There is a particular kind of fatigue caused by a poor business system. The work itself may not be difficult, yet the person keeps stopping: searching for a document, asking whether it is safe to press a button, keeping a separate note, and checking whether the system has actually remembered what was done.
The answer to how technology can be humanised is not to make a system appear friendly on the surface. A system should help people understand the current state, the consequence of a decision, and the route back when they make a mistake. Only then does a digital workplace reduce human burden instead of merely moving it from paper to a screen.
A report can look orderly. The process is digitised, fields are completed, and steps exist. The burden often becomes visible only when we sit beside the person who performs the work every day. Where do they stop? What do they check twice? What information do they look for outside the system before confirming an action?
That is where humanising technology begins.
When a user keeps a parallel spreadsheet, it is easy to conclude that they are resisting change. Sometimes the explanation is simpler: the spreadsheet gives them a sense of control that the system does not. It may show what is waiting, whom they contacted, what is missing, and what will happen if something is delayed.
Banning that spreadsheet before having a conversation usually removes the sign of a problem, not the problem itself. It is more useful to ask:
The answer does not always need to be a new feature. Sometimes a status is missing even though it could already be displayed. Sometimes the next step needs a clear owner. Sometimes the same data is entered in two places, and no one knows which record to trust. Only after observing real work does it make sense to decide whether to change the tool, the process, or the allocation of responsibility.
An exception is not disobedience either. Real life does not behave like demo data. A customer changes a requirement, goods arrive incomplete, a colleague is absent, and a decision must be made before all information is perfect.
A human-centred system recognises that exceptions will exist. It does not allow everything without control, but it clearly shows who can decide, what information must be recorded, and how the case returns to an orderly flow. This protects both process discipline and people's ability to resolve a real situation responsibly.
The message "Invalid entry" says that something is wrong, but it does not help a person continue. They do not know which field is the problem, why it matters, or whether they will lose what they have already entered.
A better message can be very plain: "Delivery date is missing. Add it before sending the order. Other details have been saved."
This language does not patronise the user. It respects their time. It describes the state, the reason, and the next step. Where relevant, it should also explain what may happen if the action is not completed now and where the person can continue later.
Recovery from a mistake is as important as preventing one. No one works without mistakes, especially during interruptions, time pressure, or handovers between colleagues. It is therefore useful to check:
The same logic applies during the introduction of a new solution. If we tell people that the new system is simple, then treat their questions as resistance, we lose trust. It is more honest to say that some things will be easier, some different, and some temporarily slower while the process settles. That conversation gives people realistic preparation and gives the project more useful feedback.
In business software, one click can reserve stock, trigger a posting, close a work order, or send a document to a customer. A button is not simple just because it says "Confirm".
Before an important action, a system should show:
Not every action has the same risk. Too many confirmations can slow routine work and encourage people to read them mechanically. A warning should therefore relate to a real consequence, rather than be added to every small action. A frequent and easily corrected action may only need a clear status. An action affecting stock, a financial record, a deadline, or an external recipient needs a more understandable view of its consequence.
This is particularly important when AI suggests a decision. The person should see what the suggestion is based on, what data may be missing, and what happens if they reject it. An AI suggestion can help identify a pattern faster or prepare the next step, but it should not hide uncertainty or replace the responsibility of the person making the decision.
Before releasing a new function, it is worth observing one real task from start to finish. There is no need to begin with a large research exercise. It is enough to note several concrete moments during the work.
Observe hesitation: where does the person have to ask, guess, or look for guidance?
Observe context switching: how many times do they move to another screen, document, email, or personal note to complete the same task?
Observe consequence: do they understand what happens after confirmation, or do they discover the result only later?
Observe error recovery: can they recover without another person's help and without re-entering the entire case?
Observe exceptions: is there a legitimate route outside the standard flow, with clear accountability and a recorded reason?
Observe confidence: must the person keep a parallel record simply to trust the system?
The number of clicks can be a useful signal, but it is not enough. Five clear steps can be easier than two unclear ones. Likewise, a short extra step can be justified when it prevents an error with a greater consequence. The aim is not to remove every effort, but to remove unnecessary uncertainty and repetition.
Repetition, copying data, and searching for documents are good candidates for automation. Conversation, judgement, and a decision with consequences require context that should not be hidden.
A good system takes over the tedious part, keeps a trace, and warns of risk. It leaves people with understanding and responsibility. This is a healthy relationship between people and technology: the tool makes work easier, but does not force a person to become its translator, controller, and backup system at the same time.
When ORKA maps processes, it also looks at human burden: where a person stops, what they manage outside the system, and what they fear before confirmation. This information often points to a better direction than one more request for a new feature.
As a next step, choose one frequent task in your digital workplace and walk through it with the person who actually performs it. Record three moments of uncertainty and check whether they can be resolved with a clearer status, a better message, or an orderly route for an exception. For a broader view of this work, visit the ORKA Knowledge Centre or consider ERP and process screening .