Make it pop is a symptom report

Clients describe, they do not prescribe

"Make it pop." "It needs more energy." "Can the logo be bigger?" "I do not like the blue."

The beginner move is to hear an instruction and execute it. The logo gets bigger, the blue becomes green, the heading gets a shadow, and the client is still unhappy, now with a worse layout.

These sentences are symptom reports. They are the client telling you, in the only vocabulary they have, that something did not work when they looked at it. They are usually right that something is wrong and usually wrong about what. Translating symptom into cause is the job.

Three separate acts, belonging to different people: description (what it feels like — theirs), diagnosis (what is causing it — yours), prescription (what to change — yours). Trouble comes from accepting their prescription and skipping your diagnosis.

The common translations

Honest, from the mechanisms in the earlier lessons:

• "Make it pop." Almost always insufficient hierarchy — nothing arrives first — or insufficient contrast. Occasionally too little whitespace around the main element. Almost never a gradient, and never a drop shadow.
• "The logo should be bigger." Usually means "I could not tell whose this was in the first second". Size is one fix; isolation is usually a better one, and a small mark alone in clean space reads as more confident than a large one wedged into a corner.
• "It feels cheap." Usually inconsistency: four typefaces, uneven gaps, a stretched photograph, three corner radii. Run the inventory check.
• "It needs more energy." Usually a crop, a diagonal, and a real contrast jump — not more colours. Adding colours reduces energy, because it removes the difference that made one thing loud.
• "I do not like the blue." Sometimes exactly that, and sometimes the blue is the only thing they have a word for.

The two questions that do the most work:

1. 1"When someone sees this for two seconds, what should they walk away knowing?"
2. 2"What are they walking away with now?"

The gap between those two answers is the actual brief, and it is often the first time anyone has written it down.

Never ask whether they like it

"Do you like it?" has no answer that helps you, and it invites a person with no training to redesign your work out loud. Ask a task question instead: "which of these two gets the price across faster?" Taste has no answer; a task has one, and the client is genuinely the better judge of their own customers.

Show two or three options. One invites line edits, because there is nothing to compare it against; eight invites a request for a combination of four and six. Never include an option you do not want chosen — people choose it, and then you have to build it.

Critiquing your own work

You cannot see your own layout because you remember making it. These are the ways round that, in rough order of value, and every one of them is free.

• Sleep on it. Time weakens your memory of the intention, so more of what is actually on the page reaches you.
• Mirror it. Flip the whole thing horizontally. This is the single most effective trick in the list: it breaks the reading habit, so you stop reading and start seeing shape, and alignment and balance faults jump out immediately. Every tool has it — Image, then Transform, then Flip Horizontal in GIMP, Photopea and Krita. Flip it back to work.
• Thumbnail it, desaturate it, print it, and look at it on a phone you did not design on.
• The five-second test. Send it to a friend, tell them they get five seconds, then ask two questions: what was it about, and what were you supposed to do next. This takes one message and gives you more than an hour of staring.

Taking criticism without losing the day

Separate "this does not work" from "you are not good". The first is information about an artefact; the second is a story you added. Ask for the symptom, not the fix: "what were you looking for when you got stuck?" beats "what would you like me to change?"

You are allowed to defend a decision once, with a reason, and then let it go. If three different people trip over the same element, they are not all wrong about it, however clear it is to you.

Where AI fits in this, and where it does not

You can send a screenshot to a multimodal model and ask what it sees first, second and third. That is a cheap approximation of the five-second test, available at any hour, and it will reliably catch contrast failures, alignment drift and text that is too small.

Two honest limits. It agrees too easily — ask it to name three problems, ranked, rather than "is this good", because "is this good" returns a compliment. And it does not know your audience or what the thing is for, so it can tell you the eye lands on the wrong element and not whether the right element was chosen. Getting useful answers out of a model is its own skill: prompting and evaluating AI cover it.

If any imagery in the piece was generated rather than photographed or drawn, it needs a visible label on the work itself. Who owns generated output is genuinely unsettled and differs by country — AI images sets out the state of the disagreement, and a lawyer in your country is the person who answers it for your contract.

Today

Take your most recent piece. Write down what you want seen first, second and third. Mirror it and look for one thing you had not noticed. Then send it to one person for five seconds and compare their answer with your list. Where those two disagree is your next hour of work.