davidkarolyi/landing-page-skill

Landing Page

You are not describing a product. You are recognizing a person.

Most landing pages are written from inside the company. They list what was built, explain how it works, compare it to alternatives, and hope the visitor does the math. The visitor does not do the math. They arrived from a link, mildly curious, mostly distracted, carrying a problem they have not fully put into words. They give the page a few seconds. In those seconds they are not evaluating features. They are asking one question: does this thing know what my life is like?

A great page tells them what the thing is in one plain line, then answers yes. The visitor knows what they are looking at, reads the next line and thinks "that is exactly it," then another, and by the end they are leaning in, because someone finally said the thing out loud. That is the whole job: orient them instantly, then lead them through a story they recognize.

Sit in the visitor's chair

Before writing a word, become them. Not a demographic, a person in a moment.

• Who is this, and what were they doing thirty seconds ago? Scrolling a feed. Closing a support ticket. Arguing with a spreadsheet at 11pm.
• What is the frustration they carry that they have half-given-up on solving?
• What have they already tried, and what did those things get wrong?
• What do they secretly wish someone would just say to them?
• What would make them feel relief, and what would make them feel sold to?

Write the answers down as they would say them, in their words, with their tone. If your answers sound like a marketing brief, you are still in the company's chair. Keep going until they sound like a message a friend would send.

The page is built from these answers. The product is only allowed on the page where it fits into their story.

One emotion, one lever

Every product people love has a personality, and that personality is almost always a single simple message built on a single emotion. Not a positioning statement. A feeling. Relief. Calm. Control. Finally. Freedom from a specific small misery.

Find that emotion. Then find the simplest lever that expresses it: the one plain observation that is almost embarrassingly obvious once said. Great pages state the obvious thing nobody has said out loud yet. They do not reach for clever. Clever makes the reader work; obvious makes them nod.

Once you have the emotion and the lever, every headline, every section, every image serves it. If a line does not point at the emotion, it is not on this page.

Resist explaining mechanism. "Every run is a receipt" describes how the system works. There is no feeling in it, and the reader has to decode it. Mechanism belongs in docs. The page carries only what the reader already feels and the promise of it going away.

Say what it is, first

The most common failure is a page that is all concept and no object. "AI hallucinates too much." "Work should feel lighter." "Rethink research." The reader finishes the hero and still does not know what they are looking at. Abstract is not the same as emotional. Abstract is just vague.

The first line names the thing. Plainly. What category it belongs to, and the one reason it is different or the one reason to care. Both, in a few words, in the same breath. "The AI web research tool that doesn't make shit up." Category, differentiator, done. It can be blunt or provocative, that is fine and often good. What it cannot be is unclear.

The reader should be able to look at the hero for three seconds and answer two questions: what is this, and why would I care. If they cannot, nothing below the fold gets read, because they have already left.

Every section is a visual and one short line

This is the rule for the whole page, and the hero is just the first place it applies.

A section is a picture with a line on it. The line is four to eight words. The visual carries the feeling; the line carries the fact. Nothing else in the section competes with either. If there is a second line, it is small, quiet, and optional, and most readers will never see it.

The hero is the same: one visual, the tagline, one button. The tagline is the four to eight words that say what the product is and why it matters. No supporting paragraph. No three value props. No feature carousel. The reader who scrolls no further should still leave knowing what the thing is and having felt one thing about it.

Recognition over persuasion

Do not sell. Recognize.

Once the reader knows what the thing is, the page stops describing the product and starts describing them. Their situation, in their words. A small specific moment they have lived. The feeling they have in that moment. Then, lightly, the release from it.

The page is not trying to convince anyone the product is better. It does not say "unlike other tools." It says the thing the reader already thinks, so accurately that they feel seen. The persuasion happens by itself, because people trust whoever understands them.

Each line should pass one test: would the reader say "yes, that is so true"? If a line only makes them say "okay, I see," it is informational and belongs somewhere quieter. If it makes them say "sure, you would say that," it is selling and it goes.

The page is a story

A landing page is not a stack of sections. It is one storyline, read top to bottom, that leads the reader through a sequence of feelings. Each beat lands one thing. The beats add up to an arc.

The arc opens with what the product is and why it is different, so the reader is oriented. Then it walks them through the problem, felt not explained, and through the relief, shown not claimed. As the problem gets more specific, the product gets more specific alongside it, always in the same plain register. By the end, the reader has felt the whole thing and the action at the bottom is obvious.

A typical arc for a research tool:

1. "The AI web research tool that doesn't make shit up." Orientation. What it is, why care.
2. "AI search is just as confident when it's wrong." The problem, named in one blunt line. Recognition.
3. A short animation of a confident answer citing a page that does not exist. The problem, felt. No words needed beyond a caption.
4. "Only live pages. Nothing from memory." What the tool does, stated as plainly as the problem was.
5. A visual of a real citation resolving to a real page. The relief, shown.
6. One button, phrased as what they get.

Notice each beat says exactly one thing. No beat carries two ideas. No beat explains mechanism. The words per beat are countable on one hand, and the visual does the rest.

One thing per section

Nothing on the page will be read except the few words you force into focus. So force them.

If you have two things to say in a section, you have two sections. If a sentence packs three concepts, you have not found the sentence yet. Say one thing, plainly, and let the visual carry the emotion around it.

Keep the communication channel thin on purpose. The thinner it is, the more certain you are that what gets through is the thing you chose.

Few words, light sentences

Write short. Then shorter. The reader will not read paragraphs, and every extra word dilutes the nod.

The tone is effortless, never forced. No exclamation marks doing the work the words should do. No stacked adjectives. No jargon, no internal names, no category words the reader would need to look up. Say it the way the reader would say it to a friend, then remove half the words.

None of it should feel like marketing. If a line could appear in a press release, rewrite it until it could only appear in a text message.

Everything visual, everything animated

The reader sees the page before they read it. Layout, motion, and imagery carry the feeling; the words carry the fact. Every section is a visual with a line on it, not a line with an illustration beside it.

Show the problem happening. Show the relief happening. A short animation of the thing going wrong, then the thing going right, does more than any copy. Motion should match the product's temperament: calm products settle, playful products bounce, precise products snap. Animation draws the eye to the line, never decorates around it.

Each section has exactly one loudest element. Space is generous. The page breathes.

What to cut

These come from the company's chair. Delete them on sight:

• Abstract heroes that state a theme but not the product.
• Feature grids and icon-plus-blurb tiles. They describe the system, not the person.
• "We are better because" and comparison tables. The reader did not ask for a debate.
• Explanations of how it works, architecture, pipelines, or clever internals.
• Buzzwords, category jargon, and anything that needs decoding.
• Testimonial walls, logo strips, and badges that ask for trust instead of earning it.
• Stacked value props trying to cover every possible reader. Pick the one person.
• Any section that says two things. Split it or cut it.
• Any line that exists because the team is proud of it rather than because the reader needs it.

Before you call it done

Read the page cold, as the person from the chair, on a phone, with something else on your mind.

• After three seconds on the hero, can you say what the product is and why you would care?
• Does the tagline carry both the category and the differentiator?
• Is every section a visual plus one line of four to eight words, including the hero?
• Does the page read as one storyline, each beat leading to the next?
• Does every section say exactly one thing?
• Does every line pass the "yes, that is so true" test?
• Is anything explaining mechanism instead of naming a feeling?
• Is there a paragraph anywhere? Why?
• Does the visual carry each section on its own if the words were removed?
• Is there one action, and is it obvious?
• What did you cut? If nothing, the page is still the company talking.

The feeling to chase: the reader knows what this is in three seconds, feels understood by the second screen, and reaches the button thinking "finally." Nothing else on the page matters.