Take apart the thing that worked on you

The habit that separates people who improve

Most people consume a great deal of visual work and learn almost nothing from it, because watching something is not studying it. The habit that changes this is small: when something works on you, stop and find out why while the effect is still fresh.

The crucial part is on you. Not a piece that won awards, not a piece somebody told you was good: a piece that actually made you stop scrolling, or stay to the end, or believe something. That reaction is a measurement you have access to and nobody else does, and it is about to fade.

The deconstruction, in seven questions

Take one piece and answer these in writing. It takes fifteen minutes.

1. What did the first second promise? Write the promise as a sentence.
2. Was it paid, and where?
3. What are the beats? List them and write the joining word between each pair: but, therefore, or and then. Count the *and then*s.
4. What is level one? Squint or blur it. What survives?
5. Where did my eye go first, and what put it there? Size, contrast, position, isolation, direction.
6. What was left out? This is the hardest and the most valuable: the shot they did not use, the information they withheld until later, the thing a weaker version would have explained.
7. What would I have done that they did not? Then ask why they did not, and take seriously the possibility that they had a reason.

Question six teaches more than the other six together, because addition is visible and subtraction is not. Most of the craft in a strong piece is in what is absent.

A file of examples with reasons

Collecting work you admire is common and mostly useless. A folder of screenshots is a mood board; it teaches taste by exposure and nothing else.

What works is the same folder with one sentence per item naming the mechanism: not “nice poster” but “one message survives the blur because everything except the date is at level three”. Now the file is searchable by problem. When you are stuck on hierarchy you can look up six examples of hierarchy with notes, rather than scrolling a wall of images hoping to be inspired.

Practice needs feedback, and creative work does not supply it

The conditions under which practice reliably produces improvement are well described: a specific sub-skill rather than the whole activity, repetition just beyond current ability, and feedback close enough in time to connect to what you did.

Creative work fails the third condition badly. You make a piece, it goes out, and what comes back is a number affected by timing, luck and distribution, weeks later, aggregated across every decision you made. There is no way to learn from that which choice was the good one. This is precisely why people can do this work for years and get no better, not from lack of effort, but because the loop never closed.

So you have to construct the feedback:

• Redo with a known answer. Take a piece that worked, recreate your own version from the same raw ingredients, and compare. The comparison is the feedback.
• Predict before you look. Write down what you expect the retention curve to do and where. Then look. Being wrong in a specific place is worth more than any amount of general viewing.
• Ask one person for one thing. “Where did you get bored?” answered honestly beats an hour of general praise.
• Separate the sub-skill. Spend a week on titles alone. Retitle twenty of your own old pieces. Nothing else.

Four drills that cost nothing

1. Retitle ten pieces you have already published, applying the front-loading rule, and note which original titles were doing nothing.
2. Re-cut a 90-second film in your head from its own material, writing the beat list and your joining words, then compare to the real order.
3. Redraw a bad diagram (a leaflet, a form, a sign) by first writing the one thing it must convey.
4. Blur-test five designs a day for a week, from anywhere: a shop window, a packet, a poster at a station. Say what survives before you unblur.

The limitation

Deconstruction is reconstruction. You are inventing an account of somebody’s intentions from the outside, and a good deal of what you attribute to a decision was an accident, a constraint, a client demand, or the only shot that came out usable. Your explanation will often be more coherent than what happened.

That is fine for learning and dangerous for arguing. The exercise builds your vocabulary for deliberate choices, which is its purpose. It does not entitle you to say what anyone did or why, and a portfolio of confident reverse-engineered explanations is a recognisable kind of wrong.

Hold your accounts loosely, keep collecting them, and judge them by whether they help you make the next thing rather than by whether they are true about the last one.