Part three

The open ground

It was 2025, and AI had started eating into search.

We'd just done some of our best work for a lender we worked with. A new content strategy, an upgraded website, tools that helped buyers work out what they needed. By every early sign we watched, it was working. People were finding the site, spending time on it and using the tools.

Then the leads started to dry up, and the deals went with them.

I couldn't work it out. Nothing we'd built had changed. Every engagement number we reported was still healthy. We were doing the hard work, and doing it well. But the numbers that mattered to the client were falling, and they pressed us hard to account for it. Fair enough, too.

When they asked, I told them the truth: "I'm honestly confused, and a little bewildered." Soon after, they halved our budget.

Shortly afterwards I found the culprit. AI answers were taking the searches our content used to win. Buyers hadn't all moved, but enough of them were getting their answers from an AI assistant to cut the client deeply.

The answer came too late to save the account. Within a few months we'd handed the work back.

I had to admit I was out of my depth. Getting a brand into AI answers was a new discipline, and I set out to learn it.

Before the foundations, an admission. I looked hard for a company that has published a measured case of winning at every stage. I didn't find one. That doesn't mean nobody does it well. It means nobody publishes it: every team owns a number and nobody owns the choice.

Why winning the whole choice is rare

The arithmetic shows why it's rare. Take a considered or committee purchase, where the buyer needs all seven stages to hold.

Suppose a brand wins each stage four times in five. It wins all seven about one time in five.

In a shortlist of five suppliers, one in five is only fair share. A brand that's good at every stage is merely average.

A single weak stage does more damage. Let Believed slip to one in two and the whole falls to about one in eight.

Treat that as an illustration, not a model. The stages aren't independent, and buyers don't pass through them in order. Not every purchase needs all seven, either. But for considered and committee purchases, the point stands. A brand that's good everywhere can still lose most of the time.

What AI changes

AI raises the stakes, because a machine now writes more of the shortlist before any supplier is called. It also makes the stages easier to see: you can ask the machine what it knows, believes and prefers. It doesn't make them easier to win. Your rivals can ask the same questions, and the machine leans toward names it already knows.

So the ground is open to whoever manages the whole choice, rather than one stage of it. The rest of this book is how.

What it takes: three foundations and seven stages

Being selected isn't a campaign. It rests on three foundations, which every stage draws on, and then on seven stages won one moment at a time.

• Hold one truth. A single account of who you're for, what you claim and your proof, told the same way everywhere. Chapter 10.
• Have a reason. A reason to choose you that a sensible rival wouldn't claim. Chapter 11.
• Know your moments. Which buyers, at which steps, in which way of buying. Chapter 12.

Then come the seven stages, one chapter each, each built on one line.

Each foundation chapter ends with something to do this afternoon and this quarter. Every stage chapter ends with its core tool, printed in full; the rest are free in the Selectability Playbooks at rocmode.com/playbooks.