When I started my marketing career at MediaCom, a WPP agency, I was running PPC campaigns for household names. Everything was manual back then. Yes, I'm old.

Even tracking URLs had to be built keyword by keyword. Ad copy was written and reviewed one ad at a time. The whole thing came together through spreadsheets and various uploads. There was a craft to it. A beautifully constructed campaign was, in its own slightly nerdy way, a work of art.

As campaigns grew more complex, automated bidding became increasingly important. I remember meeting one of the early pioneers of that shift. As a physics graduate, I had an immediate affinity for the maths behind what they were doing. Here was a way to apply more computational power to decisions we'd been making manually.

Perhaps that explains why I saw an opportunity to wield more powerful tools for growth. I think something else mattered too: I understood my job as helping clients deliver value for their businesses. How I did that was open to evolution.

The task can change. The purpose can endure.

That experience shapes how I think about AI today.

Some of the tasks we've spent years mastering are becoming cheaper, faster and easier to perform. Taking pride in that expertise is understandable. We've invested time in learning the craft, and doing it well has mattered.

The hours I spent constructing tracking URLs were necessary at the time. They helped us understand what our campaigns were doing. Automating that work gave us a reason to ask what deserved our attention next.

A person's worth doesn't fall because a task can be automated. The commercial case for spending their time on that task can change. We need to be honest about both.

Repetitive work can be essential. Accuracy, consistency and reliable execution matter. When a tool can deliver those things more efficiently, the useful question is what our experience allows us to contribute next.

Give the time you save a job

For marketers, that contribution needs to be specific.

If a weekly report takes less time to assemble, use some of that capacity to investigate why conversion has fallen. If campaign production becomes faster, spend time understanding the customer problem the campaign is supposed to address. If bid adjustments need less manual attention, examine whether the target reflects the customers and returns the business needs.

Each of those activities has a commercial purpose. Better decisions, fewer costly mistakes and stronger customer understanding give the time saved somewhere useful to go.

Producing more work is easy to count. Judging whether the work deserves to exist takes more thought.

Leaders have a responsibility here too. If automation releases capacity, name the business problem you want people to tackle and give them room to learn. A team needs time to test new methods, challenge outputs and develop its judgement. Filling every recovered hour with more of the same leaves that opportunity unexplored.

Choose one task to rethink

Start with one repetitive task you do each week. Write down what it helps the business achieve and what can go wrong when it's done badly.

Try a tool on a limited piece of that work. Check its output against the standard you'd expect from yourself. Include the time spent reviewing and correcting it when you judge whether it helps.

Then choose where any saved time will go. That might mean investigating a performance problem, speaking to customers or designing a test that could change an investment decision. Be clear about the result you're trying to improve.

This won't make anyone immune to changes in the employment market. It does give us a practical way to keep developing our contribution as the tools change.

My advice is to get involved in changing how your work gets done. Learn the tools. Challenge their outputs. Find out where they help and where your experience exposes their limitations.

I wanted to help businesses grow. That purpose has given me plenty of reasons to change how I work.

If the repetitive half of your work disappeared tomorrow, what would you want to be paid for?