In 1952, a psychologist named Ernest Dichter sat behind a one-way mirror and he watched a woman talk about cake. She wore a floral dress and she spoke with great conviction to the men in suits. She told them she wanted a cake mix that was fast and she said she wanted a cake mix that was easy.

The company listened to her and they made a mix that required only water. It was a technical marvel and the product failed in every grocery store in the country. Dichter realized the woman felt guilty because she did not work for the cake and he told the company to make her add an egg.

They changed the instructions and they made the woman crack a fresh egg into the bowl. The woman felt like a baker and the sales went up and the company became very rich. The woman did not know why she had lied and the company did not know how to ask the right question.

The Ghost in the Keyboard

We see this every day in the digital world. A creator sends an email to her subscribers and she asks them what they want to see on her channel. The subscribers sit at their keyboards and they feel like the best versions of themselves. They write down that they want thirty-minute documentaries on the history of trade routes and they say they want to learn about the complexities of tax law.

The creator reads these answers and she feels inspired and she spends four weeks and three thousand dollars on the salt trade video. It gets 842 views in its first week. The subscribers are not at their keyboards anymore and they are on their couches and they click on a video of a man throwing a microwave into a swimming pool.

This is the gap between the stated preference and the revealed preference. The stated preference is a ghost and it is the person we wish we were when someone is watching us. The revealed preference is the truth and it is the thing we do when we are alone with a mouse and a screen. The survey measures the ghost and the algorithm measures the truth. If you build your strategy on the ghost you will find yourself in an empty room and you will wonder why the people who promised to come are not there.

The Sophistication Trap

I design virtual backgrounds for people who speak on the internet. I once worked for a man who insisted on a background that looked like a modern art gallery. He said his audience was sophisticated and he said they valued minimalism and clean lines. I built the background and it was very white and it was very cold.

His engagement dropped by 31 percent in two months. I turned the project off and on again and I gave him a background that looked like a messy study with old leather chairs and a green lamp. He hated it and he said it was cluttered but his retention time went up by twelve minutes.

The audience did not want a gallery and they wanted a friend they could trust in a room that felt warm. The man had asked his audience what they liked and they had given him the answer they thought a sophisticated person would give.

The machine that runs the video platform does not care about what people say. It only cares about what they do. Here is how the process actually works. When a user clicks a video a server in a cold building receives a signal. The server identifies the IP address and it starts a timer.

If the user leaves early the machine records a failure. If the user stays the machine records a success. The machine does not read the comments where people say the video is educational and important. The machine only sees that the user stayed or the user left. The tally in the database is the only reality that the system understands.

The problem for a new creator is that the machine needs data to start the tally. If you have no views the machine has no information and it will not show your video to anyone. You can ask your friends to watch but they will lie to you because they love you. They will say the video is great but they will turn it off after twenty seconds to go get a sandwich. This ruins your retention data and the machine thinks your video is bad.

When the machine sees that 1,342 people have watched the video it begins to take the video seriously. It starts to show the video to real strangers and then the revealed preference begins to take over.

The Honesty of the Click

We are all bad at knowing what we want. We think we want the salad but we buy the burger. We think we want the documentary but we watch the drama. If you are a creator you must stop asking the audience for their opinion. You must look at their tracks in the snow.

If they click on the video with the red thumbnail and they ignore the video with the blue thumbnail you must make more red thumbnails. It does not matter if they tell you in the comments that red is an aggressive color. They clicked the red and the click is the only honest thing they have ever given you.

“The things we plan are often the things that people ignore. The things we do not plan are the things that reveal the truth of the human heart.”

I have made many mistakes in my work. I once spent three days perfecting the lighting on a virtual shelf of trophies for a client. I thought the trophies would show authority and I thought they would make the client look like a winner. The viewers did not look at the trophies and they looked at a small cat that walked across the floor in the background. The cat was not part of the design and the cat was an accident but the cat made the engagement go up.

Shouting into the Storm

The algorithm is a mirror of our worst and best impulses. It shows us that we are curious and it shows us that we are bored. It shows us that we are easily distracted and it shows us that we are hungry for connection. A survey cannot find these things because a survey is a formal conversation. A click is a private act.

If you want to grow you must become a student of the private act. You must watch the data and you must ignore the praise and you must ignore the criticism. The only thing that matters is the tally in the database.

When you start a channel you are shouting into a storm. You think that if you are loud enough the people will hear you. But the storm is made of millions of other people shouting at the same time. You need a way to make the machine hear you above the noise. You need the social proof of the view count to tell the stranger that your video is worth their time.

People are like sheep and they will follow the crowd even if they say they are individuals. If they see a video with ten views they will pass it by but if they see a video with ten thousand views they will stop and they will wonder what the fuss is about. This is the truth of human behavior and it is a truth that no one will ever admit in a survey.

We spend our lives building pedestals for an audience that only exists in the ink of our own imagination.

You must test everything. You must test the title and you must test the thumbnail and you must test the first ten seconds of the video. You must not ask for permission and you must not ask for advice. You must look at the retention curve and you must see where the line drops.

If the line drops at the three-minute mark you must look at what you said at that moment. You will find that you were being boring or you were being too clever or you were talking about yourself. The audience did not tell you they were bored but they left the room and that is the most important piece of information you will ever receive.

I have turned my own career off and on again many times. Each time I learned that the things I thought were important were usually the things that nobody noticed. I thought the texture of the virtual wood on a desk mattered. It did not matter. The only thing that mattered was whether the person on the screen looked like they were telling the truth.

Finding Your Egg

The audience can smell a lie even if they cannot describe the smell. They know when you are making the video they think you want and they know when you are making the video you actually care about.

The egg in the cake mix was a trick but it was a trick that respected the human heart. It gave the woman a way to participate in the reality of the baking. Your job as a creator is to find the egg. You must find the thing that makes the audience feel like they are part of the story.

You will not find it in a questionnaire and you will not find it in a focus group. You will find it in the data of the click and the tally of the view and the silence of the people who stayed until the very end. The survey is a liar but the machine is a witness. You must choose which one you are going to believe.