Insights

Thinking you can check my work against.

Practical notes on campaign decisions, customer understanding and AI-assisted work, grounded in my marketing experience.

Generic AI output is a context problem, not a prompt problem. Build the reference library, then check what the tool can read.

AI can write a campaign in seconds. It can’t decide who the campaign is for.

AI answers don’t need a new bag of tricks. The fundamentals still apply, and they’ve become less forgiving.

Lessons from portfolios of five, nine and fourteen brands: distinct positioning, shared infrastructure and budget that follows evidence.

I use AI every day. Four marketing jobs it still can’t do: read the room, make the call, earn trust and know when to stop.

Acquisition gets the applause; retention pays the bills. Lifetime value, the flows that matter and honest measurement.

Whatever happens to cookies, own the relationship. Collect data people are happy to give, and use it to serve them better.

Win the research months, show the machine working and stay present between purchases.

AI made the in-house vs. agency debate louder, not simpler. Start with the work, then decide who should do it.

Forget the fixed split. Ask whether each line moves something now, and whether it keeps working after the campaign ends.

AI agents can run multi-step tasks. Which jobs to hand them, what they must check and where a person still approves.