Landing page best practices: eight zones, one job each.

Every landing page is one argument in three moves.

A landing page is not a brochure with a button attached. It makes a promise, supports that promise with something the reader can check, and asks for one thing. Every element on the page is doing one of those three jobs — or it is standing in front of something that is.

That is the whole test, and it is more useful than any layout. Point at any block on a page you already have and ask which of the three moves it belongs to. The blocks that cannot answer are the ones costing you conversions, and they are almost always there because somebody had them lying around: the company history, the second offer, the awards row, the newsletter box that competes with the actual ask.

Write the one-job sentence first

Before you open a builder, finish this sentence: this page exists to get [a specific person] to [one action]. One person, one action. A page that asks for a download and a call and a newsletter subscription has not offered three chances to convert; it has asked the reader to make a decision about which decision to make, which is a slower and less pleasant thing to be asked.

The one-job sentence also decides the kind of page you are building. A page collecting an email address for something free is a different object from a page selling a two-thousand-dollar program, and the difference is not decoration — it is how much argument the page has to carry. Sales page or landing page works through which page suits which job.

The eight zones below are where those three moves land on a real page. A short page runs all three in one screen; a long page spends six screens on proof and objection handling. The order does not change, and neither does the rule that every zone has exactly one job.

Answer “what is this, and is it for me” in one screen.

The first screen is not a decoration on top of the page. It is the whole page for anybody who does not scroll, and it is the reason most of them do not.

A reader arriving from an ad, an email or a post is carrying a half-formed expectation and about as much patience as it takes to read two lines. The first screen either confirms the expectation or contradicts it. If it confirms it, they keep going and every other zone gets a chance. If it contradicts it — different words, different offer, a heading about your company rather than their problem — they leave, and no amount of excellent argument further down the page is ever read.

Message match is the cheapest fix on this list

Whatever sentence sent them — the subject line, the ad headline, the link text — the first screen should continue it, ideally in the same words. This costs nothing and is skipped constantly, usually because the page was built before the campaign and never revisited afterwards. If you run one page behind several different sources, match the biggest source and accept the rest, or build a variant for each of the ones large enough to deserve it.

The fold is a device, not a pixel

There is no fixed line. What matters is the first screen on the device a real reader is holding, which is usually a phone in portrait with a browser bar eating the top. Check it there first. A hero that looks commanding on a desktop monitor frequently pushes the button, the date and the entire promise below the visible area on a phone — and phones are where most of your readers will decide.

The mistake that costs the most here is the company-shaped headline: a line that would be equally true on a competitor’s page. “Enterprise-grade solutions for modern teams” names no problem, no outcome and no person. Replace it with the sentence the reader would use to describe what they are trying to do, and the rest of the page gets easier to write.

Name the outcome, the format and the cost.

The promise block is where a reader decides whether the trade is worth making. It has three parts, and pages usually ship with one of them missing.

1. 01The outcome. What they will be able to do afterwards, written as a capability rather than as a feeling. “Run your first webinar end to end” beats “feel confident about webinars”, because the first one can be checked and the second cannot.
2. 02The format. A forty-page workbook, a sixty-minute live session, six weekly calls, a twelve-lesson course. Readers are deciding how much of their life this takes, and a page that will not say is asking them to assume the worst.
3. 03The cost. Money and time. Free things cost an email address and an hour; say so plainly. Paid things cost a price, and hiding it moves the whole decision to a page the reader may never reach.

Outcome bullets, and the rule that keeps them honest

Three to five bullets do the work here, and each one should point at something the reader will be able to do — not at something the product contains. The test is to add “so you can…” to the end of the bullet. If nothing sensible follows, you have written a feature and labelled it a benefit. The wording of those bullets is the single highest-leverage paragraph on most pages, which is why it has a guide of its own.

Specificity beats intensity

Every intensifier on a landing page is a signal that the claim underneath it is weak. “Dramatically improve your onboarding” is weaker than “cut the first-week questions you answer by hand”, because the second one names something the reader recognises from their own week. If you find yourself reaching for an adverb, you are usually missing a noun.

One more thing belongs here and is almost always missing: what the reader has to do. An offer that requires attending live, or reading something before the first call, or having an existing audience, should say so in the promise rather than in the terms. People who find out later feel misled, and they are right.

Give them something they can check.

Proof is not a testimonial section. Proof is anything that makes the promise plausible to somebody who has no reason to believe you yet — and most businesses have more of it than they think.

This zone is where new pages go wrong in the most expensive way, because the shortcut is so available. A page with no customers can be given five testimonials in ten minutes. Nobody will check them. It still should not be done, and not only for the obvious reasons: invented proof sets an expectation the delivery cannot meet, it attracts the wrong buyers, and it is the one mistake on this page that a redesign cannot undo once somebody notices.

Proof you already have

• →The work itself. A sample chapter, the first module, a real screenshot, a five-minute excerpt of the actual session. Showing the thing outperforms describing it, and it is available on day one.
• →Demonstrated expertise. The specific detail an amateur would not know, stated in passing. A reader who recognises that you understand their problem in detail has been given proof, whether or not the page calls it that.
• →Your own history, plainly stated. What you have done, for how long, with whom — without inflation. “I have run this session eleven times” is proof. “Industry-leading expert” is not.
• →A named policy. A refund window, a guarantee with terms, a cancellation policy written in plain words. A policy is checkable in a way that a reassurance never is.
• →Honest limits. Saying who this is not for, or what is not included, buys more credibility than another badge. It is the cheapest proof on the page and almost nobody uses it.
• →Numbers you can substantiate. From your own operation, with the definition attached. If you cannot say exactly what a number counts and where it came from, leave it off — a vague number is a liability, not an asset.

Placement: proof answers a claim, so it goes after one

A wall of testimonials above the promise is a set of answers to a question the reader has not asked yet. Put each piece of proof next to the claim it supports: the credential near the promise, the sample near the description of the work, the refund policy near the ask. Proof placed against the right claim does several times the work of the same proof gathered into a carousel.

Answer the objections in the order they arrive.

The body is the distance between “this sounds good” and “I believe this will work for me”. Its length is decided by price and unfamiliarity, not by taste.

A free download for people who already know you needs almost no body at all. A two-thousand-dollar program for strangers needs a long one. The common advice that “long pages convert better” and the opposite advice that “nobody reads any more” are both wrong in the same way: length is not the variable. Unanswered objections are. Write until the real objections are answered, then stop.

The order objections arrive in

Where do the real objections come from? Not from imagination. They come from replies to your emails, questions in sales calls, support tickets, and the things people type in chat during a live session. A body section written from that material answers questions readers actually have, in the words they actually use, which is also why it tends to read better than a body written from a template.

The blocks that carry a body

• →How it works, in three steps. A mechanism the reader can picture. Make the first step something they do, not something you do — it turns an abstract offer into a sequence they can imagine starting.
• →What is included. A plain inventory with quantities: how many sessions, how many lessons, how long, delivered how. Inventories reassure; adjectives do not.
• →Who this is not for. The most under-used block on the internet. It costs you readers who would have converted badly and improves the fit of everyone who continues, and it makes every other claim on the page more believable.
• →The alternative, named fairly. What happens if they do this themselves, or do nothing. Fair beats slanted here — a reader who recognises an unfair comparison stops trusting the rest of the page.
• →An FAQ, if the objections are numerous. Objection handling with a hat on. It works because readers self-select into the answer they need instead of reading five paragraphs to find it.

One action, made obvious and made safe.

The ask is where the page stops arguing and starts asking. Two rules carry it: repeat the same ask, never multiply it, and say what happens immediately after the click.

Repetition is not duplication. On a long page the same ask should appear several times — after the promise, after the proof, after the objections, at the end — because a reader who becomes convinced at screen four should not have to hunt for the button. Multiplication is different: a page offering a download, a demo booking and a newsletter subscription has three asks and converts on none of them particularly well, because it has handed the reader an extra decision before the one that matters.

Say what happens next

Most hesitation at the button is not doubt about the offer. It is uncertainty about the next five seconds: does this charge me now, does it call me, does it start a subscription, how do I get out. Answer it in one line under the button. The trial line under our own buttons is written exactly that way — the length, that nothing is due today, that a payment method is required, and that cancelling before the first charge costs nothing — because every one of those clauses is a question somebody would otherwise have to guess at.

The button is a sentence, not a label

“Submit” describes what the form does. “Get the checklist” describes what the reader gets, which is the only part they care about. Match the button to the heading directly above it, so that reading the two together makes one sentence rather than two fragments. Button copy patterns are in the copywriting guide, along with the microcopy that sits underneath.

On pages where the ask is a purchase rather than a form, the same rules carry through to the checkout itself — and the checkout is where a surprising share of convinced readers still stop. Checkout page optimization picks the page up from there.

Catch the people who scrolled past the button.

Somebody who reaches the bottom of a page without clicking is not uninterested. They are unresolved. This zone exists for them, and it is treated as a dumping ground on most pages.

The reader who scrolls past your ask is doing one of two things: looking for the answer to a specific worry, or looking for a reason to trust you. Both are addressable. What they will not do is scroll back up through the argument they have already read — so the bottom of the page has to carry both the remaining answers and another way to act.

The FAQ is often the most-read block on the page

Not because readers love reading questions, but because it is the one block where they can find their own worry without reading anyone else’s. Use it for genuine objections, in the reader’s words, answered without defensiveness. An FAQ full of questions nobody asked — “why is your solution the best?” — is marketing copy in a costume, and readers spot it immediately.

One structural note: the page should end with something to do. A page whose final element is a link farm has spent its last impression telling the reader to go somewhere else. End with the ask, and put the legal links underneath it in the type size the law requires rather than the type size that invites clicking.

Ask for the least you can defend.

Every field is a tax paid in completions. The test for keeping one is simple: name the workflow that reads it. If you cannot name a workflow, the field is collecting data for a report nobody opens.

Run that test out loud, field by field, on a form you already have. “Company” — who reads it? “Phone” — what happens when it is filled in? A field that survives the question earns its friction. Most do not, and the ones that do not are usually there because somebody thought the data might be useful one day.

The mechanics that quietly lose completions

• →Labels above fields, and kept visible. Placeholder-only labels vanish the moment somebody types, which turns a review of the form into a memory test.
• →The right keyboard on a phone. An email field that opens an alphabetic keyboard adds work to every single submission. Field types and autocomplete attributes are ten minutes that keep paying.
• →Errors that say what to fix. “Invalid input” is not a message. Name the field, name the problem, and never clear what they already typed.
• →A visible, changed state on submit. A button that does nothing for two seconds gets clicked again. Disable it, change the label, and make success unmistakable.
• →Captcha only if you have a spam problem. It is friction applied to everybody to solve a problem you may not have. Check the actual submissions first; if you do have one, take the least invasive option that works.
• →Consent written like a sentence. Say what you will send and roughly how often, next to the field that collects the address. It is both a legal matter and the first honest thing you tell them.

Multi-step forms, honestly

Splitting a form across steps helps when the first step is trivially answerable, because starting is easier than finishing and people who start tend to continue. It does not help when the offer is unwanted — a form nobody wants to complete is not improved by being served in instalments. Treat multi-step as a remedy for length, not as a remedy for a weak promise.

Deliver immediately, then set the next expectation.

The page is not finished at the submit button. The zone after it is the one people build last, build worst, and then wonder why a converted reader never engaged again.

Three things have to happen in the first minute. The thing you promised has to arrive. The reader has to be told what happens next, from whom, and when. And the conversion has to be recorded against whatever produced it, because the source that brought this person is the only way you will ever know which of your channels is working.

1. 01Deliver on the page where you can. If the thing is a download, put it on the confirmation page as well as in the email. Making somebody switch to their inbox to receive a PDF adds a step where attention is lost and deliverability becomes your problem.
2. 02Send the email anyway. It is the durable copy, it proves the address works, and it starts the relationship in the channel you will use later. Say who it is from and what arrives next.
3. 03Set the next expectation precisely. For a dated event: the date, the time in their timezone, and that a reminder is coming. For a purchase: what happens now, and when access begins.
4. 04Make the one extra ask that is allowed. The conversion is banked, so this is the safest moment on the whole journey to ask for one more small thing — a calendar entry, a follow, a next-step offer. One, not a page of them.
5. 05Record the source on the record. Not just the conversion count. The person, the page and the source together, so that a purchase later can be traced back to the page that started it.

The confirmation page deserves more thought than “Thanks, check your email”. It is a page with a captive, positively-disposed reader and no competing demands — thank-you page ideas covers what to put on it and in what order. If your one extra ask is a paid one, the mechanics are in order bumps and upsells.

Whatever the first email says, it has to arrive. Email deliverability basics covers the sending setup that decides whether it does, and the welcome sequence covers what the next few days should contain. For a registration page, the follow-up window afterwards is its own playbook — what the registrations are worth is decided there rather than on the page.

Run it on a page you already have.

Twenty checks, four zones of the page, about ten minutes. Run it on the page as a stranger would meet it: on a phone, on an ordinary connection, without being logged in.

The order matters. Fix the first screen before anything else, because a reader who leaves in the first few seconds never saw the work you did further down. Then fix the ask, then the form, then the middle. It is the reverse of the order most people work in, and it is the order the returns arrive in.