Somewhere in most organisations is a system that cost a lot of money and nobody uses. The licences are still being paid. The training happened. People smiled through the launch, then quietly went back to the spreadsheet.
It is almost never the software
The instinct is to blame the tool, and occasionally that is fair. Far more often the tool is fine and the problem sits around it. Adoption fails for human reasons, and no feature release fixes a human reason.
Nobody explained what changes for them
Launches usually explain the system. What they skip is what any individual should now do differently on a Tuesday afternoon. People are not resisting, they genuinely do not know what the new expectation is, so they default to the thing that already works.
The old way still works
If the spreadsheet still exists and nobody minds, the spreadsheet wins. It is faster for them, they already know it, and there is no cost to carrying on. A new system only gets adopted when the old route is genuinely closed, or when the new one is plainly better for the person doing the work, not just for the person who bought it.
Training happened once, at the wrong time
Most rollout training lands weeks before anyone needs it, covers everything at once, and is never available again at the moment someone is stuck. That is not a training problem so much as a timing one. People need help at the point of use, not at the point of purchase, which is why a short reference someone can find in ten seconds often beats a two-hour session nobody remembers.
Nothing rewarded the change
If using the new system takes longer, and nothing acknowledges that, people will do the fast thing. If managers still accept the old format, the old format survives. Adoption follows what the organisation actually rewards, which is rarely what it announced.
What actually works
Decide the specific behaviour you want, and how you will know it happened. Make the new route genuinely easier than the old one, or close the old one properly. Put support where the work happens rather than in a session months earlier. And have the people who set expectations use it themselves, visibly.
None of that is about software. It is the same reason good learning goes unused: the thing was built well and the conditions around it were never designed. When we build a tool, or the training that goes with it, that is the part we plan for first, because a system nobody uses is the most expensive kind.