Scaling Expertise Is Hard

Scaling Expertise Is Hard

The hardest part of running a marketing agency? Scaling expertise.

Building WebSavvy taught me this lesson early.

No matter how many SOPs & guidelines you have, smart people will want to develop their own approach. And you want to have 'standard' ways of working, but also encourage creativity & new methods. It's a fine line.

  • Documentation becomes outdated quickly (we all know this in digital marketing!)
  • Training programs can't cover every scenario or industry
  • Individual experts can only (properly) manage so many accounts
  • Experts often have equally valid but different approaches

This is where agents could really change things. But I'm getting ahead of myself...

The Journey from Scripts to Intelligence

When I first started using scripts, the goal was simple: automate repetitive tasks. But as my agency grew, we needed more than automation. Gen AI didn't exist then, so we hired & trained & managed.

Humans.

Messy, wonderful, intelligent, unpredictable humans.

A decade ago, one of my team asked me if they could have their own assistant. How much more high-level work will that enable you to do, I asked. "Oh lots!" was the assistant-hiring answer.

It was a model that worked.

Back in 2016 I taught my 10-80-10 model to hundreds of other agencies & they scaled too.

I didn't know it then, but this model works exceptionally well in a world of AI Agents.

The 10-80-10 Model

Your goal: to focus on the first & last 10% of a task/project.

The first 10% is about defining what great looks like. How will we know when we get there? What are the milestones? What exceptions do I need to be notified about? What skills are required for this task/project. What's the real outcome we want?

The last 10% is all about improvement. Feedback loops & training. Making sure the work gets done even better next time. It's the opposite of what many people do when hiring a VA for the first time... they get the big red pen out, start making corrections & think "this would have been faster to do myself".

Well yes it would, but if you can hire 3 people who each do a 60-80% version of what you would have done AND then you train them, so over time they get to 90-95% of 'you'. You've freed yourself up to spend time on your business, or with your family, or riding a mountain bike through a forest.

This same model applies perfectly to working with AI agents.

Share: