When to Replace Your Old System: Five Signals

You already have a system, and using it is exhausting. That is a completely different problem from "we still run on Excel, should we get a system?" Replacing an old system is hard because it still works, all your data is in it, and the cost of replacing it is visible while the cost of keeping it hides in payroll. This post breaks down five signals that it really is time: someone's job is moving data between systems, reports have to be exported and recalculated, the vendor has stopped updating, every small change needs a new quotation, and new business models don't fit. And one more thing that matters more: hitting all five doesn't mean you should replace it now.

Johnny Yang12 min read
When to Replace Your Old System: Five Signals

"We've been using this system for almost ten years. Yes, it's genuinely painful to use. But it still runs every day, and just thinking about replacing it makes me break out in a cold sweat." I heard this in a conference room at a metal processing company. The owner then added: "And the last implementation already put me through hell."

I'm Johnny Yang of Brightstream Technology in Taiwan. I've led implementation teams for seven years, across 252+ projects in 32+ industries. This post is not about "we still run on Excel, should we get a system?" That's zero-to-one, and the pain is obvious. Companies that already have a system struggle with something else. The old system still works. Ten years of data is inside it. The cost of replacing it is a bill you can see. The cost of not replacing it is scattered across payroll and overtime hours, and nobody has ever added it up for you. So most owners stall. They stall until the system actually goes down and they're forced to deal with it, and that is usually the most expensive way to deal with it.

To decide whether to replace an old system, don't ask "is it easy to use?" Ask "what do we pay every month just to keep it usable?" There are five signals I actually count: someone at the company spends their job moving data between systems; important reports have to be exported and recalculated in Excel; the original vendor or reseller has stopped updating it and you can't even find someone to answer a question; every small change needs a new quotation, and the next upgrade overwrites it anyway; new business models can't be squeezed into the existing architecture. What these five have in common is that the losses keep happening, and they grow as the company grows. But more important: hitting all five doesn't necessarily mean you should replace it now. The real time to replace is when "the signals hold" and "you can absorb the chaos of the switchover" are both true. If you have the first without the second, you end up in the worst state: the old system is shut off and the new one isn't up.

Where the owner's instinct is right

Plenty of consultants walk in and say "your system is too old, you need to replace it." Half of that sentence is about making a sale. Your instinct, "if it still runs, keep running it," is right in a lot of cases.

Old is not a crime. I've seen systems in use for over a decade with hideous interfaces and report formats stuck a generation back, but the company's operating rhythm has worn into them well and the books reconcile. Replace a system like that and you get months of pain plus the whole company relearning how to use it, and what you get back may just be a nicer-looking interface. That math doesn't work, and I'll tell you straight not to replace it.

The real problem isn't age. It's that the system's architecture no longer matches how you do business now. Mismatch means the company has grown a whole set of extra manual processes just to accommodate the system, and those processes are eating headcount and response speed. That's what decides whether to replace it, not the system's age.

Five signals it's time to replace

I check the five below one by one. The point is that each one has to map to a specific person, file, or event. If it doesn't, it's a feeling, and a feeling is not a reason to replace a system.

Signal one: someone's job is effectively "the courier between systems"

Is there someone at your company whose fixed task every morning is to export a file from System A, fix the formatting, and paste it into System B or another spreadsheet? Maybe it's pasting online store orders into the shipping system. Maybe it's handing production reports to accounting for booking. Maybe it's typing stock count results into the main system one line at a time.

This person usually doesn't show up on the org chart as that role. Their title is sales support staff, accounting assistant, or production control, but a big chunk of their time goes to moving data. Worse, over time it becomes "only they know how." They know which fields don't match and need manual fixes, which customer has different IDs on each side, which entry at month-end needs special handling. All of that knowledge lives in their head. When they take leave, the whole company starts shipping late. You're not just paying a salary. You're carrying a single point of failure with no backup.

But this signal has a cheap fix: don't replace the system, just connect the two. If both sides still have vendors maintaining them and can offer an integration method, integrating first is far better value. So the first step isn't shopping for a new system. It's asking your current vendors whether they can connect. Only if they can't do you move to the next question.

Signal two: important reports have to be exported and recalculated in Excel

The system has a reporting function, but nobody actually uses it to make decisions. The report that gets taken into meetings is one someone exported, pasted into an Excel file that's been maintained for years, full of formulas, a few hidden sheets, and manual adjustments only the author understands.

This means there's a gap between what the system records and what you need to manage. The gap may be in dimensions: you want gross margin by salesperson, by customer, by product line, and the system can only give you shipment value. Or costs never actually made it into the system, so every number tied to profit has to be calculated outside it.

Whether to take it seriously, I look at two things. First, how important is that Excel file? If it's just a trend chart the owner looks at, the impact is limited. If quotations, reconciliation, and bonus calculations all depend on it, then your core management runs on a file with no version control that nobody will maintain once its author leaves. Second, can adjusting the reports fix it? Sometimes the report just wasn't set up right, and the vendor can tweak it. But if the gap comes from data that was never recorded, no amount of tweaking will produce the number. That's an architecture-level problem.

If you have no system at all and the whole company runs on Excel, that's a different question. When to switch from Excel to ERP covers how to judge zero-to-one. It's a different situation from this post. Don't mix the two up.

Signal three: the vendor has stopped updating, or you can't find anyone

This signal is the easiest to ignore because it doesn't hurt day to day. The system runs and nobody bothers you. Until one day something breaks, you call, and find the engineer who built it left long ago and nobody left has ever touched your version.

What to check? When did you last get a feature update? Were the problems you reported actually fixed? What does the maintenance contract cover, and is it still being renewed? Some situations are especially dangerous: the system runs on an old server nobody dares touch, the original customization was written by an engineer who has left and there's no documentation, or the database version is so old the operating system no longer supports it. At that point you're not using a system. You're betting against time.

Among the failed implementations I've seen, one kind is especially painful: the company waited until the old database crashed, then scrambled to find someone to take over. The data came out in pieces. At that point nobody was talking about implementation quality. It was a rescue. So even if you're in no hurry to replace, first confirm you can export your complete data in a standard format, and actually export it once and look at it. Old server problems usually lead into the question of deployment model; see cloud vs. on-premise ERP.

Signal four: every small change needs a new quotation, and gets overwritten afterward

Your process changed, and you want to adjust one small thing in the system: one more field, one more filter on a report, one more approval step on a form. The vendor says they need to assess it, quote it, schedule it. Back and forth takes two or three months, and in the meantime your people hold it together by hand.

At first you'll think it's the vendor's attitude. Over time you find it's mostly structural. The original customization was hard-coded into the system's core, so changing one thing means retesting everything. Upgrades are worse: when the vendor releases a new version, all that customization has to be redone and retested. So you don't dare upgrade, you get stuck on an older and older version, and you're back at signal three.

The key is what kind of thing you want to change. If it's in a part of your process that keeps changing anyway, and every time it has to go through a custom-built quotation, the system no longer fits the flexibility you need. But if you're a company that demands the system be changed to match existing habits for every small request, the problem isn't the system. It's how you manage requirements, and a new system will just replay the same script. The ERP customization post covers how to tell what should be custom-built and what should change on your side. It's worth getting through that before you act on the urge.

Signal five: new business models don't fit

This is the one of the five to take most seriously, because it blocks the future.

It usually looks like this. You did wholesale, now you've started selling online, but the system's inventory logic has only one warehouse and can't handle sellable stock across channels. You made to order, now you want small-batch custom work, but the part number structure has no room for variants. Or you had one store and now you have three, the system has no concept of branches, so you run three copies and consolidate by hand.

When it doesn't fit, there's always a way to muddle through in the short term, and the way is signal one: put someone next to it patching things in Excel. That's why these five signals often show up together. Signal five is the cause, signal one is the effect. I weigh this one heavily because it affects decision speed. Every time the new business grows a little, the back office needs one more person's worth of manual work. Without noticing, you slow down, and you may even start avoiding certain orders. That loss won't appear on any report, but it's real.

Hitting all five doesn't mean now

Plenty of owners hear this and say "I hit all five," then ask when we can start. I usually ask three questions back.

In the next six months, is there a stretch where you can absorb chaos? A switchover always has a painful period. Shipping slows, reconciliation gets messy, people work overtime. If the next three months are peak season, or you're pushing on a big deal, or preparing for an audit, the answer is not now. Not never. After that. Force the switch during your busiest period and you'll break shipping exactly when on-time shipping matters most.

Is there one person at the company who can truly own it? Not in name. Someone who can set aside real time every week, has the authority to decide how processes change, and whom the people on the floor will listen to. Without this role, implementation turns into the vendor pushing and your team passively going along, and you end up with a system nobody believes in. If you genuinely don't have this person, the people problem is what you solve first.

How much of your current process is agreed on, and how much is everyone doing their own thing? If three salespeople have three ways of doing quotations, and two plants have two production reporting logics, replacing the system forces you to do two things at once: standardize processes and implement the system. You can do both together, but know in advance it will be much harder and take longer.

There's one more case where I'll tell you outright not to replace: you hate the system mainly because the interface is ugly, there are too many steps, and the report formats aren't to your taste. Most of that can be improved in the current system. Have the vendor adjust reports, trim required fields, rearrange frequently used functions. The effort is not in the same league as replacing the whole thing. Do the cheap things first. If it still doesn't work after that, you'll replace it with far more confidence.

Before you replace, finish these three things

If the assessment says yes and you've picked the timing, there are three things I require to be finished before you sign any contract.

First, take inventory of your data. Which data in the old system actually needs to come along, and which can just be archived for lookup? Ten years of transaction detail doesn't all need to move. Plenty of companies pay a lot to migrate history nobody ever looks at again. At the same time, verify: can this data be exported, and is the format usable? You must do this while you still have leverage.

Next, write down every gap you're patching by hand. Every intermediate Excel file, every manual transfer, every report that gets exported and recalculated is a requirement. This list is the exam you give new systems, and the baseline for acceptance later. Without it, all you'll get is the vendor's version of a feature list.

Third, decide how to switch over. Everything in one day, or module by module in phases? Phased is safer. The cost is that you'll necessarily run two systems in parallel during that time. Make it clear in advance which side is the source of truth for which data, and set an end date for running in parallel. Parallel running without an end date becomes permanent.

What you can do this week

No meetings, no calling vendors first. Find two hours this week and do two things.

One: go through the five signals one by one, and for each, write down a specific person's name, file name, or event. Don't write "reports are hard to use." Write "who exports the monthly sales report, which file they paste it into, how long it takes." Two: find the courier at your company and ask them: if you didn't have to do this transfer tomorrow, what would you do with the extra time? Their answer will tell you directly how much this is worth.

When you're done, you'll land on one of two conclusions. If the signals are scattered and can all be fixed by adjusting the current system, don't replace it yet. If the signals come in a chain and cause each other, start building a timeline and pick a period the company can handle. If you go through it and still can't tell which one you are, Brightstream offers a free 30-minute diagnostic. Bring that list and let's talk.

Found this useful? Pass it on
Prices published · free range estimate

What is that system actually costing you a year?

Lay the maintenance contract or the invoices out and count once: what is real cost, what can be cut, whether it is worth touching at all. If the answer is leave it alone, we will tell you to leave it alone.

From NT$20,000/month · scaled to system size · on-list changes at no extra cost · or just book a free 30-minute diagnostic

Johnny Yang (熊董)
Johnny Yang (熊董)
Founder, Brightstream Technology

Serial founder leading a small team that builds systems into an owner's everyday work.

As a founder
7 yrs
Projects delivered
252+
Industries served
32+
About Johnny