, from arrival to shipping to payment collected. Who owns each step, where it gets stuck, why it gets stuck. Figure that out first. You don't need a system for this. A whiteboard and sticky notes are enough to start. If you're still fighting it out with Excel, read my How long can Excel management really hold up first, and confirm whether you actually need ERP or just haven't sorted out your processes.
Mapping processes is the owner's homework, not the vendor's. If you don't get involved at this stage yourself, no amount of money later will make up for it.
2. Wrong specs: the problem isn't paying too much, it's buying something nobody can use
The biggest mistake in choosing an ERP is treating "lots of features" as "a good fit."
I saw a trading company of twenty-some people get talked into a system designed for listed factories with hundreds of employees. The result? 80% of the features were never switched on. Staff had to click through five levels of menus just to find the "issue invoice" button. Eventually everyone gave up and went back to Excel.
More features is not an advantage. Features that match your processes are. Oversized specs don't just waste money. Worse, the system gets so complicated nobody wants to use it. That's the real sunk cost.
There's the opposite trap too: going cheap, buying something that's really just an inventory-and-sales tool, and expecting ERP-level integration from it. I wrote separately about the difference in Inventory systems vs. ERP: what's actually different. Get this straight before you choose, or you'll pay inventory-software prices, expect ERP results, and end up with neither.
As for cloud vs. on-premise and how to set a budget, read How ERP implementation costs are calculated together with Cloud vs. on-premise ERP: how to choose. Those two will help you think through the numbers and the architecture.
3. The data never really made it in
Going live ≠ clean data. This is the quietest trap, and the most fatal.
All of ERP's value rests on one assumption: the data is accurate. But in a real implementation, the most time-consuming job, and the one nobody wants to touch, is data cleanup. Duplicate old part numbers, customer names spelled three different ways, inventory on the books way off from what's actually on the shelves.
Many companies rush to go live, dump the data in carelessly, and switch on. Then every report the system produces is garbage. The owner sees numbers that are wrong, trust collapses instantly, someone says "this system isn't accurate," and the whole company quietly slides back to doing things by hand.
The system isn't inaccurate. What you fed it was inaccurate to begin with. Cut corners on data cleanup and you've paid a lot for a machine that adds and subtracts wrong numbers. Go slower here, be more tedious, do a more thorough stock count. It's absolutely worth it.
4. Using ERP to record, not to decide
This is the biggest wall between "it works" and "it works well."
After implementation, ERP at many companies becomes an expensive typewriter. Every day people enter a pile of documents, and then... nothing. Nobody reads the reports, nobody uses it to make decisions, and at month-end restocking is still decided by gut feel.
Then what was the money for? The real value of ERP is that it can tell you "which customer is actually losing you money," "which material keeps piling up in the warehouse," "where this month's gross margin dropped." Those are judgments, not records.
If you're not even using basic customer data, the problem may come before ERP. I've seen too many companies' CRM end up the same way. The root cause in Why CRM so often doesn't work is almost the same as with ERP. As for the question a lot of people ask now, whether AI can help make these judgments, I broke that down in AI ERP vs. traditional ERP. But the bottom line first: if your underlying data isn't sorted out, AI just gives you wrong answers faster.
5. The owner never bought in
Get the first four right and fail here, and you still lose everything.
The most frustrating case I've seen was an owner who treated ERP as "something the IT department or some vendor handles." He didn't participate, didn't log in, didn't read the reports. Guess what the staff thought? "Even the boss doesn't use it. Why should I bother filling it in properly?"
Staff do what the boss does. In ERP implementation, that's an iron rule. Whether the owner opens the system to look at the numbers himself, whether he uses system reports as the basis for decisions in meetings, whether he holds the line when everyone wants to slack off back to Excel: staff are watching all of it.
ERP is not an IT project. It's a business reform project. It changes everyone's work habits, and only the owner can carry that kind of change. If you don't buy in, don't blame the team for not cooperating.
But "the owner has to buy in" doesn't mean "the owner has to run the project." These two get mixed up all the time, and the owner ends up doubling as project manager and doing both badly. Who should be named as the person responsible, and which pieces IT and outside contractors should each take, I broke down separately in Who should own an ERP implementation? The owner, IT, or an outside contractor.
Quick self-diagnosis checklist
Before you sign, check your company against this table. The more boxes you tick, the more danger you're in:
| Check | Warning sign |
|---|---|
| Process mapping | Still can't say who handles an order from start to finish |
| System specs | Sold by a feature comparison chart, never checked against your own processes |
| Current data | Part numbers, customers, inventory books are a mess and not cleaned up yet |
| Purpose | Only want to "record"; can't say which numbers you want to see |
| Owner's role | Planning to hand everything to the vendor and stay out of it |
| Team alignment | Staff haven't been told, and nobody has been assigned responsibility |
If you tick three or more, I'd advise you not to rush into signing. Filling in these gaps is a hundred times more useful than switching vendors. If you want to fill them systematically, start with my Complete ERP guide for SMEs. That one lays out the whole map for you.
There's another kind of failure that looks different but ends the same way. The implementation doesn't go wrong. It never finishes. Do a POC first to try it out, and after the POC nothing happens. Two years later, the evaluation starts over from scratch. These companies usually share common traits, which I've laid out in POC first? Three kinds of companies where nothing happens after the POC.
Also, these five traps will show up again almost unchanged in AI implementation, just with different names. 5 reasons AI implementation fails at SMEs applies the same logic to AI projects.
Before you sign, ask this question out loud
Whatever vendor you're talking to, before you sign, look the salesperson in the eye and ask:
"Three months after go-live, how do I measure whether this system has actually helped me? Tell me clearly, right now, which numbers we'll use to sign off on it."
A vendor who seriously discusses acceptance metrics with you is a partner who wants to get the job done with you. One who just wants you to sign fast and dodges the question? You already know. This one question nails down what a successful implementation means before you spend any money.
FAQ
Q: Our ERP implementation failed. Will switching vendors fix it? Not necessarily. If it failed for the reasons above, processes not mapped, data not cleaned, owner not involved, it doesn't matter who you bring in. First figure out whether the problem is the system or us, then decide whether to switch.
Q: How long should an implementation normally take? Depends on scale. But my point isn't speed. It's this: don't sacrifice data cleanup and staff training to hit a date. Going live one month earlier in exchange for six months of data chaos is a very bad trade.
Q: Do small companies even need ERP? Isn't Excel enough? Depends where you're stuck. If it's just a few people and order volume is low, Excel can still hold up. But once you start seeing "several people editing the same sheet at once" or "inventory and books don't match," that's the signal. The limits of Excel management covers how to tell.
Q: Can the money from a failed implementation be recovered? Often, yes. The system usually isn't as bad as you think. The problem is the processes and the data. Sort those two out again and retrain the team. I've seen plenty of projects that had been written off get rescued and keep running fine.
If you're considering an implementation, or the ERP you already have is stuck and you want someone to look at where it's actually going wrong, book a free 30-minute diagnostic and talk to me directly. Just pick a time that suits you online. No sales pitch, no courses to sell. I'll just help you think through what needs thinking through before you touch the system. Not burning money on a mistake is worth more than anything.




