The short answer
- People revert when the system is slower than the workaround, not because they resist change
- Make the correct path the fastest one, or it will not be taken
- Roll out one function at a time, starting with intake
- A shadow spreadsheet means the system is missing something — find out what
- Half-entered data is worse than none, because people stop trusting it
Shops buy software and then discover, six months later, that half the shop is still working the old way. Orders get taken on paper, statuses go stale, and someone maintains a private spreadsheet that is more accurate than the system everyone paid for.
That is almost never about people disliking change. It is about the system being slower, at the moment of use, than the thing it replaced.
Why people work around it
| Reason | What it looks like |
|---|---|
| Slower than the old way | Orders still written on a pad |
| Too many required fields | Junk entered to get past a screen |
| Not available where the work is | Statuses updated at the end of the day, from memory |
| Nobody showed them | One person does everything in it |
| It was wrong once | Quiet loss of trust; parallel records appear |
| No consequence to skipping | Gradual drift back |
The third row is specific to production and frequently overlooked. If updating a job means walking to an office, it will be done in a batch at five o'clock from recollection — and recollection produces exactly the errors the system was meant to remove.
People do not resist software. They take the fastest route available, every time.
Make it the shortest path
The correct path has to be genuinely quicker than the workaround, at the moment somebody is deciding which to take.
What that means practically
- A screen or scanner where the work happens, not only in the office
- Required fields cut to what is genuinely required
- Defaults for everything predictable
- One-click status updates rather than a form
- Search that finds a customer from a partial name
- The job sheet printable in one action
Required fields deserve scrutiny. Every mandatory field is a small tax at the moment of entry, and a form demanding twelve things when four are known produces either abandonment or invented data. Both are worse than a record with four accurate fields.
Barcoding is the strongest version of this for production — scanning a job at each stage is faster than writing anything, so status stays current as a by-product of the work. That is what our production management is built around, and it is the difference between a board that reflects the shop and one that reflects intentions.
Roll out in order
One function at a time, each embedded before the next. A shop switching everything on one Monday will be running two systems by Friday.
| Stage | Function | Embedded when |
|---|---|---|
| 1 | Order intake | No orders on paper for two weeks |
| 2 | Job status | The board is right without being asked |
| 3 | Proofs and approvals | No proofs sent from a personal inbox |
| 4 | Purchasing | No purchase orders typed into a portal |
| 5 | Reporting | Numbers are trusted enough to act on |
Intake first, always. Everything downstream depends on the order existing as data, and a system fed by retyped emails will always be a step behind reality. That is the same reasoning as the order you automate in.
Training that works
Not a session. A session teaches a system in the abstract, and people forget it by the time they need it.
What actually sticks
- Train on real jobs, in the week they will be doing them
- One person per role, not everyone on everything
- A one-page reference at each workstation, not a manual
- A named person to ask, who is not always you
- Ten minutes a week for the first month to catch bad habits early
- Train the same thing again after a fortnight
The one-page reference is the highest-value item and the least often produced. Three or four steps for the tasks that person does daily, printed and stuck up. It removes most of the questions that otherwise come to you, and it is what makes the next hire straightforward.
Repeating the training after a fortnight catches the habits that formed in week one. People invent workarounds early and then keep them, so the second session finds far more than the first.
The shadow spreadsheet
Somebody in the shop keeps a private record — a spreadsheet of jobs, a notebook of prices, a list of who owes what. It is a reliable signal and worth investigating rather than banning.
It means the system is missing something they need. Usually it is a view rather than data: everything is in there, but not arranged the way that person works. Sometimes it is a genuine gap.
How to handle it
- Ask what it holds and why — without implying they are at fault
- Determine whether it is a missing view or missing data
- Fix the system so the spreadsheet is unnecessary
- Then agree to stop maintaining it
- Never ban it before removing the reason for it
Banning it first produces the same spreadsheet, hidden. And the person keeping it is usually the most organised person in the shop, which makes them the best source of information about what is wrong.
Keeping it clean
Trust is the whole asset. Once people believe the system is wrong, they stop relying on it, and parallel records reappear.
Two habits maintain it. Fix errors visibly — when somebody reports something wrong, correct it and say so, because a reported error that is not fixed teaches everyone not to report the next one. And keep entry honest: a status marked complete when it is not, or a date changed without a reason, does more damage than a missing record.
Individual logins matter here for the same reason they matter for security — when a change can be traced to a person, the standard of entry rises without anyone being asked. That is a permissions decision with a cultural effect. The rest of the software archive assumes the system is the fastest way to do the work.



