Ron Johnson stood in a JCPenney in 2012, looking at a store he had stripped down and rebuilt from scratch. No coupons. No sales racks. Open floors. Clean lines. The Apple store dropped into a department chain. He was not wrong about design. He was wrong about who was in the room.

Same-store sales fell 25% in one year. Revenue dropped $4.3 billion. The board fired him after 17 months. His replacement brought back coupons on the first day. The old customer came back within a quarter.

"Define your ideal customer avatar before you build."

Every coaching program, every positioning course, every business book for solo operators asks you to fill out a worksheet. Age. Income. Pain points. Dreams. You sit alone in a room and imagine a person. Then you build for that person.

It sounds right. It is not.

Johnson ran Apple Retail before JCPenney. He opened over 300 stores that earned more than $6,000 per square foot. When a reporter asked why he never tested the new format with real shoppers, his answer was plain: "We didn't test at Apple."

He defined the JCPenney customer from his own prior frame. The Apple shopper. The person who pays full price and prefers a clean store with no red tags. That person was not in JCPenney. The person who was there loved coupons, hunted for markdowns, and came because of the deal.

Johnson built a $4.3 billion mistake by imagining who he wanted to serve instead of reading who was already there. That is not a strategy problem. It is a measurement problem.

There is a name for this in the literature. Projection bias. You assume others share your preferences, your frame, your way of seeing the world. The avatar worksheet turns this error into a formal process. It asks "who do you want to serve?" and skips the one question that matters: "who is already here, and what do they do?"

The problem is not imagination. The problem is that imagination replaces observation.

Read behavioral data before you build a profile. Not after.

In early 2026, Alesia Zakharova ran a test on her Substack. Her bio at the time said she wrote for "general AI readers." Growth was flat. She had no budget for ads and no large audience to pull from.

Instead of filling out an avatar worksheet, she did something different. Once that method is clear, three moves follow from it.

Zakharova pulled up every person who had shared, commented on, or restacked her posts in the past 30 days. Not all readers. Just the ones who moved. She built a list of names, not a list of traits she hoped for.

This is the part that takes honesty. You are not asking who you wish would show up. You are looking at who already did, and letting that answer be enough.

She gave the list of engaged readers to an AI model and asked one question: what do these people have in common? Not age. Not income bracket. Not dreams. Behavior. What do they do, over and over, that brought them here?

The answer came back clean. Every engaged commenter was a newsletter writer. Not a demographic. Not an industry vertical. A shared action. A pattern that was visible the whole time, buried under an avatar she had made up at her desk.

She rewrote her bio to speak to newsletter writers. One line. One afternoon. No redesign, no rebrand, no new product. The same content, pointed at the real audience instead of the made-up one.

Within 30 days she had added 213 subscribers. Open rates rose on posts she had published months before. The same writing worked better once the positioning matched the people who were already reading it.

Running this for 30 days does something the avatar exercise never did:

It shows you who already cares, not who you hope will care. It shows you what behavior your existing audience shares, beneath the surface of job titles and age brackets. It shows you where your positioning has been lying to the people already in your data. And it costs nothing but one afternoon of honest reading.

At the end of 30 days, ask three things.

→ What moved the number? Which line or phrase in the new positioning pulled readers from the same behavioral pool?

→ What looked like progress but left no trace? Which content got likes but produced no shares, no comments, no subscriptions?

→ What friction showed up more than once? Which type of reader kept arriving but never engaged past the first visit?

That is the difference between advice that sounds right and a system that proves itself.

Johnson's replacement brought back coupons. The existing customer returned within a quarter. The audience was never missing. It was ignored.

Your audience is not hiding inside a brainstorming exercise. It is sitting in your data, in the comments, in the shares, in the pattern of who already moved. The worksheet was never going to find them. They were already there.