When your team won't use the new system: an adoption rescue

The system has been live for months, and people on the floor still have that old Excel file open. The owner's first reaction is usually 'our people resist change.' What I see on implementation sites is different: the new process really is slower per entry, running old and new systems in parallel makes people do the work twice, and the data in the system is wrong often enough that every use ends in embarrassment. This post breaks down the three real reasons employees don't use a new system, a diagnostic you can finish in one week, the rescue steps in the order you should take them, and the one situation where you should admit the mistake and cut your losses.

Johnny Yang12 min read
When your team won't use the new system: an adoption rescue

“The system has been live for almost three months. Last week I walked up behind our sales support staff member, and she still had that old Excel file open on her screen.” I hear this many times a year. The person saying it is usually holding back anger, and the next sentence is almost always: “Our people just resist change.”

I'm Johnny Yang, Brightstream Technology, Taiwan. I've led implementation teams for seven years, handled 252+ projects across 32+ industries. Let me start with something you may not like hearing: most of the time, employees are not resisting change. They are voting with their feet, and that vote is very honest — on their part of the process, the new system really is slower than the old way. If you define this as an “attitude problem,” you will never fix it. You'll keep holding meetings, running briefings, writing policies, while the reason she keeps opening Excel stays exactly where it was, untouched.

In the projects I've seen, when employees don't use a new system, the cause is rarely attitude. It is usually three things you can measure. First, the new process takes more time per entry than the old way, and you have never actually timed it with a stopwatch. Second, you let the old and new systems run in parallel, so employees have to enter the same data twice, and they rationally keep only the faster track. Third, the data in the system is wrong. Employees follow the system once, get burned by a customer or a manager once, and the second time they stop trusting it. So the rescue order is: go to the floor and time a single entry, then kill the parallel track, then fix the data until it can be trusted. Once those three are handled, adoption usually climbs on its own. If all three are handled and nobody uses it, then it's time to talk about people — or to have the nerve to admit the implementation scope was wrong from the start.

Where the owner's instinct is right, and where it's wrong

The right part first. There is a kind of employee who genuinely resists. Usually someone who has been at the company a long time and holds a way of doing things only they understand. A new system puts what's in their head on the table where everyone can see it. For them that is a real loss of power, not a matter of psychological adjustment. These people exist. Your instinct isn't wrong.

The wrong part is the proportion. On the floor, these people are a minority, usually one or two, and you can name them. But if you walk around and find a whole department not using the system, that isn't resistance. That's a process problem. One or two people not cooperating is resistance. A whole department not cooperating is a design failure. The two are handled in opposite ways: the first is a management problem, the second means you have to go back and change things yourself. The most common mistake owners make is using the first method (meetings, pressure, performance reviews) on the second problem. The result: employees nod in the meeting, go back to their desks, keep opening Excel, and never tell you the truth again.

The three real culprits behind an unused system

When I go in to diagnose, I check these three one by one. What they have in common: all of them can be measured, and none of them require a heart-to-heart with employees to find the answer. You can find it yourself by walking the floor.

Culprit one: the new process really is slower, and nobody measured it

This is the most common one, and the easiest to verify. What was the old way? The sales support staff opens an Excel sheet, drags down row by row, copies and pastes, handles ten entries at once, and never waits for anything to load. What is the new system? One form per entry, five required fields, two of them dropdowns that have to pull data from the server when clicked, then scroll to find the item, then a confirmation pop-up on save. Sit down and do ten entries yourself.

When I go in to diagnose, the thing I do most often is take out my phone stopwatch, stand next to her while she does one entry, and record the time. Then I ask her to do one the old way and record that too. It takes under twenty minutes, and almost every time it produces a number that ends the argument. If the new way takes a minute and a half longer per entry and she handles eighty entries a day, you are asking her to work two extra hours a day and then blaming her for not cooperating.

Let me be clear: a new system being slower per entry is sometimes a reasonable price, because it saves time later on reconciliation, stocktaking, and chasing orders. The question is who pays that price. If the person doing the input bears all the cost and all the benefit goes to accounting and the owner, nobody will use it voluntarily under that structure. Either you simplify input until it is close to the old way (cut required fields, add batch import, turn dropdowns into type-to-search), or you make sure the person doing input also gets a visible benefit — for example, she never again gets chased by sales asking “did that order ship or not.”

Culprit two: running both systems in parallel means doing the work twice

This is the most damaging one I've seen, and it is almost always the owner's own decision, made with good intentions: “Do both for now, to be safe. If the system breaks, we still have Excel as backup.”

It sounds responsible. In practice it is a death sentence for the whole implementation. Running in parallel means the same data gets entered twice, and the two sides slowly drift apart. Employees aren't stupid. She knows which version the owner will actually look at at month-end — usually the Excel, because the report format is the one the owner is used to. Since the Excel is what actually gets reviewed, the system becomes “I'll fill it in when I have time,” and there is never time. Three months later you look, the data in the system is full of gaps, and you conclude “the system is bad.” The truth is you never made it the only path.

My approach has always been: set a cutover date. After that date the old sheet is read-only. You can look up history, but nobody writes new data into it. This will hurt. Expect one or two weeks of chaos, and plan for someone to be on the floor during that time to fix problems as they come up. But that pain happens once. The pain of running in parallel happens every day, and it keeps hurting until you give up.

If you feel you don't dare cut over yet, that usually means your implementation scope is too big for one go. Which brings me back to what I keep saying: which process you start with decides whether the implementation survives. I go into this in more detail in which process to implement first. In short: pick a process that hurts enough, is independent enough, and won't paralyze the whole company after cutover. Get it working, then grow outward.

Culprit three: the data is wrong, and every use ends in embarrassment

The third culprit is less visible, but it is slow poison. An employee follows the stock count in the system and tells a customer “we have it in stock,” and the warehouse doesn't. She uses the unit price the system pulls in to write a quotation, and it's a price from three years ago. The printed invoice shows the customer's company name from before they renamed.

Let this happen two or three times, and trust collapses. Once trust collapses, her behavior becomes: check the system once, call the warehouse to confirm once, flip through the Excel to cross-check once. You think she's using the system. In fact she treats it as one of three sources, and the least trustworthy one. A while later, she just skips the first step.

Wrong data usually comes from three places: the opening data wasn't cleaned before import, changes after go-live didn't all make it into the system (usually the parallel track's fault again), or nobody was assigned to maintain the master data. The third is the most common and the easiest to fix — you need one named person responsible for the customer master and the part-number master. That person has the authority to change it and the obligation to change it. Not “if anyone sees something wrong, just fix it.” Data without an owner always rots. If you're still managing these masters with a few shared Excel files, the problem started earlier than you think. There's a fuller discussion in still managing with Excel.

A diagnostic you can finish in one week: skip the meeting, go to the floor

The worst way to diagnose this is to call everyone into a review meeting. Nobody tells the truth in a meeting room, especially with the owner at the head of the table looking unhappy. You'll get a pile of “we're still adjusting” and “it's mostly fine,” and walk away with nothing.

The right way is to schedule time to sit next to each key role for thirty minutes, say nothing, and watch how she works. Watch how many windows she has open, which step makes her stop and think for a long time, when she picks up the phone to ask someone. Then ask one question: “How long would this step take the old way?” Most people are very willing to answer, because someone has finally asked the real question.

This is the table I use on the floor. The symptoms look alike, but what you need to check is completely different:

What you seeMost likely real causeWhat to check
Only one or two people don't use it, everyone else is finePersonal factors or access not set upTalk to that person directly; first rule out an account problem
A whole department doesn't use it, but the department next door doesThe workflow design for that process doesn't fit how work is actually doneTime a single entry with a stopwatch, compare old and new
Everyone logs in, but data is only half filledToo many required fields, or fields asking for things they don't knowGo through each field: who is able to fill it, who should fill it
Used right after go-live, dropped after a monthGot burned by bad data, or the parallel track wore people downCheck what errors happened during that period
Front-end staff enter data, but nobody downstream picks it upThe next person in the chain wasn't included in the implementation scopeFind which handoff the process breaks at

Four rescue steps, in this order

First: kill the parallel track, set a cutover date, make the old sheet read-only. This hurts the most but goes first, because if the cutover isn't clean, everything after it is wasted effort — you won't be able to tell whether employees don't use it because the system is bad or because they simply have no reason to.

Second: cut the time per entry. Go back through the required fields and ask of each one, “what happens if this is left blank?” If nobody can name a concrete consequence, make it optional. Default whatever can be defaulted. Carry over whatever can be carried over from the previous order. If something can be batch imported, build an import function first so she has a way out. In this phase your mindset is “get her willing to use it,” not “the data has to be complete from day one.” Data completeness can be tightened later. Once the habit breaks, it's very hard to get back.

Third: fix the data and assign master-data owners. Pick the master files that are referenced most and go wrong most — customers, part numbers, unit prices. Set aside a day to clean them, then announce who owns each one. After cleaning, verify it in front of everyone: pull a few records on the spot, check them against the warehouse, and if they match, say so out loud. Trust grows back one verification at a time, not through announcements.

Fourth, and the only one that's about people: make “who is using it” visible. Not to punish anyone. To find problems. Look at usage by role once a week. When someone drops off, go ask her what happened. You'll usually dig up a workflow sticking point you didn't know about. The project lead should do this, not the managers. Otherwise it immediately becomes a performance-review tool, everyone starts clicking around the system to hit the numbers, and the data you get is worse than having none.

One situation where you should stop and admit you got it wrong

I don't like saying nice things that aren't true, so I'll say this too. If you've done all three — killed the parallel track, simplified the workflow, cleaned the data — and the floor still doesn't use it, you need to seriously consider one possibility: this process should never have gone into the system.

Some work is inherently highly variable. Every case looks different, and it depends on human judgment, not rules. Force it into a form and you get a pile of messily filled fields, then spend time every month cleaning up the garbage. In that case, going back to a well-designed form or a simple tool, and saving the system for the parts that are truly repetitive and high-volume, is the more honest decision.

Among the failed implementations I've seen, the saddest weren't the ones that picked the wrong vendor. They were the ones where you could see by month three that the scope was wrong, but they pushed on to month eighteen because “we've already spent the money,” and in the end the whole thing was shut down — taking the two parts that actually worked down with it. Cutting your losses is not shameful. Keeping the good parts and continuing to use them is the responsible thing to do.

What you can do tomorrow morning

Don't hold a meeting. Tomorrow morning, pick the role you think should be using the system most and is using it least. Bring your phone, sit next to her for thirty minutes. Watch her do one complete task and record the time. Then ask her to do one the way she did it before the system, and record that time. Then ask her three questions: which step gets stuck the most, which number in the system she doesn't dare trust, and if she could change only one thing, what it would be.

In half an hour you'll get more than three months of meetings gave you. Once you have it, deal with the parallel track first, then speed, then data. Don't change the order. If after this round you still can't tell where the problem is, or you want someone to walk through the current state with you, Brightstream offers a free 30-minute diagnostic. Come talk to us.

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