“I bought the system. It’s been live for six months. But last week I walked over and my sales support staff still had her Excel file open on screen. I asked her. She said, ‘I do enter it in the system. I just keep my own copy too. It’s faster.’”
I’ve heard this too many times. The tone is usually mixed: half angry, half guilty. Angry because the money got spent and nobody uses it. Guilty because, somewhere inside, the owner already suspects that Excel file really is easier to use.
Let me say this first: your guilt is right. Your anger is wrong.
I’m Johnny Yang, Brightstream Technology, Taiwan. I’ve led implementation teams for seven years, across 252+ projects in 32+ industries. Of the companies I’ve walked into over these years, roughly six in ten hit the same scene within six months of go-live: the system is open, Excel is open, and the same record is stored in two places. And every time an owner asks me “how do I get them to just use the system,” my answer annoys him on the spot. They’re not disobedient. What you bought really is slower than Excel at the thing they do every day.
Conclusion first.
When staff secretly use Excel, nine times out of ten it’s not an attitude problem. It’s a speed problem. In repetitive work, people automatically take the shortest path. That’s instinct, not rebellion. Your job is not to issue a ban, run audits, or call people out in meetings. Your job is to sit on the floor for a day and count how many clicks it takes them to walk one order from start to finish. If the system takes seven clicks and Excel takes one cell, usage will never recover — unless you cut those seven clicks to three or fewer. And before you touch anything, one thing matters more: if you yourself don’t read numbers from the system, then no matter how many times you say it, the whole company knows the system isn’t real.
Start on the employee’s side: secretly using Excel is usually the rational choice
This part may be uncomfortable to read, but I have to write it first. If the owner doesn’t accept this, none of the fixes after it will work.
That Excel file isn’t laziness. It’s a tool she built herself
Look at the Excel file being quietly maintained and it’s usually ugly. Messy columns, colors everywhere, a pile of abbreviations only she understands. But look closely and every one of those things has a purpose: yellow fill means the customer hasn’t replied, red text means chase the payment, and the hand-typed note in the far-right column says “last time they said they’re changing the address.”
That’s not a spreadsheet. That’s a working method one person grew over several years. She knows which customer pays late, which part number often goes wrong, which order to do first. All of that knowledge is stuffed into that Excel file.
Then you buy a system and tell her: from today, stop using that. Use this.
From her side, you just took away her tool and handed her one she doesn’t know. And the new one won’t let her color-code, won’t let her add a notes column, and won’t let her fix a typo. She didn’t argue with you. She just went back to using hers. That’s already a pretty graceful way to handle it.
“User resistance” is the laziest diagnosis there is
The sentence I dread most in a project review: “This didn’t take off because employees resist change.”
It’s a convenient sentence, because it pushes all responsibility somewhere nobody can see. Human nature resists change, what can you do? So everyone nods, the meeting ends, and nothing changes.
The truth is: people don’t resist change. They resist things getting more tedious. Give someone a tool that turns a twenty-minute job into three minutes, and they’ll use it without being taught, and pull their coworkers in too. Not one app on your phone got adopted because of a company policy.
So when you see “nobody uses it,” the question isn’t “why are they resisting.” It’s “did I make their work more tedious.” The answer decides whether you spend your next money in the right place, or keep spending it on management overhead.
The six real reasons your team is still on Excel
These are the six I catch most often on site. Check them against your own company. You’ll usually hit three or four.
Reason 1: the system takes seven clicks, Excel takes one cell
This is the most common one, and the most fatal.
Take the most ordinary task: sales support staff gets a customer call asking to change a quantity. In Excel, she clicks the cell, types the new number, saves. Three seconds.
Same task in the system? Log in, find the order module, search by order number, open the order, discover its status is “Confirmed” and can’t be edited, go back a level and click “Request change,” fill in the change reason (required, cannot be blank), submit, wait for a manager to approve, then come back and check it once more.
The customer is still on the phone waiting.
You don’t need to ask her why she uses Excel. Walk through this flow yourself once and the answer is obvious.
What many owners miss is that the system isn’t slow in the performance sense. The path is long. A faster server doesn’t help, because the time all goes into clicking, switching, waiting for pages, and hunting for buttons. You can’t see any of this in the pre-implementation demo, because the vendor walks the smoothest happy path and nobody is waiting on the phone.
Reason 2: the system won’t let you edit, so one mistake means redoing the whole order
The second reason is irreversibility.
Excel has one feature no system will ever match: Ctrl+Z. Make a mistake, undo, try again. No trace, no approval, nobody knows.
A system, to protect data accuracy, usually locks documents. That’s 100% correct accounting logic, but in practice it has a nasty side effect: once staff learn that “a typo is a hassle,” they start getting everything right in Excel first, confirming it’s fine, then entering it into the system once at the end.
See it? In her head, the system has gone from “where I do the work” to “where I hand in the homework after the work is done.” Once that happens, the parallel track is set, and there’s no going back.
In fairness to the system: locking documents isn’t the vendor being difficult. It’s required for financial compliance. The question is whether a reasonable correction path was designed in. Does an internal order that hasn’t shipped and hasn’t been invoiced really need three levels of approval to change one quantity? Usually not. That’s the process being set too tight during implementation, not a limit of the system itself.
Reason 3: the system’s fields aren’t the fields she actually uses
The third reason is invisible but does a lot of damage.
An order in the system has: customer ID, item name, spec, quantity, unit price, delivery date. Looks complete.
But her Excel file has three more columns: “send the PI (proforma invoice) to this one for confirmation first,” “contact changed, call extension 217,” “last batch had color variance, send photos this time.”
Those three columns have no place in the system. If you force them in, they go into “Notes,” but the notes field can’t be searched, can’t be filtered, and gets cut off when you print reports. So her conclusion is simple: the important things don’t fit in the system, so I still need to keep my own copy.
This is why I keep saying field inventory isn’t the IT department’s job. You have to ask the people who actually use it. If you don’t, a shadow system grows after go-live to fill the gap, and that shadow system is Excel. I wrote the full context in Where Excel management hits its limit. Excel is hard to replace precisely because it can grow a new column any time.
Reason 4: the report the manager wants can only be made in Excel
The fourth reason is pushed down from the top, and it’s the most ironic.
The system has standard reports, but the one the manager wants isn’t there. He wants “this month’s order value by salesperson, minus returns, grouped by customer tier, compared to the same period last year.” The reporting module might be able to do it, or it might need custom work, and that quotation isn’t cheap.
So the manager says: just pull it in Excel for me for now.
The moment that sentence is said, half the implementation collapses. Staff instantly understand two things: first, the system’s data isn’t enough; second, what the boss actually looks at is Excel. If it ends up in Excel anyway, why not do it in Excel from the start and just patch the system afterward.
You can’t require staff to use the system while asking them for an Excel file yourself. You get to pick one.
Reason 5: nobody ever told them why they’re switching
The fifth reason is the easiest to fix and the most often skipped.
Here’s how most companies announce an implementation: one day in a meeting, someone announces a new system goes live next month, please attend training. The training covers “what happens when you press this button.” Done.
At no point does anyone tell the front line: why are we switching? What’s in it for you?
If nobody says it, staff fill in the blank themselves. And the answer they fill in is usually one of three: the boss wants to monitor us, the company is preparing layoffs, or someone is getting a kickback. None of those makes anyone want to use the system properly.
The real reason is usually simple and easy to say: “Sales won’t chase you anymore asking where an order is. They can look it up themselves.” “No more three hours reconciling at month-end.” Saying that does more than ten training sessions. But it has to be true, and you have to deliver it. Otherwise it’s the second broken promise.
Reason 6: the parallel period drags on until both sides are doing everything
The sixth is a design flaw in the process, not anyone’s fault.
During implementation, vendors and consultants often recommend “running old and new in parallel for a while to be safe.” The advice itself isn’t wrong. The fear is data going bad with no way back. The problem is the parallel period has no clear end date.
At first it’s one month. A month passes, a few orders still don’t reconcile, so extend another month. In month two, some report still isn’t ready, extend again. By month four, “parallel” isn’t a transition anymore. It’s the new normal. Everyone is used to doing it twice, and nobody remembers which side was supposed to stop on which day.
If you try to shut down Excel now, the resistance is ten times what it was on day one — because Excel holds six months of complete data and the system holds fragments. Would you dare switch it off?
Running both in parallel: the most expensive failure an SME can have
If you ask me what kinds of implementation failure there are, I’ll tell you the most expensive is not “failed go-live,” and not “couldn’t get it working so we shut it off.” Those are ugly, but at least the bleeding stops.
The most expensive is running both tracks: you keep paying for the system, you keep doing Excel, and both stay alive.
Where the parallel track actually costs you
Many owners underestimate the cost of running both because it doesn’t show up on the books. The monthly fee gets paid, nobody’s salary went up, so it seems like nothing is lost. In reality you’re bleeding in at least three places.
First, duplicated labor. The same record entered twice. That’s the most direct waste, and it happens every day. You bought the system to save labor, and now there’s an extra job.
Second, inconsistent data. The numbers on the two sides will not match. That’s a certainty, not a possibility. Worse, when they differ, nobody knows which one is right. Checking means manual comparison, which is more extra work. Eventually people develop unwritten rules: “inventory, check the system; customer data, check Excel; prices, ask Ms. Wang.” Your company now has three sources of truth.
Third, and most expensive, bad decisions. The numbers on your management reports come from the system nobody seriously maintains. You make calls based on them: whether to stock up on materials, whether to go after new customers, whether to hire. The real situation is in that Excel file. You paid for a system so you could see clearly, and what you see is blurry.
One table: a healthy transition vs a chronic parallel track
| What to check | Healthy transition | Chronic parallel track (warning) |
|---|---|---|
| Deadline | Clear end date, written down | “Let’s do it this way for a while,” nobody remembers to stop |
| Data direction | System is primary, Excel is a backup for cross-checking | Excel is primary, system gets filled in afterward |
| Who maintains it | A few named people reconcile both sides | Everyone keeps their own copy |
| When there’s a dispute | System wins; if Excel is wrong, fix Excel | Excel wins, or the argument never ends |
| When a manager wants a report | Export from the system | Ask someone to pull one in Excel |
| How hard to shut one side down | Can stop any time, data is complete | Afraid to stop, system data may be incomplete |
Check this table against your company. If you hit three or more items in the right column, you’re not in a transition. You’re in a chronic parallel track, and it needs to be handled as its own problem, not left to fix itself.
A chronic parallel track doesn’t fix itself. It only gets harder the longer it drags on, because the Excel side gets thicker every day.
How to fix it: no bans, no audits. Go sit on the floor for a day
Once they spot the parallel track, most owners reach for management tools first: announce that Excel is banned from a certain date, add system entry to performance reviews, have managers spot-check weekly.
I understand the impulse, but I’ll be direct: this is the least effective and most damaging approach.
A ban addresses “the behavior you can see,” not “why they do it.” Ban it, and they won’t come back to the system. They’ll just hide the Excel — on their personal cloud drive, in their own LINE notes, or simply on paper. The only thing you achieve is that from now on you can’t see what’s really happening.
Audits also turn the problem into a confrontation. It started as a question of “is this tool any good.” You turn it into “do employees obey.” Once it’s the second one, nobody tells you the truth anymore.
The same logic holds elsewhere: when nobody follows an SOP, it’s usually not a discipline problem either. It’s that the SOP was never written from how the work is actually done. I break that down in Nobody follows the SOP? The problem isn’t your staff.
Step 1: go sit there for a day
Every time I take on a “nobody uses the system” case, the first thing I do is the same: pull up a chair next to the person most often blamed for “not cooperating,” and watch her work. For a full day.
No meetings, no surveys, no asking her to “write up the problems you’ve run into.” Just sit next to her and watch.
What you see will be completely different from what she says in a meeting. In a meeting she’ll say “it’s okay, just a bit slow.” Sitting beside her, you’ll see her repeat the same switching action forty times in one morning. You’ll see her type “none” every time in a required field because she doesn’t know what goes there. You’ll see her print a system report, circle the key items by hand, then type them back into Excel.
These details won’t appear in any requirements interview. You only see them sitting right there.
Step 2: find the top three high-frequency tasks and cut each to three steps or fewer
After a day, you’ll have a list of “what she did today.” Sort it by frequency and take the top three.
It’s usually things like: checking an order’s status, changing a quantity, printing a shipping list. Those three take up most of her day.
The next job is clear: find a way to bring each of those three tasks down to three steps or fewer.
There are many ways, and the point is not to rush to blame the system. Some can be solved with settings — make frequent searches into shortcuts on the home screen, give required fields default values, cut three approval levels to one. Some need permission changes — she never got access she should have had, so she has to find someone every time. Some really do need custom-built work, so assess whether it’s worth it. I wrote down how to judge that in Should you customize your ERP.
But you’ll be surprised: more than half the sticking points are settings problems, not feature problems. Many companies, aiming to “get it all right in one go,” set processes much stricter than needed at implementation, and nobody went back to loosen them after go-live.
Step 3: let her see the changes herself
This step is the easiest to skip, and it decides whether the first two worked.
After the change, walk over yourself and tell her: “Last time I saw you needed five clicks to look up an order. I had it put right on the home screen. Give it a try.”
What that sentence does isn’t about the feature. It says one thing: someone listens to what you say, and things actually change.
Once she believes that, she’ll bring up every next problem on her own. You go from “one person guessing” to “the whole company finding problems for you.” That’s the real turning point for adoption, and it works better than any review system.
Comparison: the two ways to fix it
| The ban approach | The on-site approach | |
|---|---|---|
| First action | Issue a notice, set penalties | Pull up a chair and sit there for a day |
| Assumption | Staff won’t cooperate | The tool is hard to use |
| How it’s measured | Spot-check who’s using Excel | Step count for high-frequency tasks |
| Staff reaction | Hide the Excel | Start reporting problems on their own |
| Three months later | Compliant on the surface, parallel track underneath | Usage climbs on its own |
| Information you see | Less and less | More and more |
| Cost | Management overhead, ongoing | One-time adjustment cost |
When to admit you picked the wrong system, not that users are the problem
Everything above is about cases that can be saved. But I won’t lie to you: some can’t, and the longer you drag them out, the more they cost.
As a consultant, I think I owe you a way to tell. If these signals show up together, the question isn’t how to raise usage. It’s whether this thing was the wrong pick from the start.
Signal 1: whatever you ask for, the vendor’s answer is always “that needs customization.” Not occasionally — nearly every requirement. That means the system was never designed for your industry. You’re not implementing. You’re paying the vendor to build a new product.
Signal 2: changing something small takes more than a month. Response speed is the thermometer for fit. If adjusting a field or changing an approval level means waiting in a queue for scheduling, ongoing optimization after go-live is close to impossible. And implementations survive on ongoing optimization.
Signal 3: the people on the floor can’t name a single thing that’s “better now.” Ask five actual users. If none of them can name a concrete benefit the system brought, it’s not that they’re ungrateful. There really isn’t one. If even one step got better, someone would remember. If nothing did, nothing did.
Signal 4: more time goes into working around the system than into the system. When the company fills up with patches like “do that in Excel first, then paste it in” and “the system can’t do this, we use a LINE group,” your system has already been demoted to a data warehouse.
Signal 5: swap in a different person, and they get stuck in the same place. This one is key. If the sticking point only happens with certain people, it’s a training or habit problem. If everyone gets stuck at the same step, it’s a product problem.
If you’re feeling a chill reading this, I suggest reading Why ERP implementations fail and sorting out the type of failure first — wrong selection, wrong implementation method, or wrong expectations. Each is handled completely differently, and the wrong medicine makes it worse.
I have to say something honest that works against me: if you just want to digitize a few spreadsheets, and the company isn’t complex enough yet to need data linked across departments, you don’t need to spend this money at all. Some companies use Excel well and with discipline. That’s not a problem. That’s a stage. I laid out how to tell when it’s really time to switch in When to move from Excel to ERP. Switch too early and what you buy isn’t efficiency. It’s a pile of extra steps.
How fatal it is when the owner doesn’t use the system
This last section is the most important in the whole post, and the one most consultants don’t dare tell clients.
I’m saying it anyway.
If you don’t read your numbers from the system yourself, the system will never take off. No matter how much you spend, how many meetings you hold, or how many people you penalize.
Your behavior is the company’s spec sheet
You might think it’s reasonable not to go into the system yourself — you’re the boss, you need conclusions not line items, an assistant can prepare it for you.
But here’s what staff see: in meetings, the boss is holding an Excel printout. When the boss asks questions, it’s “did you update that sheet from last time.” The boss has never, in any setting, opened the system screen.
So the whole company gets a very clear message: the system is for us to fill in, not for the boss to read.
That settles what the system is. It’s a chore, not a tool. And since it’s a chore, people find the least-effort way to deal with it: fill it all in at month-end, fill it in carelessly, or just skip it. Your audits can catch missing entries. They can’t catch careless ones. And careless data is more dangerous than no data, because it makes you think you know.
Three things you have to do yourself
I’m not asking you to learn how to enter orders. The owner doesn’t need to operate it. But you have to do three things:
First, the numbers you look at can only come from the system. No Excel versions accepted. This forces every real problem into the open. If the system can’t produce the number you want, that’s the top priority to fix, not something to route around.
Second, in meetings, the screen shows the system. Not slides, not printouts. It doesn’t matter if the data is ugly and incomplete at first. Look at that ugly screen in front of everyone. Everyone will immediately know this is for real.
Third, when the system’s numbers don’t match what the floor tells you, chase down why instead of just believing the floor. Every time you choose to trust the system, its credibility goes up a notch. Every time you go around it, it drops a notch. There’s no middle setting.
This is the same thing that kills CRM
If you’ve implemented a CRM, this pattern will be familiar: sales doesn’t log visits, the manager nags, sales logs a pile of filler, the manager finds it useless and looks even less, sales notices nobody looks and logs even less. It’s a closed death spiral, and the only way to break it is for the people at the top to actually look. I break this logic down further in Why CRM fails. You’ll find ERP and CRM die for almost exactly the same reasons.
To put it more bluntly: you can’t outsource implementation to the vendor, and you can’t outsource it to IT. They can build the tool. But whether “this counts in this company” is something only you can define.
FAQ
The system has been live for six months and nobody uses it. Is it too late to fix?
It’s not too late, but first accept one thing: you can’t fix it by demanding harder. Still running both tracks six months after go-live means the sticking points that weren’t solved back then are still there, and staff have built habits to work around the system. The order of the fix: go watch actual work on site, find the top three high-frequency tasks, cut them to three steps or fewer, then personally tell users what you changed. After those four steps you’ll usually see a clear difference. The worst move is announcing “Excel is banned starting next month.” That just moves Excel from the desktop to a personal cloud drive, and you can’t even see where the problem is anymore. Also be prepared: if there’s no improvement at all within three months of the adjustments, start assessing whether it’s a selection problem.
Can I just ban Excel and force everyone onto the system?
I don’t recommend it, and in the cases I’ve seen it almost never worked. A ban changes the behavior you can see, not how staff actually work. They’ll save Excel somewhere private, or switch to paper or LINE notes. You lose the ability to see what’s really happening, and the problem is still there, just underground. The worse side effect is that it turns a technical question — is this tool any good — into a management question — do people obey. After that, nobody will tell you what’s wrong with the system. The right order is to first make the system genuinely faster than Excel on a few high-frequency tasks, let people choose it themselves, and Excel will fade out on its own. You won’t need a ban.
Excel can do everything the system does. Why should I switch?
If Excel can handle your current size, then you really can hold off, and I won’t lie to you to close a deal. The point where Excel actually breaks usually isn’t “more data.” It’s “several people working at the same time, different departments looking at the same data, and needing to track who changed what and when.” When three people each maintain their own version, month-end reconciliation takes hours, or a mistake happens and you can’t find who changed it — that’s when to consider it. On the other hand, if you just want to digitize a few spreadsheets and your company’s data flow isn’t complex enough to get tangled, switching systems mostly gives you extra steps, not efficiency.
How do I tell if we’re in a normal parallel transition or if it’s already out of control?
Look at four things. First, is there a written end date for the parallel period, or just “let’s do it this way for a while.” Second, which way does the data mainly flow — system primary with Excel as backup, or Excel primary with the system filled in afterward. Third, when the two sides disagree, which one does the company agree is correct; if the argument never ends, it’s out of control. Fourth, when a manager wants a report, is it exported from the system or does someone pull one in Excel. If three or more of these land on the second option, you’re not in a transition. You’re in a chronic parallel track, and it needs to be handled as its own problem. It won’t fix itself.
I’m the owner. Do I really need to use the system myself? I’m busy.
You don’t need to know how to enter orders or configure anything. Hand that to others. But you have to do three things: the numbers you look at come only from the system, no Excel versions accepted; in meetings, put the system on screen directly, not a polished slide deck; and when the system’s numbers don’t match what people on the floor say, chase down why instead of just taking the floor’s word. Together these cost you a few hours a month at most, but they decide how the whole company sees the system. Staff never judge whether something matters by what the boss says. They judge by whether the boss does it. Every time you go around the system, its credibility drops a notch. There’s no middle setting.
If you happen to be stuck here
The hardest part of “nobody uses the system” isn’t technical. It’s that it’s hard to tell whether you have an implementation problem you can fix or a selection problem you can’t. The prescriptions are opposite: the first means keep investing in adjustments, the second means anything more you put in is sunk cost.
If you’re at that fork, do two things first. One, take the “healthy transition vs chronic parallel track” table in this post and check yourself against it, so you see clearly what stage you’re at. Two, go through the five wrong-selection signals one by one. If you hit three or more, don’t rush to spend more money on optimization.
If your situation is that the whole system stalled right after go-live and adoption never picked up, that’s not just an Excel problem, and the order of steps is different. I wrote that rescue process up separately in Staff won’t use the new system? A hands-on adoption rescue.
We regularly turn what actually happens on implementation floors into case studies and podcast episodes. They’re all about what to do “after the money’s already spent,” including what can be saved and when to cut your losses. If you’d rather work through it yourself first, start with the SME ERP buying guide and fill in the questions that got missed during selection. A lot of the time the answer isn’t in the system. It’s in the few questions nobody asked clearly at the start.




