How to Limit AI Output So CRO Work Actually Gets Implemented

I’ve spent days “prompting”…

And then, implementing nothing.

It’s because AI is brilliant at giving us a fire hose of information.

But it is terrible at understanding that we need time to implement.

• We need time to process.

• We need time to think.

• We need time to plan.

• We need time to prioritise.

• We need time to implement.

• We need time to measure

So the biggest ‘hack’ is to set limitations on LLMs so they only give us the information we need, when we need it.

Research Your goal here is to get clean, applicable data that is organised and ready for audit and analysis. You are not looking for suggestions yet. You are just turning messy inputs into something usable.

Audit/Analyse Your goal here is to understand what is happening. Not what you should change. You want clear observations and insights, not ideas for improvement.

Ideas/recommendations Your goal here is to turn observations into actionable tasks that have clear requirements and deliverables.

Prioritise Your goal here is to understand what is most important, what will give you the highest return on investment in the shortest possible amount of time.

Execute Your goal here is to successfully implement the recommendations in accordance with the requirements.

Measure Your goal here is to learn. Did it work, did it not work, and what does that mean for the next change? Then rinse and repeat.

Chat soon,

Peter