01 / Another dashboard?

We automated the task. Now someone has to watch the automation.

Automation normally starts with the right intentions. Remove repetitive work. Make information easier to find. Help a team respond before a small issue becomes a bigger one. All sensible things to want.

Then, somewhere along the way, someone suggests a dashboard.

Don't get me wrong. I like dashboards. I've spent enough time around digital products and operational tools to appreciate a good one. But I also wonder how often we solve one problem by creating another place someone is expected to visit, monitor and maintain.

We've saved people ten minutes of copying information between systems. Brilliant. Now they have a new tab to check every morning. And perhaps a notification telling them to check the tab. At some point, the dashboard itself starts needing a dashboard.

The question I find more useful is not simply what can we automate? It is where should the result of that automation appear, and what should someone be able to do with it?

Sometimes a dashboard is exactly the right answer. Sometimes it is an unnecessarily elaborate way of saying, "This task is ready for you."

02 / Start with friction

The technology can wait. Show me the frustrating bit.

Before talking about platforms, integrations or all the clever things a system could do, I want to understand how the work happens today.

Where does someone copy the same information twice? Where does a request sit because nobody is quite sure whose turn it is? What gets forgotten? What makes a person open three different systems just to work out whether something has happened?

Those little interruptions are easy to miss in a process diagram. They are not so easy to miss when you're the person dealing with them every day.

Rather than automating the entire process in one heroic effort, I'd start by choosing one repeated hand-off. Understand what triggers it, what information is needed, what should happen next and where the exceptions live.

And sometimes the discovery is that we don't need automation at all. Perhaps the process needs a clearer owner, one unnecessary approval removed or a little less confusion about the next step. There is no shame in solving a problem without creating a new system. I promise.

03 / Where work happens

A useful result should arrive where the work already happens.

If a team already manages its work in a task system, why make them visit a separate dashboard to discover that a task is ready?

A clear update in the existing record, a new item in their current queue or a message in a collaboration tool might do the job. A reminder can sometimes solve a hand-off problem more effectively than an entirely new workflow.

The important part is that the output arrives at the point where someone can understand it and act. Not hidden behind three clicks and a filter called "Pending miscellaneous".

Good design is not always about creating another interface. Quite often, it's about making the existing experience work a little better.

Of course, there are situations where teams genuinely need to inspect a changing system, compare trends or investigate incidents. That's when a dashboard earns its place. But it should be a deliberate choice, not the default souvenir from every automation project.

04 / Useful notifications

Please don’t notify me just because you can.

Once something can send a notification, it becomes surprisingly tempting to send lots of them.

But an alert that says "A status has changed" often creates more work than it removes. Which status? On what? Does it matter? Am I meant to do something, or am I simply being kept informed against my will?

I'd rather every important notification answer three questions: What happened? Why does it matter? What happens next?

For example, instead of announcing that a request has moved to another status, tell the person responsible that the request is ready for review, include the reason and link to the exact item they need.

And if nothing needs anyone's attention, perhaps record the event quietly. Not everything needs to interrupt a person who is already trying to get something done.

An automation that makes work invisible until someone needs to act can be more useful than one constantly reminding everyone how busy it is.

05 / Leave room for people

Automate the predictable. Keep judgement where it matters.

Some tasks repeat because they follow a dependable pattern. Others look repetitive until an unusual customer request, sensitive decision or unexpected exception arrives.

That distinction matters. Automation can gather information, check straightforward conditions, prepare a draft or highlight something that deserves attention. It does not always need to make the final decision.

A well-placed review step is not necessarily a failure to automate properly. Sometimes it is the part that keeps the system useful, responsible and understandable.

People should know what the system has done, what it has not done and where their judgement is still needed. Otherwise, we risk replacing a slightly awkward manual process with a very confident automated one that gets things wrong faster.

06 / Keeping it useful

Someone still has to look after the clever little thing.

There is a less glamorous side to automation, and I think it deserves more attention than it usually gets during the exciting part of a project.

Who owns it after launch? How will anyone know when it fails? What happens when a connected system changes a field name, an access permission disappears or the business process moves on?

A process that silently stops working can be worse than the manual process it replaced. At least with the manual one, someone usually knows where the work is piling up.

For each automated step, I'd want a simple explanation of what starts it, what it changes, who looks after it and how it can be paused or corrected. Nothing elaborate. Just enough for the next person to understand it without having to reverse-engineer the entire thing over a cup of increasingly cold coffee.

And there's one final test worth keeping: has this actually reduced friction, or have we simply moved the work somewhere else?

07 / Start small

Make one thing easier before making everything automatic.

If I were starting an automation initiative, I'd pick one repeated hand-off that genuinely gets in the way. Build the smallest workable improvement. Put the result where the right person already works. Then observe what happens.

Did the team spend less time chasing information? Was the next action clearer? Were exceptions handled properly? Did we remove a burden, or did we introduce a new one?

If the experiment works, improve it. If it doesn't, change it. A small, honest experiment teaches us far more than a beautifully designed control panel nobody really needed.

For me, useful automation is not about making every step automatic. It is about making the overall experience easier to understand, maintain and use.

And if we achieve that without adding another dashboard, I think we've earned ourselves a small celebration. Preferably one that doesn't require a notification.

Share this thought
Share on LinkedIn ↗